Watch this Ask the Expert session to learn how to refine your skills, optimize workflows and maximize the value of SAS solutions within your marketing strategies.   Image 1: Marketing AI With SAS Customer Intelligence 360 and SAS® Viya®   Watch the Webinar   You will learn:   Data science and analytics can bring incremental value to marketing opportunities. To accelerate the delivery of AI solutions to marketers. To follow a step-by-step approach in bringing your AI-powered campaigning to life. To observe SAS viewpoints on 2025 innovation themes within Martech use cases.   The questions from the Q&A segment held at the end of the webinar are listed below and the slides from the webinar are attached.   Q&A   Why is SAS building another AI solution in a crowded ecosystem of analytical solutions available in 2025?   Totally legit to say because this is the first time in the history of SAS that we're building a domain-specific AI solution for martech and customer experience. It's never been done before, but we’ve seen year after year the friction building up between the data science and marketing communities. The marketing community is very data- and insight-hungry when it comes to activation use cases.   As AI and ML continue to innovate at a fast pace—and those of us practicing in the field know that pace is accelerating—it’s becoming more challenging for newer participants in martech to keep up, especially since martech itself is evolving rapidly. We need to introduce a solution that brings deeper synergy between these two communities. Otherwise, we’ll continue to see two distinct segments within companies moving in different directions.   We don’t want AI and ML to be overlooked in marketing workflows. We believe there’s massive potential for further improvement (or incrementality). As AI and ML mature over time, and it becomes harder for some to keep up with the pace of innovation, we want to reduce that gap.   How is project retraining scheduling dependent on underlying data scheduling?   We can mirror it. There are a number of ways to approach that, but ultimately it would involve understanding your team (or brand's) desire to mirror when data is updated on the front end of a workflow that could trigger an alert for retraining.  We intend to offer a wide variety of options for how you want retraining to be scheduled—manually or autonomously. We're working with early adopters (as some of them may be watching this session today) and receiving real client feedback on how they want the software to perform.   A question like this is valuable because I’ll share that insight with my product management and supporting R&D teams. We're building this product with the understanding that this interest exists.   I’ve heard rumor that, aside from Intelligent Decisioning, SAS also develops Marketing Decisioning. Is this true?   Yes. SAS Customer Intelligence 360 in the near future will offer clients the opportunity to adopt new modules focused on Marketing Decisioning and Marketing AI. The two modules will be siblings to address batch, fast-batch and real-time customer journey use case themes. Keep your attention on this page for future updates.   How can an existing Viya installation be integrated with CI360?   We would suggest reviewing this SAS Communities article for details on this question.   Recommended Resources   SAS Thought Leadership & SAS Community Articles on Data Science & Martech SAS Learning Subscription for CI SAS Support Community   Please see additional resources in the attached slide deck.   Want more tips? Be sure to subscribe to the Ask the Expert board to receive follow up Q&A, slides and recordings from other SAS Ask the Expert webinars. ... View more

Marketers, advertisers and customer-focused brands face growing challenges in making sense of complex data to drive actionable insights. This article introduces recent SAS research efforts to develop a solution-oriented bridge to this cited gap, offering prescriptive recipes to address trending use cases for B2C (and B2B) brands. SAS is diligently exploring how data science and AI can bring forth incremental value to marketers across industries. The intent is to create synergy improvements between marketers and data scientists while elevating self-sufficiency in running analytics at scale through use case-driven solutions that package the best of SAS capabilities in a simple-to-use interface.   In this article:   Transforming Marketing Organizations Into Analytical Factories The Role of the Data Person and Configuring Actionable Recipes The Role of the Business/Marketing Person and Leveraging Projects   Image 1: Applying AI To Marketing   As a result of these recent trends, marketing research and customer analysis as a discipline should continue to explore the following considerations to unlock incremental innovation:   Treat all of your owned data assets as a priority. It is frequently mentioned in the martech industry that brands must maximize the potential of their "owned" customer data sources. However, this also means we should prioritize activating structured, semi-structured and unstructured data sources. Brands cannot deprioritize semi-/unstructured data because of a perception that these flavors of information are difficult to use, as the opportunities to harvest these ingredients for analytical innovation is front and center today.    Recipes for data and analytical models. Retail data offers different analysis experiences from mining financial banking data. Similarly, a churn model in the discretionary fashion industry, where most customer behaviors indicating disengagement is silent, will likely differ from a churn strategy for a subscription-oriented streaming service, where churn is explicit (or observable). Recipes provide comprehensive use case-specific solutions to reduce adoption friction and increase the likelihood of success in leveraging customer insights within a marketer's workflow.   Optimize customer-level treatments. Machine learning models don’t make decisions — but they do identify actionable signals in noisy customer behavioral data. It’s up to our CX and/or marketing teammates to take prescriptive insights and execute data-driven targeting/personalization. But the decision isn’t always clear. Should a brand optimize on likelihoods, engagement, propensities, or profitability? The path forward will vary by a brand's unique business model, as well as the industry vertical it operates in.   Transforming Marketing Organizations Into Analytical Factories   Numerous well-known martech vendors have attempted to aspire to this vision over the years. However, the theme across the big martech vendors has frequently gone like this. As public companies, we strive for growth year after year. There are three segments we target our software and technology towards: marketers, analysts and data scientists.   In the world today, there are a large volume of marketers, moderate amount of analysts, and a smaller subset of data scientists. Generally speaking, the theme at major martech vendors has been to automate analyses on behalf of marketing users using templates to provide AI insights while masking/hiding the manual workflow steps. While this can provide benefits in regard to perceived speed-to-market acceleration, the auto-analysis behind these templates typically do not offer customization features to conform to a brand's unique business model. The data science community understands incremental opportunity is being left on the table with solutions like this.   This trend has resulted in a compelling insight for us at SAS (a private company with different motives), and a deep exploration of the Marketing AI landscape has resulted in the realization that there is a different way to approach this emerging paradigm.   Image 2: Domain-specific AI Solutions For Marketing & Customer Experiences   Think about the magnitude of requests that come in from customer experience and marketing teams to their supporting data science and analyst teams. The wish list includes actionable scoring for topics like acquisition, upsell, retention, segmentation, next-best-action (or experience), recommendations, lifetime value, pricing personalization, attribution and net lift.   The list could go much longer, but as many readers recognize, the point remains the same. Customer experience management has an insatiable appetite for data intelligence. This myriad of desires stratifies further when considering industry context. Let's take a moment and imagine we are in a requirements gathering meeting between two teams - data science and marketing.   Image 3: Requirements Meeting Between Marketing & Data Science   The marketing and CX teams responsible for the interactions between a brand and everyday consumers speak one language. The data science and analyst group likely speaks another. Terms like acquisition, cross-sell, churn, targeting, personalization, A/B tests, conversions, and impressions are the common tongue of the martech universe. Alternatively, words such as misclassification, precision, average squared error, confusion matrices, outliers, auto tuning, neural networks, and random forests represent the language of data science.   In other words, marketers do not typically think in terms of algorithms, and analytical jargon creates confusion, friction and inefficiency for those not trained in the discipline. This can be intimidating for many working professionals, and why data and analytical literacy across the enterprise is increasing in relevance in 2025.   Image 4: Data Science Jargon Creates Friction For Those Not Trained In The Discipline     The language of marketers and customer experience is rooted in use cases and outcomes. Domain expertise, acceleration and simplifying the process of analytically injecting data-driven intelligence into marketing workflows is the desire, and year after year, SAS clients share feedback on this challenge. If this is what the martech community craves, this is a call-to-action to my brothers and sisters practicing data science across all industries.    You want to see your analytical assets bring rewarding impact to your brand, right? You want to observe your efforts making a significant positive difference in customer journeys, correct? Then the democratization of marketing team enablement via customer journey orchestration and prescriptive analytics benefits from speaking their language. Further, SAS AI development efforts targeting the martech community is to bring forth software and technology that removes technical jargon and adoption intimidation.   Image 5: The Language & Use Cases Of Martech   Our vision at SAS is to serve as the market leader in advanced audience creation & targeting, independent of channel, for enterprise brands leveraging complex, disparate data sources and wishing to consistently deliver superior understanding within customer journeys. In other words, we want to empower brands to practice responsible marketing. However, none of this aspirational messaging matters unless SAS delivers comprehensive, end-to-end use case-specific solutions to common martech challenges.   An emerging trend to combat the ongoing analysis inefficiencies cited above involve Do-It-For-Me (DIFM) prebuilt recipes representing a specific ML/AI algorithm or model ensemble, processing logic, and configuration to auto-build and execute a trained solution that comprehensively solves (or improves efforts against) specific business problems. The analytical models and data engineering pipelines are ingredients of a broader recipe that get trained on data and parameter configurations to optimize a solution's ability to contribute significant value when pivoting to customer inference and marketing strategies.   Image 6: Increase Synergy & Accelerate Analytics   There is a fair amount to unpack and explain. SAS practitioners who program, code, visualize or model data recognize that the "beginning of time on Earth" is defined as midnight on January 1, 1960. The methodology of data science workflows across high-code and low/no-code windows has largely been a "best-practice approach" with specificity across the analytical lifecycle taught in academia and software vendor education/training programs. Now, its 2025, and SAS is introducing a new approach for enterprises to unlock value from beautiful, wonderful data.   The concept of recipes and required ingredients can be outlined as:   Data – What data do I need?​ Preparation – How does it need to be transformed?​ Use-case specific – Applicable ML/AI algorithm​(s). Scoring​ - Segments, recommendations, propensities, etc. Activation – Using the scoring in journeys and channels.   From a software user's perspective, our motivation at SAS is to create an experience that unites what is special and unique about data scientist and marketer talents. To achieve this, use case-driven solutions that proactively and prescriptively guide these two types of anticipated users is the intended vision.   Data person - Users who have previous experience managing, engineering or analyzing data assets. Business/marketing person - Leverage recipes approved by "data person" to run no-code analytical projects at the velocity necessary to support customer treatment strategies and campaign activation cycles. Image 7: SAS for Marketing AI   We know what you're thinking. Is this another fully or semi-automated analytics offering that promises the lovely benefits of AI but sacrifices transparency, control and customization? The answer is a "pound the fist on the table" moment, and an enthusiastic NO! Remember, we are SAS, and for the last 45+ years, our heritage is rooted in being the founder and future of analytics. It all began when curious minds set out to answer some big questions. Is there a better way to analyze data? How can we turn data into intelligence? Who might benefit from our technology?   Years ago, lines of code were the key to something extraordinary. Now, SAS has customers around the world. We analyze billions of rows of data every second that change the way we work and live. Ultimately, we believe curiosity is at the heart of human progress. As the years pass by, SAS continues to hear in the field that marketers (not all, but many) struggle to do analytics at scale.   Our intention is to support every step of the marketing/customer analytics journey in an applied manner through functionality that will help with use case driven solutions that guide users through their challenges. Benefits will include self-service analytics run by marketing teams with minimal external support, the ability to deploy anywhere to run analytics against your data (wherever it lives), and streamline data preparation (or engineering) with accelerators and automation. The Role of the Data Person and Configuring Actionable Recipes The efforts made by data engineers everyday focused on maximizing the potential and accuracy of a brand's data assets for analytics and marketing is a critical function. Let's begin explaining how efforts by SAS in the domain of Marketing AI will enable a data-savvy individual in crafting recipes for their marketing counter-parts to leverage. How do use cases and requirement meetings begin? Well, with some type of objective that the brand aligns to. Below readers can view a screenshot of how the software will begin assisting the user through prompted screens to guide a workflow of configuring a recipe focused on customer churn (or retention). Image 8: Define Your Objective   Recipes can be given custom names, codes, descriptions, business context and specificity on customer events of interest. Ranging across a variety of business models and industries, SAS recognizes our users need the ability to adapt and apply customization to gain incremental value through the recipe's specificity.  Moving on, brands aspire to strategically manage their business through prioritizing customer convenience. This involves  anticipating and responding to customer needs, while manifesting in proactively delivered, seamless, and unobtrusive interactions. The intent is to provide personalization, assistance and valued services. However, there is a little secret in the marketing and customer analytics ecosystem that practitioners frequently will admit to when pressed for honest feedback. A massive proportion of customer & marketing analysts in 2025 continue to skew towards the wrong end of this workflow spectrum: "I spend more than 80% of my time accessing/preparing data, and less than 20% actually performing analysis."   Speed bumps like this usually emerge when customer experience teams require data to unlock and fuel advanced marketing insights.   Image 9: Simplifying Accessing Data In A Complicated Martech Ecosystem   For example, have you ever tried to extract event (or raw HIT) data from your preferred marketing cloud vendor? It tends to be challenging for data engineers based on numerous reasons (data volume, structure, storage, etc.). Our viewpoint going forward is to remove high-code requirements or manually connecting to APIs (Application Programming Interfaces) to transform the user experience within the software by providing simple point-and-click prompted steps to complete data absorption into SAS across a variety of martech vendors. Image 10: Flexibility Across Data Source Selections   Users can select data originating from different martech vendors (assuming brands have contractional relationships with these organizations that allow access). This will include features to access, query, filter, profile and more when allocating data sources to a recipe configuration. Accelerating user workflows through the known challenges (mentioned earlier) residing in the phase of DataOps has been a key research area of interest at SAS. Image 11: Joining Data Tables From Different Sources   Table joins are one of many examples in the DataOps phase for users to begin the process of acceleration through improving data quality, reducing time-to-value, and fostering collaboration between data engineers, data scientists, and business/marketing stakeholders.  Image 12 below highlights the completed steps of joining multiple tables using variants (inner, left, right, etc.) of join definitions. Image 12: Table Joins Without Sacrificing Capabilities And Offering A Variety Of Join Methods   Preparing analytic base tables (or ABTs) is the process of organizing data into a flat table schema that's traditionally used by analysts for building analytical models and scoring (predicting/inference) the future behavior of a customer. A single record in this table represents the subject of the prediction (such as a customer or anonymous visitor) and stores all data (variables, features or predictors) describing this subject. An example of this workflow includes aggregating transactional records associated with one customer to "flatten" the table and meet the expectations of the downstream algorithms that will produce the prescriptive scoring for marketing. Readers take note - users will will have features available across the spectrum of what is expected for best practices within data science and engineering. Image 13: Defining Relevant Numeric and Categorical Predictors By Recipe Type   If this concept is new to any readers, here is a marketing-centric example of how an ABT is important in the context of attribution analysis.  Although there are many sub-topics related to the acceleration of data preparation, the intent of this article is to introduce readers on leveraging a guided interface that assists in the areas of automated aggregations, data quality handling, and feature engineering. The next topic of interest is defining the business objective of the recipe, or in data science jargon, providing the brand's custom definition of the target (or dependent) variable. Image 14: Defining the Business Objective of a Recipe   The target variable is the data signal that you're modeling or predicting. It's also known as the dependent variable, response variable, or y variable. The target variable is important because it defines the type of problem you're solving (regression or classification) and determines how to evaluate a model's performance. Here is a contextual example: Regression Use Cases: Predicting house prices (target: house price), predicting stock prices (target: stock price). Classification Use Cases: Predicting whether a customer will churn (target: churn - yes/no), predicting whether an email is spam (target: spam - yes/no). The target variable should be well-defined, measurable, and relevant to the problem you're trying to solve. An example can be framed in a credit risk model, where the target variable might be whether a borrower will default on a loan (1 = default, 0 = no default). A value proposition to bring to your attention is how SAS simplifies and automates workflow steps by removing the user's responsibility to simply know which algorithms should be used in a recipe based on the target variable's type (numeric or categorical).  Image 15: Flexibility in Defining a Target Variable Definition   The formation of a model-ready ABT with a customizable target objective leads us to the topic of project variants. Before a brand's data person can share pre-configured recipes with their marketing counterparts, SAS is bringing forth another value proposition in automating the readiness of addressing difficult customer experience scenarios. Take for example, churn. The definition of a churn event of interest can carry different time lines on when to flag (or censor) the customer's exiting behavior. Project variants, as a user feature, will automate the formation of sibling ABTs to enable marketers choice on which time-series scenario is more relevant to their use case. Image 16: Project Variants For Different Recipe Strategies   After the user is satisfied, the next step is to kick off the job for preparing the table variants. The concept of local analytical modeling agents as a service needs to be introduced to highlight what is unique about this area of innovation.  Local analytical modeling agents as a service refer to AI-powered systems that perform data analysis tasks locally on a user's computer or network, rather than relying on a cloud-based service. These agents can be accessed as a service, meaning they can be deployed and used without the need for extensive coding or infrastructure setup. Image 17: Preparing Tables With Local Analytical Modeling Agents As A Service   The agent runs on the user's device or local network, minimizing reliance on external services, thus providing local execution and enhancing data privacy by keeping data within the brand's control. Local agents are a major part of the reason why customization of user features to suit specific analytical needs can be tailored to use cases and recipes. Other benefits to call out include:   No movement of data. Brands simply need to provide SAS access to their cloud structure and connection to owned data. Extreme Agility. SAS manages the software's design environment and pushes the necessary logic to run in a brand's data environment. Service. Analytical Agents are delivered by SAS as part of our solution for Marketing AI. Image 18: Transparent Job Views and Logs   It's vital to indicate that users have transparency to review and assess jobs that have been kicked off. In the image above, we can observe an example where data connections have occurred, data extraction steps, and every analytic base table variant that is generated (or failed) with user access to information within detailed logs. Image 19: Table Preparation Summary and Status Dashboard   For the benefit of recipe configurations, users can confirm when a job has completed successfully across table variants, and can step forward to customizing column names and designating scoring eligibility. These user features simplify the proposition of delivering customizable prescriptive information to marketing team counterparts to action on. Image 20: Customizing Data Display and Defining Scoring Eligibility   SAS emphasizes incorporating fairness and oversight at every stage of the analytics journey, from development to deployment. This ensures that Marketing AI models are not only accurate but also fair and equitable in their predictions and decisions. Users will have tools for assessing fairness and bias in recipes, identifying potential variances in model performance for different groups. These assessments help uncover biases that might be embedded in the data or the model itself. Image 21: Configuring AI Fairness   Detecting bias is only half the solution, and SAS will enable users to auto-mitigate bias during model training, helping organizations create more equitable and fair predictions. The software, as needed, will adjust algorithms, rebalance datasets, or use other techniques to reduce the impact of bias. SAS recognizes the importance of explainability in building trust and confidence in AI models. Explainable AI helps users understand how to recommend strategic decisions and identify potential biases or areas for improvement.  The Role of the Business/Marketing Person and Leveraging Projects   Projects is the transition from a configured recipe to unleashing the opportunity for marketing teams to train, activate, and manage models for a given use case. The intent is for SAS to reduce the complexity and automate as much of this process as possible. For this section, we will utilize both live demo video snippets with screenshots to bring the user's experience to life for readers. In this short introduction, the business/marketing user enters the Projects section of the interface where configured recipes await them. In a production deployment at a brand's site, the user's welcome screen would likely look similar to the Image 22 below. Image 22: Project Users - Welcome Screen   Analytics and data science do NOT need to be constrained, and should be applied to what MATTERS to the brand. At the end of the introductory project user video above, we were presented with choices to optimize the project. Image 23: Optimize On What Matters   As marketers, what do you care more about in the context of churn or retention?    Balancing the effort to save existing customers at risk of leaving while avoiding unnecessary retention actions. Optimizing your strategy to ensure retention efforts generate the highest net profit. Ensuring the predictions to target/not target are correct as often as possible. Drive the highest total revenue. Maximizing precision for targeting customers who are truly at-risk. Guess what? SAS offers choice to select the strategy that aligns with your brand's preferences. This is a wonderful enhancement that transforms a user's thinking to shift from probabilities and likelihoods to maximizing net profit or revenue. Regardless of industry type, SAS believes CMOs and CFOs likely care more about monetary terms which improve the state of the business. Let's strive forward into another software demo snippet which will showcase how a business/marketing user can take a configured project recipe and start training the solution. The important takeaway for readers is the user is benefitting from the configuration of the recipe prior to accessing the project, allowing them to quickly train an automated solution without technical friction or distraction. Optimizing models for real business impact—like revenue, retention, and profitability - has never been easier or faster. Speeding up analytical workflows with ready-to-use (yet customizable) templates for common marketing challenges is the objective to ensure data-driven insights never get overlooked again due to the velocity of martech requirements today. As training completes, the project experience receives a number of auto-benefits for the user. Image 24: Project Recipes, Optimization and Responsibility   When a user receives an indication that project training has completed, the software's left menu pane confirms a series of automated outputs are now available for review. Let's begin by demonstrating what the training results screen offers users. After viewing, let's discuss why this matters to business and marketing users.   Natural language generated explanations and co-pilot helpers assist throughout the user's experience to help accurately understand the output (to reduce any possible intimidation issues that prevent business adoption). Interactive audience cutoff selection for inclusion/exclusion prior to marketing activation. Modeling results summary and key performance measures to increase the confidence of the user before campaign activation occurs. Data science technical terms like accuracy, precision, recall and specificity are explained in common business language to help users understand, appreciate and make better outcome decisions. Assessment plots and butterfly charts to gain transparency on which data signals (or predictors) matter (both positive and negative correlations) in accordance with a brand's defined recipe objective. Confirmation if the trained solution will generalize as expected, alleviating concerns associated with under- and over-fitting models. Fairness and ethical AI checkpoints to alert users if bias and/or other concerns exist. There is a theme we hope readers are detecting in the area of simplifying the data science workflow. Now, let's pivot and review a walkthrough of the eligibility, scheduling and output user interface screens. Once again, let's revisit the question of why this matters. Interactive eligibility provides two key benefits involving customization of the inclusion/exclusion filtering of customers who will (or will not) be part of an activation-oriented audience, as well as the assistance of the software to proactively prescribe selection recommendations. Furthermore, the demo video cast a spotlight on the simplicity of scheduling retraining of the use case solution to ensure marketers are activating on the latest (and freshest) customer scoring. Finally, the output screen examples shown provide users the ability to select from a variety of destinations within SAS Customer Intelligence 360 and external martech vendors, such as Amazon Redshift/S3, Microsoft Azure SQL, Google BigQuery, Oracle, SalesForce, Adobe and Snowflake.   We have reached the final phase of the project user's workflow, which relates to monitoring specific metrics prior to activation. The reasons SAS built this functionality into a Marketing AI software solution's user work flow includes:   The software will guide and prompt the user with recommendations (while still allowing interactive customization). Metric monitoring supports automation on when to alert users of decaying health of the trained solution and provides data-driven evidence when the solution needs to be revisited/refreshed, as well as when fairness becomes an issue and should be considered for mitigation (manually by a user or auto-addressed by the system itself).   Marketing teams will hopefully agree that using predictions and segments that are out-of-date and making unexpected mistakes on treatment recommendations will negatively impact KPIs. Scoring insights and dashboards will auto-populate on behalf of users when activation takes place, and labeled data refreshes to support scheduled (or alerted) training updates for the project solution. Here is a brief walkthrough of what users can expect. We have spent a great deal of time discussing the notion of activation. To be transparent, we will now introduce readers to how an analytically scored Audience eligible for activation can be seamlessly absorbed into SAS Customer Intelligence 360 Journeys. In the context of SAS, an Audience is the starting point for a journey orchestration use case (churn, acquisition, next best action, etc.).    In other words, it's time time for the big finish! Although this last demonstration example was be handled comprehensively by SAS martech capabilities, selecting an external vendor destination is always an option for users as well. Image 25: SAS Customer Intelligence 360 and Marketing AI   Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here. For those who want to dive deeper into the current state of the marketing/customer analytics technology ecosystem, check out fresh (and unbiased) research here. ... View more

The last few years have upended the way customers select to interact with brands. Digital engagement has led to higher customer expectations, rising demands on marketers for personalized interactions across all channels, and an increased likelihood that customers will jump ship if brands don’t deliver.  Generative AI (genAI) is THE buzzword of the last two years, being discussed everywhere, and the customer analytics ecosystem is no exception.   Image 1: Consumer Demands Are Higher Than Ever   The idea of designing an LLM-enabled user interface within a software product to help with user democratization is a reasonable hypothesis. However, practitioners reading this article will likely agree that genAI does nothing (or very little) to ensure the robustness of the underlying analyses. Instead of being instantly romanced by genAI's potential (which remains to be seen as use cases mature), readers should look for differentiated analytical techniques to innovate customer treatment strategies.   As a result of these recent trends, customer analytics as a discipline should continue to explore the following considerations to unlock incremental innovation:   Treat all of your owned data assets as a priority. It is frequently mentioned in the martech industry that brands must maximize the potential of theirs 1st party customer data sources. However, this also means we should prioritize activating structured, semi-structured and unstructured data sources. Brands cannot deprioritize semi-/unstructured data because of a perception that these flavors of information are difficult to use. Many of us have heard that ~80% of the world’s data is unstructured, and the opportunities to harvest analytical innovation is front and center today. As customers and prospects increasingly interact with brands conversationally, all the resulting unstructured (natural language) data can be stored, contextualized, and deterministically blended with structured CRM data for richer insight potential. Recipes for data and analytical models. Retail data offers different analysis experiences from mining financial banking data. Similarly, a churn model in the discretionary fashion industry, where most customer behaviors indicating disengagement is silent, will likely differ from a churn strategy for a subscription-oriented streaming service, where churn is explicit (or observable). Recipes provide comprehensive use case-specific solutions to reduce adoption friction and increase the likelihood of success in leveraging customer insights within a marketer's workflow. Optimize customer-level treatments. Machine learning models don’t make decisions — but they do identify actionable signals in noisy customer behavioral data. It’s up to our CX and/or marketing teammate to take prescriptive insights and execute data-driven targeting/personalization. But the decision isn’t always clear. Should a brand optimize on likelihoods, engagement, propensities, or profitability? The path forward will vary by a brand's unique business model, as well as the industry vertical it operates in.   Part 1: 2024 Trends & Research Insights   Now, allow me to invite readers to check out this introductory video extracted from a recent on-demand webinar summarizing the 2024 observed trends in the customer analytics ecosystem.     Part 2: User Types, Creating Synergy, Removing Friction & Recipes Think about the magnitude of requests that come in from customer experience and marketing teams to their supporting analysts. The wish list includes actionable scoring for topics like:   Acquisition Upsell Retention Segmentation Next-best-action (or experience) Recommendations Lifetime value Pricing personalization Attribution The list could go much longer, but as many readers recognize, the point remains the same. Customer experience management has an insatiable appetite for data intelligence. This myriad of desires stratifies further when considering industry context. Let's take a moment and imagine we are in a requirements gathering meeting between two teams - data science and marketing.   Image 2: Requirements Meeting Between Data Science & Marketing The marketing and CX teams responsible for the interactions between a brand and everyday consumers speak one language. The data science and analyst group likely speaks another. Terms like acquisition, cross-sell, churn, targeting, personalization, A/B tests, conversions, and impressions are the common tongue of the martech universe. Alternatively, words such as misclassification, precision, average squared error, confusion matrices, outliers, auto tuning, neural networks, and random forests represent the language of analytics.   In the Part 2 video below, we cover an array of topics across:   Domain expertise Applied use cases Acceleration Simplification of analytically injecting data-driven intelligence into CX and marketing workflows     Part 3: Recipes 101 & Acceleration   An emerging trend to combat the ongoing analysis inefficiencies cited above involve Do-It-For-Me (DIFM) Prebuilt Recipes representing a specific ML/AI algorithm or model ensemble, processing logic, and configuration to auto-build and execute a trained solution that comprehensively solves (or improves efforts against) specific business problems. The analytical models and data engineering pipelines are ingredients of a broader recipe that get trained on data and parameter configurations to optimize a solution's ability to contribute significant value when pivoting to customer inference and marketing strategies.   Image 3: Recipes 101 & Acceleration   Topics that will be covered in the Part 3 video include:   Recipe templates applied to customer & marketing-centric use cases Accessing data where it resides to avoid duplication & redundancy Flipping the 80-20 rule upside down Democratizing analytical base table (ABT) engineering Managing data quality, privacy, imbalances & irrelevancy Moving beyond the standard/typical CDP value propositions     Part 4: Projects 101 & Welcoming Everyone Else To The Analytics Party   Now, the big question that has been posed to analytical technology companies year-after-year is whether data-driven insights can bring positive momentum to mission-critical KPIs. This brings us to another important topic because I have a message for my data science and analyst brothers/sisters. There is more to activation than just scoring your model!   Image 4: Projects 101 & Empowering CX & Marketing Teams   You want to see your analytical assets bring rewarding impact to your brand, right? You want to observe your efforts making a significant positive difference in customer journeys, correct? Then let's complete this by discussing the democratization of CX & marketing team enablement via customer journey orchestration and prescriptive activation.   Topics that will be addressed in the final Part 4 video will include:   Use case-driven recipes Moving beyond customer propensity scoring/likelihoods to optimizing marketing strategy profitability Assessment, interpretability & explainability Federation of customer analytical scoring Responsible AI & governance for CX & marketing     Our vision at SAS is to serve as the market leader in advanced audience creation & targeting, independent of channel, for enterprise customers leveraging complex, disparate data sources and wishing to consistently deliver superior understanding into their customer journeys. In other words, we want to empower brands to practice responsible marketing.    Image 5: SAS for Responsible Customer Engagement   Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here. For those who want to dive deeper into the current state of the customer analytics technologies ecosystem, check out fresh (and unbiased) research here.   ... View more

Data continues to flood every organization, both in size and in speed. Sometimes more data is better, but the challenge can be that critical decision-making information gets buried and regretfully, never gets unearthed. Skilled analytical talent with domain experience in modern marketing is the key to move a brand from reactive to proactive. Thus, varying flavors of IP, technology features and automation are critically important in accelerating the equation of time-to-insight-to-value:   Do-It-For-Me (DIFM) automation Do-It-Yourself (DIY) customization Prebuilt machine learning recipes for common customer use cases   Within the martech industry, there are several factors that contribute to the challenges surrounding brands and their decision-making processes. Obviously, customers and markets are more competitive and demanding. When you step back and reflect on this, it's a linear trend upward year-after-year when it comes to consumer expectations. This means, to satisfy that demand, it's well recognized that brands need to respond quicker, but it's often overlooked that accuracy and quality hold equivalent weights of importance.   In addition, every software vendor claims to be a CDP, do magic with AI, and activate across an ever-growing list of customer communication touchpoints. However, that doesn't mean every technology and solution provider is the same, and it's vital to look past high-level buzzwords displayed in large font sizes of slide-based presentations. Transparency is key in meeting a brand's data, analytical & activation requirements.   But don't take it from me. Fresh publications from Forrester Research released on the subjects of customer analytics (Q2, 2024) and real-time interaction management (Q1-2024) amplify the present day importance of brands leveraging technology that prioritizes performance, productivity, trust – the trifecta of undeniable AI platform value in delivering precision within martech, audience targeting and personalization.    In the context of customer propensity analysis, SAS strives to provide:   Algorithms that produce better analytic scores and accuracy Automation of machine learning that aligns with greater productivity Embedded analytics, making AI more impactful and consumable Human-like interfaces, creating approachability Trust, which is critical when using propensity scoring effectively   Image 1: Considerations For DIFM Prebuilt Recipes & Propensity Scoring Use Cases   As cited in an introductory article explaining how SAS Customer Intelligence 360 enables DIFM prebuilt machine learning recipes, there is a little secret in the customer analytics ecosystem that practitioners frequently will admit to when pressed for honest feedback. A massive proportion of customer & marketing analysts in 2024 continue to skew towards the wrong end of this workflow spectrum:   "I spend more than 80% of my time preparing data, and less than 20% actually performing analysis."   Speed bumps like this usually emerge when customer experience teams require advanced insights for propensity scoring, algorithmic segmentation, retention strategies or next-best-actions. Accelerating through this challenge has been a key area of interest at SAS, as we recognize customer experience management has an insatiable appetite for data intelligence.   Domain Expertise & Propensity-Based Use Cases   Think about the magnitude of requests that come in from customer experience and marketing teams to their supporting analysts. The wish list includes actionable propensity scoring for topics like:   Lead scoring Acquisition Upsell Retention Win-back Supervised segmentation Next-best-action (or experience)   This myriad of desires stratifies further when considering industry context. For example, retail brands commonly desire to optimize their app's shopping experience and increase the efficiency of conversion rates. Alternatively, non-profit brands want to segment their digital traffic and personalize experiences based on one's likelihood to donate vs. research and/or engage.    Propensity-Based Use Cases & DIFM Prebuilt Recipes    An emerging trend to combat the ongoing analysis inefficiencies cited above involve Do-It-For-Me (DIFM) prebuilt recipes representing a specific ML/AI algorithm or model ensemble, processing logic, and configuration to auto-build and execute a trained solution that comprehensively solves (or improves efforts against) specific business problems. The propensity models and data engineering pipelines are ingredients of a broader recipe that get trained on data and parameter configurations to optimize the solution's ability to contribute significant value when pivoting to customer inference.   Training is the process of learning patterns and insights from labeled data. A trained model represents the actionable output of a model training process, in which a set of training data was applied to the model instance. The benefits of a trained model include scoring, inference and the opportunity to create an intelligent customer service.     Supervised learning algorithms are trained using labeled examples (conversion vs. non-conversion), such as an input where the desired output is known. The learning algorithm receives a set of inputs along with the corresponding correct outputs, and the algorithm learns by comparing its actual output with correct outputs to find errors. It then modifies the model accordingly. Supervised learning is commonly used in applications where historical data predict likely future events (another way of describing the value of propensity scoring). For example, supervised learning can anticipate when an insurance customer is likely to file a claim, or identify a patron has a higher likelihood to be interested in a particular excursion (versus alternatives) for their cruise experience.   Image 2: Supervised Learning & Propensity Scoring   SAS supports two types of supervised learning problems:   Classification – When the data are being used to predict a categorical target, supervised learning is called classification. This is the case when assigning a label or indicator (for example, a customer selects to subscribe or not subscribe). When there are only two labels, this is called binary classification. When there are more than two categories, the problems are called multi-nominal classification.   Regression – When the data are being used to predict interval targets (for example, the donation amount as opposed to whether someone will or will not donate), the problems are referred to as regression.   For more details on how SAS supports supervised learning in martech, please go here. Now, with that backdrop, we can transition our thinking about customer propensity-based recipes through a stepwise (or itemized set of workflow steps) that can be templated, shared and if a user desires, customized further. Let's walk through a use case where brands can leverage SAS Customer Intelligence 360 and Viya to benefit from this concept.   SAS Customer Intelligence 360 + Viya for DIFM Prebuilt Propensity Recipes   The structured data model available to users of SAS Customer Intelligence 360 provides 1st party digital customer data views suitable for a variety of customizable analysis purposes. The data model and it's associated schema is an opportunity to leverage propensity use case-driven recipes.   Image 3: Propensity-Driven Marketing Relies On High Quality Ingredients   Here are examples of what an analyst can do with data originating from SAS Customer Intelligence 360 when instrumented on a brand's website, mobile app and/or any channel-based communications/campaigns.   Leverage session-based behavior where structured information about all visitors (identifiable and/or anonymous) including device type, browser, date, time, location and other dimensions/metrics is accessible to feed into a propensity-scoring recipe. Use journey-based customer behavior where a consolidated view of all sessions, attributes and activities across all 1st-party cookies and devices is available to derive propensity scores related to any important macro- or micro-goals. Detect bias and mitigate (or remove) such negative and unintended effects on customer propensity scores. Build a product offer engine rank-ordered by propensities for data-at-rest or data-in-motion use cases leveraging 1st party customer behaviors. Benefit from a variety of supervised learning techniques embedded within propensity-oriented recipes.   The list above is simply a sampling of ideas on how the SAS Customer Intelligence 360 data model can be leveraged for customer propensity use cases. Next, let's take a high-level look at the types of tables that reside within the data model's schema.   Image 4: Data Model Snippet & Contextualization   As analysts within SAS, the concept of accessible data of any type within SAS Libraries applies to users of both SAS 9 and SAS Viya. We are accustomed to creating connections to where data resides within on-prem and cloud-based locations. Using this same concept, SAS Customer Intelligence 360 makes 1st party digital interaction data between consumers and a brand available to SAS analysts in the same exact manner.    The screenshot below shows an instance of SAS Studio & Flows within Viya 4. On the left-side, SAS analysts will immediately recognize a library with data tables. One specific table, entitled FIN_ACTIVITY_CONVERSION is opened to highlight the available dimensions, metrics, date and time-based variables. The Pre-Process Code node in the first swim lane of the Flow contains a short snippet of code to enable the analyst to create a data connection. It simply contains the instructions/credentials highlighting how SAS Customer Intelligence 360 data residing within a Snowflake environment can be accessed in a repeatable fashion. After running the node, the connection to the server residing in Snowflake is established and the library containing all of the data model's tables are available for analysis. Outside of Snowflake, readers should know the same approach can be used for SAS Customer Intelligence 360 data landed for long-term storage in other platforms (hyperscalers, on-prem databases, etc.).   Image 5: Creating A Customer Intelligence 360 Data Connection In Viya 4   Analysts are not limited to simply selecting ALL the data. The next screenshot shown below opens up a second swim lane within this Flow which provides no-code user features to select:   Specific area(s) of the data model (dimension, event or reporting)  Acute categories of data Data-at-rest snapshots filtered on time windows Image 6: No-code User Features For Selecting Customer Intelligence 360 Data Views   The beauty of this example is enabled in SAS Viya 4 using a Custom Step feature. For readers unfamiliar with this, SAS Studio ships with several SAS Steps, which are available from the SAS Steps tab in the Steps pane (see screenshot below). For example, one use case for a Custom Step enables analysts to create an interface for no-code users at your organization to complete a specific task. Custom Steps can be saved to your SAS Server or in SAS Content.   The SAS Server is any file system that SAS can access. Users can save Custom Steps on a local network, in Git, and other shared mounted systems. Analysts can access these steps from the Explorer pane within SAS Studio on Viya. Steps that are saved can be shared with other SAS users at your organization.   Image 7: Custom Steps For Accessing & Selecting Data From SAS Customer Intelligence 360   The Shared tab shown on the left-side in the screenshot above lists all the Custom Steps available in my demo environment. It includes Steps for simplifying the process of accessing data originating from SAS Customer Intelligence 360. This list includes any Steps that an analyst authored, any steps that are saved at a location where users have access, and any Steps that have been shared. This is a critical moment in this article's reading.   For every brand that selects to use SAS Customer Intelligence 360 and Viya 4, the deployment of the technology can include Custom Steps to remove any high-code requirements and users can accelerate their access of important 1st party digital customer data originating from interactions with a website, app, marketing channels and paid media. Removing the friction of working with SAS APIs (although high-code users are still welcome to leverage this alternative) is the value proposition to accelerate usage and amplify analysis efforts.   For brands who have this no-code preference, simply communicate this request to your supporting SAS account/support team members, and the Custom Steps highlighted in this article can be shared with your user team.   Moving on, now that an analyst has selected the relevant data from SAS Customer Intelligence 360 to work with, we can explore the value propositions of customer propensity recipes and the associated ingredients for leveraging ABTs (analytic base tables) & supervised learning algorithms. The benefit of prebuilt machine learning recipes for propensity scoring can be very helpful to data scientists and developers so they don’t have to start from scratch. If users prefer, they can adapt the proposed prebuilt recipe described below to their needs (or use it as inspiration to start from scratch to build your own custom recipe).  Once analysts train/tune a recipe, creating an intelligent activation layer doesn’t require a developer - just a few clicks and marketers are enabled to build targeted, personalized customer experiences using the SAS Customer Intelligence 360 Audiences API. More on this in a moment...   The concept of a recipe may be a new term to some readers. Let's break this down:   Recipe A recipe is a term for a data-driven solution for a particular use case and is a holistic asset representing a specific ML/AI algorithm or ensemble, processing logic, and configuration required to build and execute a trained model and hence help solve specific business problems via inference. Model A model (prediction/ML/AI) is a recipe ingredient that is trained using historical data and configurations to solve for a specific use case. Training Training is the process of learning patterns and insights from labeled data. Trained Model A trained model is a recipe ingredient representing the executable output, in which a set of training data was applied to the algorithmic solution. The trained model is suitable for scoring and creating a customer treatment strategy.  Scoring Scoring is a recipe ingredient representing the process of generating actionable insights (inference) from data using a trained model. CI360 Audiences Service A deployed service is a recipe ingredient which exposes functionality of an advanced algorithm (originating from SAS and/or open-source) through an API so that it can be consumed and activated by SAS Customer Intelligence 360.   With that said, let's dive into an example. We will start with a recipe for generating customer propensities, where SAS strives to accelerate how our users can make intelligent decisions regarding customer treatments.   Image 8: Input Tables & Mappings For Customer Propensity Recipe   The customer propensity recipe begins with input tables originating from SAS Customer Intelligence 360's data model. For this example (screenshot above), we are using five tables:   Product Views: The PRODUCT_VIEWS table provides information about what products visitors view. The table is sourced from the product view event in SAS Customer Intelligence 360. Session Details: The SESSION_DETAILS table provides information about web and mobile sessions. This information includes a wide range of data about your users. For example, you can use this table to determine which browsers are used to access your content, identify the geographic location of web or mobile app users, and identify the traffic sources that brought a user to your website. Identity Map: The IDENTITY_MAP table stores the associations between anonymous users and identified customers, and is updated when an anonymous user is identified by SAS Customer Intelligence 360. An anonymous user can be identified through an identity event or a data import. These examples illustrate when an entry is added to this table: When an anonymous user signs in to your site (or app) and triggers an event that captures the login_id or customer_id, that anonymous user becomes an identified user. When the same person is identified by SAS Customer Intelligence 360 across different devices, an entry is added. For example, a user first signs in to your mobile application, then the same user navigates to your site without signing in. The user is identified on the mobile app, but anonymous on the site. After the user signs in to your site, SAS Customer Intelligence 360 associates the anonymous user with the identified user that logged in to the mobile application. If you import data that has information in multiple ID columns and one of those columns is already known in SAS Customer Intelligence 360, the system merges the IDs and their attributes into one ID. For example, SAS Customer Intelligence 360 has a customer_id for a user from an identity event and a subject_id for the same user from an external event. To associate these two IDs, you can import data that contains all of the identifying information for this user, and SAS Customer Intelligence 360 merges all the known information about this user to a single customer identity. When an entry is added to the Identity Map table, any attributes that are associated with anonymous user become associated with the identified user. Identity Attributes: The IDENTITY_ATTRIBUTES table contains information about a person’s identity. Data is stored in this table when any of these actions occur: Identity events created in SAS Customer Intelligence 360 are triggered, and the events contain a customer_id or login_id. This includes identity events sent through our JavaScript API and the mobile SDK.  External events are sent to SAS Customer Intelligence 360 using either the external API gateway or the on-premises API gateway. The events must contain a subject_id, login_id, or customer_id.  1st party customer data is imported into SAS Customer Intelligence 360. Custom Events: The CUSTOM_EVENTS table captures user-defined custom events for which SAS Customer Intelligence 360 does not have a standard definition.   For more information about the SAS Customer Intelligence 360 data model, please visit the documentation pages here.  The next couple of steps in the Propensity Recipe involve the use of native data transformation capabilities within SAS.    Image 8: Data Transformations For Customer Propensity Recipe   Keep in mind, a Flow is a sequence of operations on data. Data and operations are represented in SAS by steps that users can access from the Steps section of the left-navigation pane. Each step in a flow is represented by a node on the flow canvas. The nodes on this Flow canvas above represent some of the steps that are available in SAS Studio. We are simply using Query node capabilities to perform various types of deterministic table joins to connect customer identities with behaviors like product viewing, session activities and custom events.   SAS Studio is shipped with many predefined steps that include queries and data transformations. The steps are organized into categories that indicate the function that they perform. With respect to the Query step acutely, users can leverage this node to select, join, filter, and sort columns from a table in a Flow.   Once the input tables have collectively been mapped to customer identities, we proceed with transposing the products customers viewed and custom event definitions they met. The Transpose Data step turns selected columns of an input table into the rows of an output table. Our desire is to manipulate the input data and reshape them as predictors in anticipation of performing supervised learning and propensity modeling.   Image 9: Data Transpositions For Customer Propensity Recipe   After these nodes complete processing, the formation of a model-ready ABT is now ready for algorithmic propensity modeling within this demo example. The reason this milestone is important is because if users can accelerate to this step of the process, it reduces the time-to-insight issue (80-20 trend) cited at the beginning of this article. In addition, this is an example of how SAS provides robust CDP+ capabilities, matching the typical requirements of a CDP solution today in martech while also extending incremental benefits of a CDP to the data science, analyst and marketing communities.   Image 10: ABT Creation For Customer Propensity Recipe   For simplicity, we will use imputation and a tree-based algorithm in the next two steps.     The VARIMPUTE procedure performs numeric variable imputation in SAS Viya. Imputation is a common step in data preparation. The VARIMPUTE procedure can replace numeric missing values with a specified value, with the mean or median of the non-missing values, or with some random value between the minimum value and the maximum value of the non-missing values.   The TREESPLIT procedure builds tree-based statistical models for classification and regression in SAS Viya. The procedure produces a classification tree, which models a categorical response, or a regression tree, which models a continuous response. Both types of trees are called decision trees, because the model is expressed as a series of if-then statements.    The predictor variables for tree models can be categorical or continuous. The set of all possible combinations of the predictor variables is called the predictor space. The model is based on partitioning the predictor space into nonoverlapping segments, which correspond to the terminal nodes (called leaves) of the tree. Partitioning is done repeatedly, starting with the root node, which contains all the data, and continuing until a stopping criterion is met. At each step, the parent node is split into child nodes by selecting a predictor variable and a split value for that variable that minimize the variability, according to a specified measure, in the response variable across the child nodes. Various measures, such as the Gini index, entropy, and residual sum of squares, can be used to assess candidate splits for each node. The selected predictor variable and its split value are called the primary splitting rule.   Tree models are built from training data for which the response values are known, and these models are subsequently used to score (classify or predict) response values for new data. For classification trees, the most frequent response level of the training observations in a leaf is used to classify observations in that leaf. For regression trees, the average response of the training observations in a leaf is used to predict the response for observations in that leaf. The splitting rules that define the leaves provide the information that is needed to score new data.   The process of building a decision tree begins with growing a large, full tree. The full tree can overfit the training data, resulting in a model that does not adequately generalize to new data. To prevent overfitting, the full tree is often pruned back to a smaller subtree that balances the goals of fitting training data and predicting new data. Two commonly applied approaches for finding the best subtree are cost-complexity pruning (Breiman et al. 1984) and C4.5 pruning (Quinlan 1993). Compared with other regression and classification methods, tree models have the advantage that they are easy to interpret and visualize. Tree-based methods scale well to large data, and they offer various methods of handling missing values, including surrogate splits.   Readers should be reminded that any supervised learning algorithm (or ensemble of algorithms) appropriate for propensity-driven use cases could be leveraged in SAS for the modeling ingredient of this recipe.    Image 11: Authoring A Tree-Based Model For Customer Propensity Recipe   Although comprehensive use case-driven recipes can be shared between SAS and our user community, if an analyst wanted to author their own custom propensity model, SAS enables no/low-code users (not just high-code users) to leverage a GUI interface to assign the relevant predictors, parameters, auto tuning method, and other criterion properties. As these inputs are made, the right-side of the screenshot above highlights how SAS auto-scripts the programming language to run the custom model.   Users can leverage the code-to-flow feature to then map in the custom authored analysis into the Flow as a Step.   Image 12: Code-To-Flow For Customer Propensity Recipe   The result completes the Flow's propensity recipe ingredient for running the tree-based model (shown below).   Image 13: Algorithmic Modeling Ingredient For Customer Propensity Recipe   Now, the big question that has been posed to analytical technology companies year-after-year from our customers is whether data-driven insights can bring positive momentum to mission-critical KPIs. This brings us to another important recipe ingredient because I have a message for my data science and analyst brothers and sisters:   There is more to activation than just scoring your model!   You want to see your analytical assets bring rewarding impact to your brand, right? You want to observe your efforts making a significant positive difference in customer journeys, correct? Then let's complete this by discussing the last recipe ingredient that ties into destinations, journey orchestration and prescriptive activation. In the screenshot below, the last Swimlane of the Propensity Recipe Flow is highlighted.   The first node provides us a view into the propensity scoring that resulted from running the tree-based model step. The first column entitled Subject_ID is the unique (and cloud-secure) identifier that enables SAS Customer Intelligence 360 users to communicate, target or personalize experiences on websites, apps and channels with individuals or audiences. The propensity scoring is embodied in the second column entitled P_CE_InterestedInCreditCards (or the propensity score associated with a customer's likelihood to be interested in a credit card offer from a financial services brand). For an overview on Identities in SAS Customer Intelligence 360, please go here to learn more.   Image 14: Scored Audience Table For Customer Propensity Recipe   Analysts should consider this table as the "prescription" for our marketing counterparts. The marketer has a need or desire for intelligent scoring of the customers they want to target with a treatment (as well as exclude those who are deemed irrelevant). The recipe scoring is the prescription (or conduit) between data science and marketing for the given use case. Using the SAS Customer Intelligence 360 Audiences API, this last Swimlane contains two remaining steps:   Query and share the relevant attributes between Viya and Customer Intelligence 360. In the screenshot below, we select to include the Subject_ID (unique & encrypted identifier for a customer) and the associated propensity scoring. All other attributes are removed since they are only relevant to data science and the associated quality of the propensity analysis. The removal of this information does not impact the marketer's activation workflow.   Image 15: Controls For Selecting Relevant Data To Upload Into Customer Intelligence 360   Push the scored data to the correct cloud-based tenant using the Audiences API where your brand's instance of SAS Customer Intelligence 360 lives. Image 16: No-code Interface To Connect To Audiences API & Customer Intelligence 360   A large reason we are excited to share recipes with our user community is validated again in this last Custom Step of the demo. For those readers who have worked with APIs, you understand they typically require high-coding skills. In essence, what we have shown here removes the friction of an analyst having to author this code themselves and simply provide inputs in a few clicks. So, what is the result after running this final Swimlane? Users of SAS Customer Intelligence 360 will see the scored Audience within the software available for activation.   Image 17: Propensity Recipe Audience Uploading Into Customer Intelligence 360   Based on a brand's SAS environment, this process will take either seconds or minutes. Once the processing is completed, users will see an updated status with additional details (as exemplified below).   Image 18: Propensity Recipe Audience Activated In Customer Intelligence 360   From here, how a brand takes advantage of analytically-derived audiences for any recipe's use case (not just propensity scoring) can be activated across one or multiple channels. Here is a sampling of what is possible:   Image 19: Examples Of Supported Touchpoints In Customer Intelligence 360   Once a touchpoint (or task) is selected (we will leverage Facebook Ads for this example), users can leverage 1st party customer data from a variety of options, including Audiences sourced from analytical activities originating from SAS Viya to support the best practices of responsible marketing.    Image 20: Leveraging Audiences For Facebook Ads Targeting In Customer Intelligence 360   After clicking on the Audiences tile button, users have the option to observe the Propensity Recipe Audience metadata.   mage 21: Propensity Recipe Audience Metadata View   In addition, users can preview the the audience data itself. Image 22: Previewing Audience Data In SAS Customer Intelligence 360   From here, users can select to target one or multiple segments that originated from the propensity analysis. For example, a gentle touch in delivering prescriptive Audiences to marketers can be to filter the propensity values associated with a Subject ID to only the customers that should be targeted. Instead of using the default output values (which include propensities both above and below a desired threshold), an analyst can simplify the experience for the marketer. For example, after interpreting the propensity analysis, an analyst may determine to filter out the low scores. Prior to uploading the Propensity Recipe Audience from SAS Viya to Customer Intelligence 360, an analyst would simply apply a filter to isolate the Subject IDs that are relevant for inclusion.   Image 23: Filtering Propensity Scores and Simplifying Marketing Activation Workflow   After applying the filter, an analyst can run the Audiences API to only load the High Propensity Audience.   Image 24: High Propensity Recipe Audience Available For Marketer   Using descriptive Audience naming conventions relevant to a marketer's workflow can further minimize adoption and activation friction.   Image 25: High Propensity Audience Available For Selection For Facebook Ads Targeting   Advancing to the next step of the marketer's workflow, the user would set up the High Propensity Audience for Facebook Ads targeting in the same manner they would target any other customer segment or group.   Image 26: Facebook Ads Connector Within SAS Customer Intelligence 360   Building upon this vision, Activity Maps for customer journey orchestration across multiple touchpoints is also "in-scope" as a value proposition on leveraging use case-driven recipes and activating attractive customer audiences.   Image 27: Leveraging Audiences For Multi-touchpoint Targeting Strategies In Customer Intelligence 360   In conclusion, the important takeaways from this article include:   The introduction of Propensity Recipes in SAS. A detailed walkthrough of using/adapting a propensity recipe across SAS Viya and Customer Intelligence 360. Other recipes exist across the areas of acquisition, upsell, retention, next-best-action (or experience), recommendations, lifetime value, pricing personalization, attribution and more. If you are interested in leveraging any of these proposed recipes, please reach out to your SAS support team and the sharing can begin!   Remember, there is more to activation than just scoring your model!    Image 28: Destinations & Analytical-Driven Journey Orchestration In Customer Intelligence 360   Our vision at SAS is to serve as the market leader in advanced audience creation & targeting, independent of channel, for enterprise customers leveraging complex, disparate data sources and wishing to consistently deliver superior understanding into their customer journeys. In other words, we want to empower brands to practice responsible marketing.    Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here. For those who want to dive deeper into the current state of the customer analytics technologies ecosystem, check out fresh (and unbiased) research here.   ... View more

Brands aspire to strategically manage their business through prioritizing customer convenience. This involves  anticipating and responding to customer needs, while manifesting in proactively delivered, seamless, and unobtrusive interactions. The intent is to provide personalization, assistance and valued services. However, there is a little secret in the customer analytics ecosystem that practitioners frequently will admit to when pressed for honest feedback. A massive proportion of customer & marketing analysts in 2024 continue to skew towards the wrong end of this workflow spectrum:   "I spend more than 80% of my time preparing data, and less than 20% actually performing analysis."   Speed bumps like this usually emerge when customer experience teams require advanced insights for propensity scoring, algorithmic segmentation,  retention strategies or next-best-actions. For example, have you ever tried to extract event (or HIT) data from your preferred marketing cloud vendor because your team grew frustrated by the limitations of digital BI & reporting (in other words, descriptive measurement and diagnostic analysis capabilities)? Those who have experienced this have witnessed firsthand the data extracts are not formatted for machine learning or AI use cases, and the time-to-value expense becomes heavily negative in the complex efforts to re-engineer that data.   Accelerating through this challenge has been a key area of interest at SAS, and the solution is referred to as analytic base tables (or ABTs). It is the process of organizing data into a flat table schema that's used by analysts for building analytical models and scoring (predicting/inference) the future behavior of a customer. A single record in this table represents the subject of the prediction (such as a customer or anonymous visitor) and stores all data (variables, features or predictors) describing this subject.   But there is more to efficiently delivering analytically-driven value downstream to teammates involved with customer experience management. It isn't just about getting customer data ready for modeling, and also involves:   Domain Expertise & Applied Use Cases   Think about the magnitude of requests that come in from customer experience and marketing teams to their supporting analysts. The wish list includes actionable scoring for topics like:   Acquisition Upsell Retention Segmentation Next-best-action (or experience) Recommendations Lifetime value Pricing personalization Attribution   The list could go much longer, but as many readers recognize, the point remains the same. Customer experience management has an insatiable appetite for data intelligence. This myriad of desires stratifies further when considering industry context. For example, retail brands commonly desire to optimize their app's shopping experience and increase the lifetime value of mobile users. Alternatively, financial service brands want to deepen customer relationships and improve stickiness through recommendation systems for upsell opportunities. Finally, entertainment brands with subscription services obsess about retention (or churn), identifying meaningful friction points within customer journeys, and alter strategic treatments on minimizing these customer events.   Image 1: Considerations For Prebuilt Recipes & Customer Intelligence   Use Cases & DIFM Prebuilt Recipes    An emerging trend to combat the ongoing analysis inefficiencies cited above involve Do-It-For-Me (DIFM) Prebuilt Recipes representing a specific ML/AI algorithm or model ensemble, processing logic, and configuration to auto-build and execute a trained solution that comprehensively solves (or improves efforts against) specific business problems. The analytical models and data engineering pipelines are ingredients of a broader recipe that get trained on data and parameter configurations to optimize the solution's ability to contribute significant value when pivoting to customer inference.   Training is the process of learning patterns and insights from labeled data. A trained model represents the actionable output of a model training process, in which a set of training data was applied to the model instance. The benefits of a trained model include scoring, inference and creating an intelligent customer service.     Another way to think about recipes is through a stepwise (or itemized set of workflow steps). Let's walk through a use case where brands that leverage SAS Customer Intelligence 360 and Viya can benefit from this concept.   SAS Customer Intelligence 360 + Viya for DIFM Prebuilt Recipes   The structured data model available to users of SAS Customer Intelligence 360 provides 1st party digital customer data views suitable for a variety of customizable analysis purposes. The data model and it's associated schema is an opportunity to leverage use case-driven recipes.   Image 2: Analytical-Driven Integrated Marketing Relies On High Quality Ingredients   Here are examples of what an analyst can do with data originating from SAS Customer Intelligence 360 when instrumented on a brand's website, mobile app and/or any channel-based communications/campaigns.   Analyze session-based customer behavior where structured information about all visitors (identifiable and/or anonymous) including device type, browser, date, time, location and other dimensions/metrics is accessible. Analyze journey-based customer behavior where a consolidated view of all sessions, attributes and activities across all 1st-party cookies and devices is available.  Model how customer behavior correlates with macro- and micro-defined goals and events using supervised learning techniques. Perform attribution and/or media mix analysis to determine opportunities to incrementally lift the ROI of paid media investments. Build a recommendation system for data-at-rest or data-in-motion use cases leveraging customer viewing behaviors of products and/or services offered. Detect bias and mitigate (or remove) such negative and unintended effects on customer insights.     The list above is simply a sampling of ideas on how the SAS Customer Intelligence 360 data model can be leveraged. Next, let's take a high-level look at the types of tables that reside within the data model's schema.   Image 3: Data Model Snippet & Contextualization   As analysts within SAS, the concept of accessible data of any type within SAS Libraries applies to users of both SAS 9 and SAS Viya. We are accustomed to creating connections to where data resides within on-prem and cloud-based locations. Using this same concept, SAS Customer Intelligence 360 makes 1st party digital interaction data between consumers and a brand available to SAS analysts in the same exact manner.    The screenshot below shows an instance of SAS Studio & Flows within Viya 4. On the left-side, SAS analysts will immediately recognize a library with data tables. One specific table, entitled FIN_ABT_Attribution is opened to highlight the available dimensions, metrics, date and time-based variables. The Pre-Process Code node in the first swim lane of the Flow contains a short snippet of code to enable the analyst to create a data connection. It simply contains the instructions/credentials highlighting how SAS Customer Intelligence 360 data residing within a Snowflake environment can be accessed in a repeatable fashion. After running the node, the connection to the server residing in Snowflake is established and the library containing all of the data model's tables are available for analysis. Outside of Snowflake, readers should know the same approach can be used for SAS Customer Intelligence 360 data landed for long-term storage in other platforms (hyperscalers, on-prem databases, etc.).   Image 4: Creating A Customer Intelligence 360 Data Connection In Viya 4   Analysts are not limited to simply selecting ALL the data. The next screenshot shown below opens up a second swim lane within this Flow which provides no-code user features to select:   Specific area(s) of the data model (dimension, event or reporting)  Acute categories of data Data-at-rest snapshots filtered on time windows Image 5: No-code User Features For Selecting Customer Intelligence 360 Data Views   The beauty of this example is enabled in SAS Viya 4 using a Custom Step feature. For readers unfamiliar with this, SAS Studio ships with several SAS Steps, which are available from the SAS Steps tab in the Steps pane (see screenshot below). For example, one use case for a Custom Step enables analysts to create an interface for no-code users at your organization to complete a specific task. Custom Steps can be saved to your SAS Server or in SAS Content.   The SAS Server is any file system that SAS can access. Users can save Custom Steps on a local network, in Git, and other shared mounted systems. Analysts can access these steps from the Explorer pane within SAS Studio on Viya. Steps that are saved can be shared with other SAS users at your organization.   Image 6: Custom Steps For Accessing & Selecting Data From SAS Customer Intelligence 360 The Shared tab shown on the left-side in the screenshot above lists all the Custom Steps available in my demo environment. It includes Steps for simplifying the process of accessing data originating from SAS Customer Intelligence 360. This list includes any Steps that an analyst authored, any steps that are saved at a location where users have access, and any Steps that have been shared. This is a critical moment in this article's reading.   For every brand that selects to use SAS Customer Intelligence 360 and Viya 4, the deployment of the technology can include Custom Steps to remove any high-code requirements and users can accelerate their access of important 1st party digital customer data originating from interactions with a website, app, marketing channels and paid media. Removing the friction of working with SAS APIs (although high-code users are still welcome to leverage this alternative) is the value proposition to accelerate usage and amplify analysis efforts.   For brands who have this no-code preference, simply communicate this request to your supporting SAS account/support team members, and the Custom Steps highlighted in this article can be shared with your user team.   Moving on, now that an analyst has selected the relevant data from SAS Customer Intelligence 360 to work with, we can explore recipe types and associated ingredients for leveraging use-case applicable ABTs (analytic base tables) & algorithms. This allows us to introduce prebuilt machine learning recipes for common business needs, like segmentation, propensity, and recommenders, so data scientists and developers don’t have to start from scratch. If users prefer, they can adapt any of the proposed prebuilt recipes described below to their needs (or start from scratch to build a custom recipe).  Once analysts train/tune a recipe, creating an intelligent activation layer doesn’t require a developer - just a few clicks and marketers/customer experience are enabled to build targeted, personalized customer experiences using the SAS Customer Intelligence 360 Audiences API. More on this in a moment...   The concept of a recipe may be a new term to some readers. Let's break this down:   Recipe A recipe is a term for a data-driven solution for a particular use case and is a holistic asset representing a specific ML/AI algorithm or ensemble, processing logic, and configuration required to build and execute a trained model and hence help solve specific business problems via inference. Model A model (prediction/ML/AI) is a recipe ingredient that is trained using historical data and configurations to solve for a specific use case. Training Training is the process of learning patterns and insights from labeled data. Trained Model A trained model is a recipe ingredient representing the executable output, in which a set of training data was applied to the algorithmic solution. The trained model is suitable for scoring and creating a customer treatment strategy.  Scoring Scoring is a recipe ingredient representing the process of generating insights (inference) from data using a trained model. CI360 Audiences Service A deployed service is a recipe ingredient which exposes functionality of an advanced algorithm (originating from SAS and/or open-source) through an API so that it can be consumed and activated by SAS Customer Intelligence 360.   With that said, let's dive into an example. We will start with a recipe for customer segmentation, where SAS strives to provide our users:   Algorithms that produce better analytic scores and accuracy. Automation of machine learning that aligns with greater productivity. Human-like interfaces, creating approachability. Trust, which is critical when using segmentation effectively. These concepts represent our guiding light on how information & derived insight are used to make intelligent decisions regarding customer treatments.   Image 7: Input Tables & Mappings For Customer Segmentation Recipe   The customer segmentation recipe begins with input tables originating from SAS Customer Intelligence 360's data model. For this example (screenshot above), we are using five tables:   Product Views: The PRODUCT_VIEWS table provides information about what products visitors view. The table is sourced from the product view event in SAS Customer Intelligence 360. Session Details: The SESSION_DETAILS table provides information about web and mobile sessions. This information includes a wide range of data about your users. For example, you can use this table to determine which browsers are used to access your content, identify the geographic location of web or mobile app users, and identify the traffic sources that brought a user to your website. Identity Map: The IDENTITY_MAP table stores the associations between anonymous users and identified customers, and is updated when an anonymous user is identified by SAS Customer Intelligence 360. An anonymous user can be identified through an identity event or a data import. These examples illustrate when an entry is added to this table: When an anonymous user signs in to your site (or app) and triggers an event that captures the login_id or customer_id, that anonymous user becomes an identified user. When the same person is identified by SAS Customer Intelligence 360 across different devices, an entry is added. For example, a user first signs in to your mobile application, then the same user navigates to your site without signing in. The user is identified on the mobile app, but anonymous on the site. After the user signs in to your site, SAS Customer Intelligence 360 associates the anonymous user with the identified user that logged in to the mobile application. If you import data that has information in multiple ID columns and one of those columns is already known in SAS Customer Intelligence 360, the system merges the IDs and their attributes into one ID. For example, SAS Customer Intelligence 360 has a customer_id for a user from an identity event and a subject_id for the same user from an external event. To associate these two IDs, you can import data that contains all of the identifying information for this user, and SAS Customer Intelligence 360 merges all the known information about this user to a single customer identity. When an entry is added to the Identity Map table, any attributes that are associated with anonymous user become associated with the identified user. Identity Attributes: The IDENTITY_ATTRIBUTES table contains information about a person’s identity. Data is stored in this table when any of these actions occur: Identity events created in SAS Customer Intelligence 360 are triggered, and the events contain a customer_id or login_id. This includes identity events sent through our JavaScript API and the mobile SDK.  External events are sent to SAS Customer Intelligence 360 using either the external API gateway or the on-premises API gateway. The events must contain a subject_id, login_id, or customer_id.  1st party customer data is imported into SAS Customer Intelligence 360. Custom Events: The CUSTOM_EVENTS table captures user-defined custom events for which SAS Customer Intelligence 360 does not have a standard definition.   For more information about the SAS Customer Intelligence 360 data model, please visit the documentation pages here.  The next couple of steps in the Segmentation Recipe involve the use of native data transformation capabilities within SAS.    Image 7: Data Transformations For Customer Segmentation Recipe   Keep in mind, a Flow is a sequence of operations on data. Data and operations are represented in SAS by steps that users can access from the Steps section of the left-navigation pane. Each step in a flow is represented by a node on the flow canvas. The nodes on this Flow canvas above represent some of the steps that are available in SAS Studio. We are simply using Query node capabilities to perform various types of deterministic table joins to connect customer identities with behaviors like product viewing, session activities and custom events.   SAS Studio is shipped with many predefined steps that include queries and data transformations. The steps are organized into categories that indicate the function that they perform. With respect to the Query step acutely, users can leverage this node to select, join, filter, and sort columns from a table in a Flow.   Once the input tables have collectively been mapped to customer identities, we proceed with transposing the products customers viewed and custom event definitions they met. The Transpose Data step turns selected columns of an input table into the rows of an output table. Our desire is to manipulate the input data and reshape them as predictors in anticipation of performing (un)supervised segmentation.   Image 8: Data Transpositions For Customer Segmentation Recipe   After these nodes complete processing, the formation of a model-ready ABT is now ready for algorithmic segmentation within this demo example. The reason this milestone is important is because if users can accelerate to this step of the process, it reduces the time-to-insight issue (80-20 trend) cited at the beginning of this article. In addition, this is an example of how SAS provides robust CDP+ capabilities, matching the typical requirements of a CDP solution today in martech while also extending incremental benefits of a CDP to the data science, analyst and marketing communities.   Image 9: ABT Creation For Customer Segmentation Recipe   For simplicity, we will use a K-Means clustering algorithm in the next step.  The KCLUS procedure performs clustering (a common step in data exploration and/or unsupervised segmentation) in SAS Viya. Analysts can use the KCLUS procedure to read and write data in distributed form, and to perform clustering and scoring in parallel by making full use of multicore computers or distributed computing environments. The KCLUS procedure performs a cluster analysis on the basis of distances that are computed from quantitative or qualitative variables (or both). The observations are divided into clusters such that every observation belongs to one and only one cluster (or segment). The KCLUS procedure uses the k-means algorithm for clustering interval input variables, uses the k-modes algorithm for clustering nominal input variables, and uses k-prototypes algorithm for clustering mixed input that contains both interval and nominal variables. Readers should be reminded that any algorithm (or ensemble of algorithms) appropriate for segmentation could be leveraged for the modeling ingredient of the recipe.   Image 10: Authoring A Clustering Model For Customer Segmentation Recipe   Although comprehensive use case-driven recipes can be shared between SAS and our user community, if an analyst wanted to author their own custom segmentation model, SAS enables no/low-code users (not just high-code users) to leverage a GUI interface to assign the relevant data inputs, standardization methods, distance measures, number of clusters, and other criterion properties. As these inputs are made, the right-side of the screenshot above highlights how SAS auto-scripts the programming language to run the custom model.   Users can leverage the code-to-flow feature to then map in the custom authored analysis into the Flow as a Step.   Image 11: Code-To-Flow For Customer Segmentation Recipe   The result completes the Flow's segmentation recipe ingredient for running the clustering model (shown below).   Image 12: Algorithmic Modeling Ingredient For Customer Segmentation Recipe   Now, the big question that has been posed to analytical technology companies year-after-year from our customers is whether data-driven insights can bring positive momentum to mission-critical KPIs. This brings us to another important recipe ingredient because I have a message for my data science and analyst brothers and sisters:   There is more to activation than just scoring your model!   You want to see your analytical assets bring rewarding impact to your brand, right? You want to observe your efforts making a significant positive difference in customer journeys, correct? Then let's complete this by discussing the last recipe ingredient that ties into destinations, journey orchestration and prescriptive activation. In the screenshot below, the last Swimlane of the Segmentation Recipe Flow is highlighted.   The first node provides us a view into the segmentation scoring that resulted from running the clustering model step. The first column entitled Subject_ID is the unique (and cloud-secure) identifier that enables SAS Customer Intelligence 360 users to communicate, target or personalize experiences on websites, apps and channels with individuals or audiences. The segmentation scoring is embodied in the second column entitled Cluster_ID. For an overview on Identities in SAS Customer Intelligence 360, please go here to learn more.   Image 13: Scored Audience Table For Customer Segmentation Recipe   After confirming the customer data has been scored, consider this table the "prescription" for our marketing counterparts. The marketer has a need or desire for intelligent segmentation of the customers they want to target with a treatment (as well as exclude those who are deemed irrelevant). The recipe scoring is the prescription (or conduit) between data science and marketing for the given use case. Using the SAS Customer Intelligence 360 Audiences API, this last Swimlane contains two remaining steps:   Query and share the relevant attributes between Viya and Customer Intelligence 360. In the screenshot below, we select to include the Subject_ID (unique & encrypted identifier for a customer) and the associated segmentation. All other attributes are removed since they are only relevant to data science and the associated quality of the segmentation analysis. The removal of this information does not impact the marketer's activation workflow.   Image 14: Controls For Selecting Relevant Data To Upload Into Customer Intelligence 360   Push the scored data to the correct cloud-based tenant using the Audiences API where your brand's instance of SAS Customer Intelligence 360 lives.   Image 15: No-code Interface To Connect To Audiences API & Customer Intelligence 360   A large reason we are excited to share recipes with our user community is validated again in this last Custom Step of the demo. For those readers who have worked with APIs, you understand they typically require high-coding skills. In essence, what we have shown here removes the friction of an analyst having to author this code themselves and simply provide inputs in a few clicks. So, what is the result after running this final Swimlane? Users of SAS Customer Intelligence 360 will see the scored Audience within the software available for activation.   Image 16: Segmentation Recipe Audience Uploading Into Customer Intelligence 360   Based on a brand's SAS environment, this process will take either seconds or minutes. Once the processing is completed, users will see an updated status with additional details (as exemplified below).   Image 17: Segmentation Recipe Audience Activated In Customer Intelligence 360   From here, how a brand takes advantage of analytically-derived audiences for any recipe's use case (not just segmentation) can  be activated across one or multiple channels. Here is a sampling of what is possible:   Image 18: Examples Of Supported Touchpoints In Customer Intelligence 360   Once a touchpoint (or task) is selected (we will leverage Google Ads for this example), users can leverage 1st party customer data from a variety of options, including Audiences sourced from analytical activities originating from SAS Viya to support the best practices of responsible marketing.    Image 19: Leveraging Audiences For Google Ads Targeting In Customer Intelligence 360 After clicking on the Audiences tile button, users have the option to observe the Segmentation Recipe Audience metadata.   Image 20: Segmentation Recipe Audience Metadata View   In addition, users can preview the the audience data itself.   Image 21: Previewing Audience Data In SAS Customer Intelligence 360   From here, users can select to target one or multiple segments that originated from the clustering analysis. For example, a gentle touch in delivering prescriptive Audiences to marketers can be a translation of the Cluster values associated with a Subject ID. Instead of using the default output values of 1, 2, 3 and 4, an analyst can simplify the experience for the marketer and use more contextually relevant tags. For example, after interpreting the cluster analysis, an analyst may determine the following naming convention:   Cluster 1 = High value Cluster 2 = Moderate value Cluster 3 = Low value Cluster 4 = Exclusion   Prior to uploading the Segmentation Recipe Audience from SAS Viya to Customer Intelligence 360, an analyst would simply apply a filter to isolate the High Value Audience.   Image 22: Filtering Cluster IDs and Simplifying Marketing Activation Workflow   After applying the filter, an analyst can run the Audiences API to only load the High Value Audience.   Image 23: High Value Segmentation Recipe Audience Available For Marketer   Using descriptive Audience naming conventions relevant to a marketer's workflow can further minimize adoption and activation friction.   Image 24: High Value Audience Available For Selection For Google Ads Targeting   Advancing to the next step of the marketer's workflow, the user would set up the High Value Audience for Google Ads targeting in the same manner they would target any other customer segment or group.   Image 25: Google Ads Connector Within SAS Customer Intelligence 360   Building upon this vision, Activity Maps for customer journey orchestration across multiple touchpoints is also "in-scope" as a value proposition on leveraging use case-driven recipes and activating attractive customer audiences.   Image 26: Leveraging Audiences For Multi-touchpoint Targeting Strategies In Customer Intelligence 360   In conclusion, the important takeaways from this article include:   The introduction of Machine Learning Recipes in SAS. A detailed walkthrough of using/adapting a segmentation recipe across SAS Viya and Customer Intelligence 360. Other recipes exist across the areas of acquisition, upsell, retention, next-best-action (or experience), recommendations, lifetime value, pricing personalization, attribution and more. If you are interested in leveraging any of these proposed recipes, please reach out to your SAS support team and the sharing can begin!   Remember, there is more to activation than just scoring your model!    Image 27: Destinations & Analytical-Driven Journey Orchestration In Customer Intelligence 360   Our vision at SAS is to serve as the market leader in advanced audience creation & targeting, independent of channel, for enterprise customers leveraging complex, disparate data sources and wishing to consistently deliver superior understanding into their customer journeys. In other words, we want to empower brands to practice responsible marketing.    Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here. For those who want to dive deeper into the current state of the customer analytics technologies ecosystem, check out fresh (and unbiased) research here.                                 ... View more

When brands select to invest precious budgetary dollars to perform a customer campaign, it’s not always desirable to target an entire marketable universe. Whether budgetary constraints stand in the way, or communicating with irrelevant customer segments wastes money, there are rationale reasons to consider efficiency. This means brands may choose to focus on the top 10/20/30% of customers with the highest likelihood to convert for a campaign effort. This is typically referred to as response modeling.   The main goal of this approach is to develop an analytical solution that selects the customers who are most likely to respond to a marketing offer. This action can be a purchase, upgrade, registration, donation, subscription or offer redemption.  Several modeling techniques, typically within the category of supervised learning, can be applied such as logistic regression, decision trees, neural networks, etc. in order to provide us with insight. The key in response modeling is that the algorithm an analyst selects will predict which customers have the highest chance to take a desired action based on historical campaign data. This labeled data contains a set of customers with their individual characteristics and their associated response to the historical campaign offer. However, is this really a brand's goal?   Marketing teams aim to contact the customers most likely to respond. In addition, they desire to contact customers who respond to offers because of the campaign. This leads us to an approach which goes under many names, such as net lift/uplift modeling, differential response analysis, etc.. The net lift modeling approach assumes that there are customers:   Who will always take an action. Who will never take an action regardless of the campaign. Who react negatively to a campaign regardless of their original feelings.   Image 1: Net Lift Modeling Value Propositions   To illustrate this hypothesis, let's provide an example. A brand decides to contact a segment of customers to convince them of a subscription upgrade. A subset of these customers have the intention of making an upgrade, and thus, these are the customers a brand could save budgetary dollars on by not contacting. There is another subset of customers who will be  annoyed by an outbound marketing intervention, and this could cause an unexpected negative outcome. Lastly, yet another subset of the customer group isn’t currently considering the upgrade, and analytically has a likelihood to convert. This is the group that interests the marketing team. If the marketing intervention can convince them to upgrade, this creates incremental value.    Similar to response modeling, analysts can use multiple modeling techniques to build a predictive net lift solution. The main differences are determined by the type of historic data one uses and the maximization function. Net lift modeling attempts to optimize the difference in response of the customers who have received the campaign intervention with those who didn’t. In this way, a brand still ends up with customers who are most likely to respond, but also with an audience who reacts specifically to the campaign thus avoiding the subset who would convert regardless of the campaign.  This saves on the cost of the outbound campaign effort for customers who are already convinced. Moreover, it avoids the concern that a marketing team scares away customers (churn, attrition, etc.).    When one compares traditional response modeling with net lift analysis, a number of value propositions become apparent. The biggest advantage of net lift modeling is that it gives analysts a true answer to the question if marketing is making a positive impact (or not). Important considerations for performing this type of analysis include:   Is it possible that a subset of customers can react negatively to a brand's marketing interventions? The high cost related to losing an existing customer is a well known problem. What is the macro-cost of a campaign? If a brand takes on a response modeling approach, it will also contact customers who would have reacted positively regardless of the offer intervention. In this situation, a brand is pushing budgetary dollars into the campaign without gaining a return on the investment. Before proceeding, check out this introductory video extracted from a recent on-demand webinar summarizing the net lift & incrementality approach as a viable business use case.     Now, let's proceed to a more detailed example. The objective is to target a product offer to customers on a brand's website or mobile app. Impressions of the targeted intervention would occur on the brand's product-specific web pages or app screens. Whether the customer selects to immediately convert or not, the brand will monitor the customer's behavior related to this intervention for 90 days. The test design results of this campaign would look like this:   Targeted test group: Received an offer. Non-targeted (or holdout) control group: Did not receive an offer.   Let's assume the overall customer 90-day conversion rate for this test resulted in 1.5%.   Customer Cell Conversion Rate Test Group 5.01% Control Group 5.00%   So, why did we not see any campaign lift?   Customer Segment Outcome Not Interested Will never purchase the product. No point in marketing to them. Very Interested Likely to purchase the product on their own. Marketing could even have an adverse effect. The campaign targeted too many of these clients. Influenced Customers Interested in the product but need to be motivated to buy it. Target more of these clients.   A solution to the problem observed above would be net lift modeling (as opposed to response modeling).   Net Score Audiences Test Control Net Lift Top 20% 6.10% 3.90% 2.20% Bottom 80% 4.75% 5.28% -0.48%   To close out this example, let's summarize the differences between net lift vs. response modeling in the context of this use case.   Net Conversion Rate = Test Group Conversion Rate - Control Group Conversion Rate   Propensity Modeling   Most common approach. Targets the clients with the highest probability of making a purchase following a marketing contact. Maximizes the test group conversion rate.   Net Lift Modeling   Targets the customers that can be motivated by marketing. Maximizes the incremental conversion rate and profitability.   Image 2: Marketing Intervention Strategy Centered On Incrementality     Net lift models predict which customer segments are likely to make a purchase ONLY if prompted by a marketing undertaking, as well as maximizing return on campaign investment. With that said, let’s pivot to an industry-specific demonstration on how SAS can be leveraged as an end-to-end solution for campaign management, return maximization and incrementality in the context of net lift-driven marketing.   Casino Gaming & Hospitality Demonstration: Optimizing Player Reinvestment with Machine Learning   The use case we walk through here shows how to decide which casino player in our database receives the most profitable offer within a campaign. One industry-specific term used in this demonstration is theoretical spend or “theo,” the amount of money a player is expected to spend on the casino floor. It is the theoretical revenue and profit numbers which drive many casino marketing decisions.   The first step in the net lift modeling process reviews the results of the last 90-day marketing campaign. The two main targets of this analysis are player visits (our conversion event) and theoretical spend (our revenue metric). The aim is to maximize the combination of these two metrics to determine which is the recommended offer to send to each individual player.   The demo video below demonstrates the exploratory visual analysis and assessment of this net lift modeling example for casino gaming.       Now that we have explored and summarized our visual modeling outcomes, let’s next demonstrate the model pipelining process to show how analysts and data scientists have the optionality of deeper control and transparency in authoring machine learning assets. The focus of this demonstration is on the first goal of the net lift process: predicting a player’s likelihood to return within the next year. The second goal of predicting a player’s estimated spend would be accomplished using a similar workflow.     As mentioned in the demonstration, machine learning models must be transparent and fair to be implemented into business processes. To showcase one plot provided for the demographic variables we selected to assess for bias in the model project, let’s review the performance bias chart auto-generated by SAS of a player’s gender. In Figure 1, SAS compares how accurate the champion model is for each gender group – thus we can determine if it is treating one group of people differently than another. Here, we see limited differences in bar size across our groups – thus the model performs at a similar accuracy rate across all levels of gender in our data.    Image 3: SAS for Modeling Performance Bias Detection & Interpretation   Now that we have confirmed our model is fair, we need to ensure it is transparent for those decision makers dependent on the results. One way to do such is to look at an individual player to understand the key influences of his/her predicted outcome. In the HyperSHAP plot shown below, when looking at this individual, the most significant variables which positively contributed to their high likelihood of return are the number of total trips, days since last trip, and loyalty tier code, while the one which most negatively impact return probability is the player tenure.   Image 4: Champion Model HyperSHAP Plot   We now have a machine learning model in which we are confident in. To implement it into a customer decisioning process, we want to integrate all available domain knowledge to ensure our model serves our bespoke business use case. Thus, we need to integrate the modeling output with a brand's business rules (or best practices) to generate a final treatment decision for our players: who is eligible for a specific marketing offer?   Realize the Marketing Vision With SAS Customer Intelligence 360: Activating Prescriptive Analytics To Improve Profit   SAS Customer Intelligence 360 enables brands to use first-party data to make better customer decisions using predictive analytics and machine learning in conjunction with business rules across a hub of channel touch points. As your brand's journey into analytical marketing use cases progresses, usage of modeling intellectual property cannot be under-exploited. It’s competitive differentiation awaiting to be deployed.    All of this culminates into a brand's guiding light framed around scalable customer decisioning. Rooted in a variety of analytical approaches that can be leveraged within a wide set of marketing use cases,  it doesn't matter if one or multiple technology solutions serve as the bridge to the finish line. There is innovation being served from the software industry to appreciate, explore and experiment with. We (at SAS) simply want to help our customers through partnership and adoption, and whether it involves augmenting a 3rd party martech application to derive incremental value, or using SAS standalone to resolve challenges in orchestrating profitable customer experiences, one theme is clear.   2024 is showing a strong propensity for how marketing divisions will fall in love (again) with analytics, machine learning and AI for an array of new customer use cases. Our last demonstration video below will summarize how SAS analytically-driven martech capabilities broaden/enhance the value propositions of our casino gaming and hospitality use case.     Activating Prescriptive Analytics To Improve Profit With 3rd Party Applications   Decisioning use cases can either be executed on a schedule or in real-time, depending on the business need. For informing marketing campaigns of eligible players (or customers) with specific offers, a scheduled job may suffice to deploy those insights into a brand's 3rd party marketing hub when needed. Real-time execution receives live interactions or data points, processes them, and outputs the decision at the time and in the application necessary for the targeted consumer. Let’s go through one example summarizing when a brand would like to deploy the customer decisioning treatments into a 3rd party web app. This last demo will also provide an opportunity to go deeper and learn more about SAS Intelligent Decisioning.     Our vision at SAS is to serve as the market leader in advanced audience creation & targeting, independent of channel, for enterprise customers leveraging complex, disparate data sources and wishing to consistently deliver superior understanding into their customer journeys. In other words, we want to empower brands to practice responsible marketing.    Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here. For those who want to dive deeper into the net lift marketing use case highlighted in this article, check out a recently published webinar available on-demand here.   ... View more

If one doesn’t take a closer look into how digital analytic technologies differentiate from one another, it is easy to assume they do similar things. For example, both Google Analytics and SAS Customer Intelligence 360 can create 1st party digital data assets for a brand to utilize. But is that it? No, and if one investigates this topic, it is evident that brands have incremental opportunities to generate business value and insight from their digital properties.    It wasn’t long ago where brands operated across the customer lifecycle of acquisition, cross-sell, and retention using assumptions based on years of practitioner experience. Then up and coming analytic talent prior to the data science explosion began challenging those conventions with hard quantitative data. The shift to today is the monetary expenditures associated with capturing and storing 1st party customer behavioral data has decreased significantly opening a new realm of innovative possibilities.   Image 1: Digital Analytics   Let’s start by sharing a generalized definition for "digital analytics" platforms:   In short, they represent specialized analytic applications used to understand and improve digital channel user experience. Their focus ranges across customer acquisition and behavior, as well as optimizing marketing campaigns. The emphasis is on digital channels and techniques yet use cases can evolve to connect with offline data. They are end-to-end platforms performing functions from data collection through analysis and visualization.   After approximately 15 years, Google officially modernized their well known digital analytic software offering from Google Universal Analytics (UA) to Google Analytics 4 (GA4). This occurred in July of 2023, the previous UA version no longer functions and brands had to perform a new implementation if they desired to leverage GA4 going forward. Given how many years brands depended on the UA version of the software, numerous readers will agree with various areas of concern associated with this change:   Risk of historical data loss Lack of feature parity between GA4 and UA Learning to use a new product Resources required for re-implementation   Building upon the importance of this historic event, Adobe released their annual holiday digital shopping forecast. The Q4 - 2023 edition of this report summarizes the analysis of 1 trillion visits to retail sites and over 100 million SKUs in the United States. For context, that's one industry within a singular geography. Two key findings to be aware of:   Despite a challenging macro-economic climate, consumers will drive up positive spend growth and generate over $221.8 billion (4.8% YoY) for online retail this season. ~5% YoY growth may seem small, but keep in mind this is incremental growth after the Covid pandemic when everyone was at home and utilizing digital devices at a much higher rate. Suddenly the ~5% metric is impressive. Mobile device will overtake desktop this holiday season, accounting for 51.2%of online spend for the holiday season and amounting to a record$113 billion.   Now, extrapolate on market opportunities outside of the retail industry driven by progressive growth. The importance of digital customer experience and the analysis that supports B2C interactions is massive. Hopefully we have your attention.    Image 2: Online vs. Offline   Think about why people visit a website or open an app. A few examples could be:   Existing customers looking for help Segments of people with unique behaviors and interests Not everyone comes to you to buy something…maybe they are interested in an organization’s philanthropic efforts   Digital customer analytics helps brands understand why people come to your website or app. What do they do? Do they leave happy? Did the brand make money during these interactions? The main purpose of digital customer analytics in any industry is to help brands understand their progress toward particular business objectives. Considerations include:   Deciding which objectives to focus on Identifying the metrics that drive the chosen objective at each stage of the customer journey Defining a target metric to work towards   Now, for readers who are not familiar with Google Analytics 4, it is a digital analytics service that tracks and reports website and mobile app traffic, currently as a solution inside the Google Marketing Platform.    Image 3: Google Analytics Primer   The most common use cases for Google Analytics 4 revolve around websites and mobile apps. GA4 collects and stores user interactions with your website or app as events (page views, button clicks, user actions, etc.). Before we begin contrasting GA4 with SAS Customer Intelligence 360, it is relevant to note the evolving state of digital customer analytics associated with the frequently used term "digital intelligence". This category is represented by specific sub-categories making up a logical hierarchy:   Data management technology Tag management systems Customer data platforms (CDPs) Data management platforms (DMPs) Digital analytics technology Location intelligence Application analytics & performance management Web analytics Interaction (CX) analytics Product intelligence Predictive modeling Social analytics Internet-of-Things (IoT) Experience optimization technology Behavioral targeting Recommendation systems Testing and experiments Digital touchpoints Web Mobile app Email Paid media Social, etc. Customer and business context Personalization Relevance Conforms to a brand's unique business model and customer measurement objectives   To learn more about digital intelligence as a theme, it is recommended to review the latest research on the subject available here. Moving on, rapid digital transformation trends have resulted in a hybrid consumer engagement model that is changing how brands shape, manage and deliver customer experiences (CX). Throw in the forthcoming loss of third-party cookie tracking and marketers must double down on the information they do have: owned first-party data.   Image 4: The Importance Of 1st Party Digital Customer Data   Brands can address B2C challenges by deriving real intelligence from first-party data as opposed to misleading insights from inferred sources (third-party data) while acting with the speed and agility necessary to meet consumer expectations. With that said, let’s explore the commonalities between GA4 and SAS before addressing how they are different.   Google Analytics 4 & SAS Customer Intelligence 360: Commonalities   Both Google and SAS create 1st party digital customer data assets for your brand to leverage by offering JavaScript tracking code & SDKs as the technologies that natively collect data from websites and apps while providing brands the capability to address which user interactions to focus on, as well as what to defocus. Each time the tracking mechanisms are triggered by a consumer’s behavior, such as a page load on a website or a screen view in a mobile app, GA4 and SAS record that activity, and then package the data. Once created, the data is transmitted either to the Google Analytics collection servers within the Google Cloud Platform (GCP) or SAS collection servers on Amazon Web Services (AWS) tenants. Examples of the type of data GA4 and SAS capture from a digital property include dimensions and metrics associated with the Page URL, Screen Views, Browser Info, Language Type, Device, Operating System, Traffic Sources and more.   Image 5: Digital Customer Data Collection   In the context of sites, to add the JavaScript tag to a website's code, users have a few options:   Use a tag management solution Manually add the tag to a website's code Provide the tag to a website builder service   Heading into 2024, most brands desire a user-friendly solution for tag management. One example would be Google Tag Manager. It allows organizations to efficiently add and update all of their website tags in a web interface to better understand conversions, site analytics, and more. This makes tag updates much simpler for anyone who’s not familiar with coding in JavaScript. Most importantly, it works with both Google and non-Google tags, including SAS Customer Intelligence 360.   Alternatively, users can manually add either the Google or SAS data collection tag to a website by copying and pasting it in the code of every page immediately after the <head> element. No matter how a brand approaches website set up, once added, this establishes the connection with the digital property and data collection begins!   Before we move on, let's address mobile app data collection. Both Google and SAS use software development kits (or SDKs) to add support for event collection and managed content to native mobile applications. Marketers, business analysts, and designers can use collected events to understand how the mobile app is performing, customer insights and targeting users for content distribution. Whether using Google's Firebase SDK or the SAS Customer Intelligence 360 SDK, if your brand’s app is on both iOS and Android, users need to create a customer data stream for each platform. App developers must implement the SDK within the brand's app before analysts can measure or model customer/prospect activity.   Once configured for an app, a number of customer behavioral events like opens, in-app purchases & screen views will become available for analysis. In addition, users have the option to collect custom events that are relevant to their company's business. While the method of data collection differs slightly between web and app, analysts can research all this data together in separate or consolidated views to understand how customers navigate across different touchpoints and properties.   As the tracking executes, both GA4 and SAS carry one more similarity which is creating anonymous unique identifiers to distinguish between new and returning digital visitors. There are different ways an identifier can be created, and industry best practices continue to utilize persistent 1st-party cookies.   Image 6: Identifying New & Returning Users   Google Analytics 4 & SAS Customer Intelligence 360: Differences   Now that we have outlined the similarities, this is the moment when the topic of data contextualization emerges to showcase the differences between digital customer analytic technologies. After digital behavioral data is collected, both Google and SAS transform the raw information using the settings (or configurations) controlled from their respective administrative user accounts.  During data contextualization, the technologies take steps to transform the semi-structured raw event and interaction data from collection using these rules-based settings and configurations. These settings underscore the importance of a solution’s implementation, as they help align your 1st party digital data more closely to your brand’s measurement plan and business objectives.   Image 7: Configuration & Processing   It should be noted how data is transformed has direct implications on the type of analysis, as well as the sophistication of the use cases, that can be performed by an analyst team. In the case of Google, after it has finished processing what it has collected, let's dive in and provide visibility into how this works.    With each customer interaction to a brand’s website or app, an event and associated parameters containing information about the interaction are sent to GA4. An event could be when a user first opens an app, watches a video, or views a page on a website. Events are sent with additional data called event parameters for interpretative context. For example, when someone watches a video on a website, an event is fired when the user clicks play. Event parameters, like the name of the video and how long the video was watched, add context. Along with events and event parameters, user data, like geographic location and the device being used, are also sent. These are called user properties.   To visually represent this data in reports, GA4 uses dimensions and metrics. A dimension is an attribute of your data. It describes your data, and it's usually text as opposed to numbers. A metric is a quantitative measurement, such as an average, a ratio, and a percentage. It's always a number as opposed to text. All GA4 reports are based on different combinations of “dimensions” and “metrics”. When viewing data in the analysis interface, users can think of this environment as a layer on top of the data that allows analysts to organize, segment, and filter. When opening a GA4 report, a query is sent to the aggregate tables stored in the Google Cloud Platform (GCP) that are populated with prepared data to enable the analyst with descriptive insights. Image 8: GA4 Descriptive Reporting   One of the strengths of GA4 is represented by prebuilt descriptive analytic & measurement views available in the "reports" section of the user interface (which can be viewed in Image 8 above on the left-side menu of the screenshot). Let's dive into the GA4 Audiences reporting view to see an example of this. The intention of this pre-made report is to identify a brand’s engaged and profitable audiences.   Image 9: GA4 User Audience Report   Google defines the term "Audiences" as a group of users from a site and/or app who have generated similar behavioral data or who share demographic or other descriptive data (e.g., same age group, same gender, were acquired by the same campaign).  A notable benefit to users of GA4 is the ability to share audiences with Google Ads, so marketing across touchpoints like search or display to specific groups of users is feasible.   For example, analysts might want to create an audience who have made a purchase of any kind (purchase event_count > 0). However, analysts might find this audience too large for the practical purposes of their brand's ad campaigns. In this case, analysts can identify smaller groups of users who have more specific behavior and demographics in common, like:   Users from California who have purchased 1-5 items Users from San Francisco, California who have purchased 1-5 items in the last 7 days Users from San Francisco who purchased 1-5 items in the last 7 days and who spent more than $100   GA4 administrators can work with their analyst teammates and define audiences broadly or narrowly. When a unique audience needs to be defined and made available in GA4 reporting views, this is performed in the admin section of the software.   Image 10: GA4 Admin Screen & Defining Audiences   Another feature to point out here is the concept of predictive audiences in GA4. A predictive audience is an audience with at least one condition based on a predictive metric. For example, analysts and/or admins could build an audience for ‘likely 7-day purchasers’ that includes prospects who are likely to make a purchase in the next 7 days. The availability of predictive audiences depends on the underlying predictive metrics being eligible for use by meeting specific prerequisites. As of January 2024, GA4 offers three predictive metrics for usage:   Purchase probability: The probability that a user who was active in the last 28 days will log a specific conversion event within the next 7 days. Churn probability: The probability that a user who was active on your app or site within the last 7 days will not be active within the next 7 days. Predicted revenue: The revenue expected from all purchase conversions within the next 28 days from a user who was active in the last 28 days. Image 11: Setting Up Predictive Audiences In GA4   Google describes that GA4 automatically enriches a brand’s data by bringing machine-learning to bear on the dataset to predict the future behavior of customers. This concept of do-it-for-me (DIFM) functionality is an evolving trend in the broader martech ecosystem of solutions, and is designed to simplify the usage of advanced analytics. Keep in mind, there are both advantages and disadvantages to friendly features like this that will be discussed in more detail shortly.   Once defined, predictive audiences are available for analysts to view or edit within the admin screen.    Image 12: GA4 Predictive Audience Example   Image 9 above shows predictive audiences displayed in the same Audience reporting views as standard audiences. However, pre-configured reports can sometimes be frustrating for analysts who want more customization within their research. Building upon this premise, GA4 offers analysts the ability to build custom explorations.  When analysts want to explore data in more detail, they can:   Perform ad hoc queries Configure and switch between techniques Sort, refactor, and drill down into the data Use filters and segments Create segments and audiences Export the exploration data for use in other tools   The screenshot below was taken in January 2024. At the time, these were the exploration techniques supported:   Image 13: GA4 Exploration Techniques   The ingredients of an exploration in GA4 are made up of three parts.   Canvas: The large area on the right displays data using the selected technique.  Variables: The panel on the left gives analysts access to the dimensions, metrics, segments & timeframe to use in the exploration. Tab Settings: Analysts can use the options in the Tab Settings panel to configure the exploration.    The types of exploratory visualization and graphing objects supported are:   Table Donut chart Line chart Scatterplot Bar chart Geo map Funnel Sankey Diagram Venn Diagram Heatmap   For example, here is a screenshot of the GA4 funnel exploration.     Image 14: Funnel Exploration Technique In GA4   Analysts can visualize the steps visitors take to complete a task and see how well they are succeeding or failing at each step. Funnel exploration enables analysts to address questions like:   How do prospects become shoppers and then become buyers? How do one-time buyers become repeat buyers?   Moving on in our quick GA4 tour, the last section to cover relates to Advertising analysis. The reporting views in this section aim to help analysts better understand the ROI of media spend across channels, make informed decisions about budget allocation, and evaluate attribution models.    For readers who are not familiar, attribution is the act of assigning credit for conversions to different ads, clicks, and factors along a consumer's path to completing a conversion. An attribution model can be a business rule, a set of rules, or a data-driven algorithm that determines how credit for conversions is assigned to touchpoints on conversion paths. These GA4 reports allow analysts to explore different attribution measurement approaches and determine which one might work best for their business.   Image 15: GA4 Advertising Report For Conversion Paths   As of January 2024, there are currently four reporting views in the GA4 Advertising section:   Advertising snapshot: Overview of business metrics, while allowing analysts to dig deeper into the areas they want to explore. Performance: Observe which channels and campaigns received conversion credit.  Model comparison: Contrast how different attribution models impact the valuation of marketing channels. Conversion paths: See customer paths to conversion, and review how different attribution models (last touch vs. data-driven) distribute credit on those paths.   Analysts can use these reports to answer questions like:   What roles did referrals, searches, and ads play in conversions? How much time passed between a customer's initial interest and their purchase? What are the most common paths customers take leading up to conversions?   Now that we have provided a detailed introduction of GA4, let's pivot to SAS. After digital consumer interactions are processed and contextualized by SAS Customer Intelligence 360, the first key difference to highlight relates to the solution's data model. While GA4 enables:   Pre-made & templated descriptive reporting views Data exploration, segmentation & filtering Advertising analysis Audiences & Google Ads integration Evolving set of automated insight features   In contrast, SAS provides an actionable data model that serves both overlapping and uniquely different requirements.   Image 16: SAS Customer Intelligence 360 - Data Model   A number of value propositions SAS is bringing forth include:   Analyst access to structured data tables at varying levels of aggregation and detail Integration with on-prem or cloud-based CRM Pre-made, templated & custom reporting Do-It-For-Me (DIFM) automated insights Do-It-Yourself (DIY) data visualization and visual no-code modeling DIFM or DIY predictive modeling, machine learning & model interpretability Pre-made ABTs (analytic base tables) for data science acceleration Support for data-at-rest and data-in-motion use cases Triggered real-time decisioning activated by customer behavioral event detection Campaign management, personalization & targeting for owned digital properties, ad media platforms & multi-touchpoint journeys   As the proportion of online to offline customer interactions with a brand continues to elevate, both digital experiences and the 1st/Zero-party data that analytic solutions collect about them will mature in ways that will challenge and benefit insight-led companies. Over the years, SAS has observed the web/digital analytic ecosystem of software solutions and targeted users bury themselves in templated measurement reports primarily descriptive in nature. As digital data related to a brand's customers continues to incrementally increase in importance, enterprise use cases for sharing insights broadly within their walls frequently attempt to absorb digital data signals into their business intelligence (BI) or data visualization software applications. However, many brands seem to be hitting an adoption ceiling.   At SAS, our perspective on this trend is the lack of customization (or personalization) derails analysts on receiving the benefits of a digital measurement solution's attempt to be easy-to-use. Augmented data visualization and discovery is appealing because it does not subscribe to a one-size-fits-all user experience. To fully extract the potential of analytic measurement, interactive visualization, predictive modeling, algorithmic segmentation and machine learning, numerous viewpoints should be considered:   Insights - Who will explore, identify & produce? How will they be explained? Most importantly, don't overlook how they will be consumed. Every brand will have unique profiles (or segments) of team members who will receive insights, and their personal levels of data/analytical literacy matters.  Analysis - Web/digital analytic standalone solutions tend to skew more towards canned/guided delivery of insight. There's nothing wrong with that if your brand finds this beneficial. But when brands link digital and non-digital data sources together, the flexibility to explore, share recipes of approach & using technology that augments/accelerates the analysis workflow in arriving to unbiased, impactful insight conclusions is critical.   SAS recognizes the critical importance of serving multiple enterprise personas through augmentation (embedded AI and machine learning to assist as analytical helpers.) This spectrum ranges from business users who want out-of-the-box benefits to savvy analysts who want to build assets from scratch. It is extremely challenging for any brand or supporting vendor to predict if a do-it-yourself (DIY) approach vs. a do-it-for-me (DIFM) approach will be more effective. SAS constantly observes, accepts and uses this challenge to inspire our software’s design principles to enable capabilities to reflect the balancing needs between marketers, analysts and data scientists within an organization.   Image 17: SAS Technology User Personas   Furthermore, what can Marketing AI do to keep improving insight-driven practices? It's natural within brands to see different flavors of analysts and martech team members express a desire to go beyond reporting, querying, data visualizations, and descriptive/diagnostic analytics. Augmentation introduces emerging capabilities worth noting:   Enabling users in no/low or high-code environments to identify key drivers or influencers on a metric, find anomalies, or project a forecast/trend supported with natural language generated (NLG) explanations to ensure accurate analyst interpretation of insights. What about data scientists and data engineers? It should not be overlooked that SAS Viya allows users to author and train new ML models in the same platform where models are going to be deployed, but it also eliminates the need to integrate multiple platforms, improving model transparency/governance, and reducing the likelihood of errors that may result from platform-to-platform data and metadata handoffs.    Image 18: Digital Business Solutions   As we mentioned earlier in the article, SAS can support website and mobile app 1st/Zero-party data collection, pixel tracking, mobile SDKs, and server-to-server APIs. Once any of these data streams are absorbed, the SAS Customer Intelligence 360 unified data model (UDM) contextualizes this information into out-of-the-box (OOTB) and customizable structured tables for analysis enabling DIFM martech user features & DIY analyst acceleration.   The subject areas of the data model include session/event-based data for visits, media, pages, products, forms, search, goals, conversions, customers, web/mobile/email/direct marketing and contact/response. SAS provides over 100 prebuilt & customizable reports (including templates, metrics & KPIs) for campaigns, journeys, content, events, ecommerce and more. Users can create recipes & share using native SAS visualization or through integration w/ MS Power BI & open-source (Python, D3) to take advantage of customizable features for color, KPIs, sizing, responsive design, cross-device/streaming analytics, user annotations/alerting, extensibility to iOS/Android mobile apps & SDKs for custom/3rd party app/websites.    But enough chatter...let's bring this to life. The forthcoming sections will provide readers the opportunity to learn how SAS can be used for a variety of customer and marketing use cases. This will provide clarity into how SAS and GA4 overlap and differentiate.   SAS Technology Demo 1: DIFM & DIY Dashboards, Reporting, Measurement & Distribution   To begin, we will briefly introduce standard OOTB reports to users of SAS Customer Intelligence 360. We will then transition to how 1st/Zero-party data captured & contextualized by SAS Customer Intelligence 360 can be made available to SAS Visual Analytics on SAS Viya. This will significantly expand the amount of customization that can be applied to dashboards and reports, as well as diversify the data sources that can be represented in these measurement assets. We will drill into the augmentation features that provide value propositions such as natural language generated insights, explanations, outlier detection, correlated measures, auto-segmentation & DIFM propensity-scoring. We will wrap up this demonstration in how these measurement reporting assets can be shared.     Within the martech industry, there are several factors that contribute to the challenges surrounding brand decision-making. Obviously, customers and markets are more competitive and demanding. When you step back and reflect on this, it's a linear trend upward year-after-year when it comes to consumer expectations. This means, to satisfy that demand, it's well recognized that brands need to respond quicker, but it's often overlooked that accuracy holds an equal weight. Personalization, targeting, segmentation, relevance and other fun martech buzzwords all rely on this.   Data continues to flood every organization, both in size and in speed. Sometimes more data is better, but the challenge can be that critical decision-making information gets lost. Skilled analytical talent with application experience in the various domains of modern marketing is the key to move a brand from reactive to proactive. Thus, varying flavors of technology and automation are critically important to augment customer analysts in accelerating their delivery's time-to-value.   Machine learning is a branch of AI that automates the building of systems that learn iteratively from data, identify patterns, and predict future results. And it does that with minimal human intervention. Machine learning shares many approaches with other related fields, but it focuses on predictive accuracy. Building representative machine learning models that generalize well on future data requires careful consideration of both the data at hand and assumptions about the various available training algorithms.   SAS Technology Demo 2: Acceleration of DIY Customer Propensity Analysis   How can we improve our conversion rate going forward? This is the question of the decade for analysts and data scientists, and I do not view your leadership team changing their interest in this topic any time soon. You can report, slice, dice, and segment away in your analytics platform, but needles in haystacks are not easily discovered unless we adapt. I know change can be difficult, but allow me to make the case for AutoML and hyperparameter tuning within the discipline of customer propensity analysis. A trendy subject for some, a scary subject for others, but my intent is to lend a practitioner's viewpoint.   Automated machine learning (commonly referred to as AutoML) involves automating the tasks that are required for building a predictive model based on machine learning algorithms. These tasks include data cleansing, feature engineering, variable importance, model selection and hyperparameter tuning, which can be tedious to perform manually. Platforms that provide this capability offer many benefits, such as empowering analysts by giving them a start at a machine learning workflow, as well as allowing data scientists to spend less time on model design and more time on making an AI-enhanced marketing campaign a reality.   Automation is not intended to replace the role of data scientists; ideally, there should be support for intervention in these systems to allow the performance of tasks such as domain-specific feature engineering, which can be a critical component of improving the performance of predictive modeling. These systems should be transparent with regard to the algorithms being used, so that users can be aware of, understand, and trust the insights being generated.   SAS provides different levels of automation that can be included in the machine learning pipeline-building process. Users can do any combination of automated tasks, such as having the system determine variable roles and levels, create the best transformation for numeric features, generate new features, and more. Alternatively, the entire process can be automated, through a graphical user interface as well as using a REST API.    With that said, this demo video will address AutoML and hyperparameter autotuning with data captured and contextualized from SAS Customer Intelligence 360.     As an analyst, when you can communicate the value of your modeling efforts in monetary terms, every executive paying attention is going to lean in and focus. Passing these insights to influence our marketing teammates will directly impact their segmentation strategies and touchpoint tactics. As a life-long student of business and marketing analytics for the last two decades, the concept of applying a profit matrix is one of the most industry-practical topics I have ever learned.    Keep in mind, supervised learning algorithms are trained using labeled examples (conversion vs. non-conversion), such as an input where the desired output is known. The learning algorithm receives a set of inputs along with the corresponding correct outputs, and the algorithm learns by comparing its actual output with correct outputs to find errors. It then modifies the model accordingly. Supervised learning is commonly used in applications where historical data predict likely future events. For example, supervised learning can anticipate when an insurance customer is likely to file a claim, or when a retail customer has a higher likelihood to be interested in an upsell recommendation.   SAS supports two types of supervised learning problems through natively-supported algorithms such as gradient boosting, forests, neural networks, support vector machines, Bayesian networks and more.   Classification – When the data are being used to predict a categorical target, supervised learning is called classification. This is the case when assigning a label or indicator (for example, labeling an image a dog or a cat). When there are only two labels, this is called binary classification. When there are more than two categories, the problems are called nominal classification. Regression – When the data are being used to predict interval targets, the problems are referred to as regression.   The reason supervised learning as a category contains a variety of algorithms is based on the notion that no model is uniformly the best, particularly when considering the deployment over time, when data changes. Analysts select a model primarily based on fit statistics and assessment graphics of performance.   Supervised classification does not usually end with an estimate of the posterior probability. For example, in binary classification problems, the ultimate use of a predictive model is to allocate cases (customers) to classes (target / don't target). This is accomplished by appropriately choosing a posterior probability cutoff. The cutoff or threshold represents the probability that the prediction is true.   SAS Technology Demo 3: Supervised Learning, Prediction & Maximizing Profit for Customer Targeting   Determining an appropriate cutoff is problem specific, and there are many ways of accomplishing this (Bayes' Rule, Central Cutoff, KS Cutoff, etc.). We will focus on one solution referred to as the Profit Matrix, which is a formal approach to determining the optimal cutoff using statistical decision theory. The decision-theoretic approach starts by assigning profit margins to true positives and loss margins to false positives. The optimal decision rule maximizes the total expected profit.   The next demo video will address the usage of profit matrices using digital interaction data and other 1st party customer assets together to benefit marketing-centric use cases.     As marketers and advertisers utilize AI and ML to elevate their brands, understanding the potential for detecting and mitigating bias within predictive analytic procedures is crucial.  Today, AI plays a progressive role in advancing the marketing and advertising mandate. Data is generated constantly in the digital ecosystem, associated with real-time customer engagement behavior, web/mobile interactions and incremental streams of desired revenue year-after-year. This gives brands the chance to apply a variety of analytical techniques to draw "ah-ha" insights that can be used in various forms of customer interaction tactics.   At a fundamental level, AI/ML helps brands improve their viewpoints on customers and subsequently, running their business. When using data science for these intentions, brands are placing immense trust that their modeling IP is providing useful insights that subsequently influence marketing treatments. Keep in mind, not every single human being on Planet Earth has the same level of data/analytical literacy, and some may consider applications of AI/ML as opportunistic innovation, free from the mistakes commonly attributed in human-driven decisions.   I hate to spoil the fun, but we need to be more prudent as practitioners. AI/ML can contain bias. Instead of ushering in a utopian era of fair decisions, AI/ML have the potential to exacerbate the impact of biases. As innovations help with everything from the identification of attractive prospects to predicting who should receive a marketing stimuli, it is important to understand every modeling application has the potential to affect separate segments of a customer population differently. When applied in martech, biased AI/ML can negate efforts to learn, understand and anticipate consumer behavior. Brands should improve their understanding of how AI bias impacts them, how to detect it and ultimately, mitigate.   SAS Technology Demo 4: AI/ML Bias Detection and Mitigation for Responsible Marketing   One of the top predictions for 2024 centers on responsible marketing. There are numerous ways in which bias can slip into customer data. Although marketers themselves may not build analytical models, it's hard to find a use case these days that doesn't benefit from propensities or probabilities. There are many perspectives to take into account when describing bias in data science. Bias can happen during data collection, data processing, sampling, model building, and so on. However, when AI/ML is applied to data that is inaccurate, it has the potential to magnify the errors and cause unintended bias in campaigns, personalization or testing. If that wasn't enough, bias can lead to irrelevant results from severely impacted KPIs, such as failing to reach the correct audience or serving up the wrong offers to a particular demographic. Ultimately, this means wasted money and resources, failure to reach relevant customers, and potential harming the reputation of a brand.   While most brands readily promote fairness in AI/ML as a principle, putting the processes in place to execute it consistently is an ongoing obstacle. There are multiple dimensions for evaluating the fairness of AI/ML, and determining the correct approach depends on the use case. In short, we (at SAS) see two parallels at the moment:   Data scientists and analysts need to continue their focus on translating the output of AI/ML in business language and storytelling to reduce stakeholder intimidation. Remember, if your models are not put into action, what was the point of your effort?  Marketing and CX decision makers may not be passionate about statistics, but nearly every use case can elevate through the usage of propensities and probabilities. While AI/ML is advertised as next-level precision, it is not 100% perfect. Therefore, the translation of propensities and probabilities into business context must be interrogated, transparent and understood.   Effective risk management is increasingly being brought to the frontline rather than functioning in the back office. When using advanced analytics, it’s becoming increasingly important to understand and measure fairness risk to avoid exploiting vulnerable customers. The next demo will feature a combination of online and offline customer data as inputs to the use case.     As 2024 takes hold, it is anticipated use cases for customer analytics and digital intelligence continue to evolve with a bottomless hunger for variety and volume of data to supercharge use cases. Check in with any of the analytical magicians within your brand, and observe their struggle to acquire the relevant input data (and enough of it) to train modeling recipes.   Sounds strange, right? The frequently used term of yester-year known as "BIG data" used to roll off the tongue as often as we hear the term "AI" today. The challenge isn't that there is enough information in general. But the intention of any analyst is to identify meaningful data signals to address business objectives. What's the point of having access to oceans of big data if it's just noise? Garbage in, garbage out. For example, I would like to build a classification model using supervised learning to improve our understanding of behaviors and drivers of conversion behavior for my B2C (business-to-consumer) brand.   - 99.5% of the customer journeys resulted in non-conversion for the past 90 days. - 0.5% resulted in a conversion.   If my analysis does a great job of accurately predicting non-conversions (high true negative rate), but does a terrible job of classifying conversions (low true positive rate), what marketing leader is going to get excited about that? Excluding special exceptions, not many. The problem here amplifies a well known problem in analytical modeling related to a rare-event of interest.   Taking it one step further, as generative AI and deep learning increase in practical application of business use cases, many observe the space as a foundational component of modern AI algorithms. But we also know to effectively train models using these algorithms, one needs a tremendous amount of data. And while it seems like we’re practically swimming in data day in and day out, we don’t necessarily always have enough of the RIGHT data for every process or behavior we’re trying to model. In other words, it's NOT the lack of choice of machine learning algorithms, but the scarcity of high-quality data.    Enter synthetic data (one of numerous generative AI approaches) which replicates, mirrors, or extracts look-alike information that allow analysts to model use cases that would otherwise be impractical. A few examples of challenges include:   - Data quality concerns - Privacy - Lack of relevant data   But I work in the martech industry, how does this apply to me?    Real data is expensive to collect and properly annotate, especially when it is in large scale. This is both a monetary and time drain for high-utility team members who support customer journey management processes. Real data can also be messy, requiring time to clean and/or extract useful features. It can be imbalanced, which makes it harder to train good models in support of journey-based analytics. It can be sensitive to share or store due to privacy concerns.     Now, marketers frequently run campaigns, tasks and activities. They target audience segments. The desire is to deliver personalization that is helpful, relevant and value enhancing within tailored customer experiences across channels. Two examples of analytically-driven marketing to consider:   Propensity scores. They are intended to identify high-likelihood audiences who will convert on your macro- and micro-conversion goals. However, imbalanced data that is used to train classification models will produce higher margins of error, or less valuable propensity scores that lead to irrelevant personalization and lower conversion rates. Look-alike audience insights. You know you love it when an analyst describes to you the behaviors, demographics, and transactional patterns of high-value customers. The actionable outcome is to hunt for look-alikes within acquisition & upsell/cross-sell marketing. Imbalanced data is like a viral infection, and reduces the opportunistic potential of the insight-driven strategies being leveraged to influence the usage of marketing budgets.   SAS Technology Demo 5: Synthetic Data Generation (Generative AI) for Martech   Synthetic data can be artificially manufactured by special-purpose machine learning models in a way that captures the data distributions and patterns, while also helping to maintain privacy without exposing real information.  For example, a Generative Adversarial Network algorithm, or GAN, can learn the patterns and relationships in existing data in order to generate new observations that are indistinguishable from real data. You’ve probably seen this used for what are known as deep fakes (creating very realistic images of people that don’t even exist). But we can also use this same technology for tabular data, which is most common for training predictive models with machine learning algorithms. The following demo will showcase the use of the SAS native tabularGAN action set to generate synthetic data.      It's readily recognized customers are digitally savvy, discerning and motivated to get the best deal. This has made it increasingly difficult for brands to develop pricing strategies that optimize net revenue.   Customer-centric pricing can be a game-changer. Whether an online or brick-and-mortar shop, pricing personalization aims to use data and analytical insight to influence what is (or isn't) offered to each prospective buyer. Is the price a part of the shopping experience? Absolutely. Price is one of the major factors that forge a consumer's buying decision and loyalty.   The direct interaction between brands and customers (think web, mobile app, email, etc.) enable the opportunity to implement personalized pricing more effectively. Brands collect 1st party data on the consumer's engagement with their  product offerings, and can use this information to develop offer strategies. Over time, our consultative engagements here at SAS with various brands have shown common challenges:   Storage and effective use of high quality, un-sampled 1st party data on user behavior, purchase history, etc. Optimizing net revenue with pricing tactics   Pricing personalization can be a useful tool to help brands reach different objectives, depending on their business model, market conditions, and customer segments. For instance, brands can increase sales volume by targeting price-sensitive customers with lower prices while maintaining higher margins from less price-sensitive consumers. Sounds logical, rationale and intelligent, right?  While pricing is often described as a science by practitioners, it is a key factor to drive conversions. And we want to do this while optimizing profit, which points to a secondary challenge in regard to which customers should receive a marketing intervention (or stimuli), and which ones shouldn't.   Setting the right price for a good or service is an old problem in economic theory. Keep in mind, there are a vast amount of pricing strategies in existence that depend on the objective sought. One brand may seek to maximize profitability on each unit sold or on the overall market share, while another company needs to access a new market or to protect an existing one. Moreover, different scenarios can coexist in the same company for different goods or customer segments. Although strategies like premium and penetration pricing have existed for many years, let's focus on the use of algorithms to address this challenge.   Algorithmic personalized pricing is a process of setting optimal offers using the power of AI/ML to maximize revenue, increase profit or address other business goals set by brands. Algorithmic personalized pricing can easily become one of the most powerful means of gaining a competitive advantage for a brand.   SAS Technology Demo 6: Pricing Personalization, Net Revenue Optimization & Marketing Interventions   One challenge brands will always face relates to the heterogeneous characteristics of their customers. Such variability can affect how customers choose to (or not to) convert from a targeted marketing intervention (such as a price increase or decrease). Understanding customer response behavior to targeted offerings is crucial for informing individualized pricing decisions. PROC DEEPPRICE from SAS offers a flexible framework for specifying and estimating customer responses to marketing treatments.   Let’s look at a demo of how an online media brand can offer targeted discounts through personalized pricing to optimize revenue.     Helping customers of your brand's owned digital properties find items of interest is useful in almost any situation. Now, recommendation analysis leverages customer interaction data to uncover hidden patterns in order to identify related products, services or content to surface for targeting and personalization. This analysis easily extends to other types of use cases to build more customer relevance, especially with recent innovations of new AI/ML techniques.     In SAS Customer Intelligence 360, users can create tasks (Web, Mobile, Email, etc.) that display different creatives based either on a product being viewed or a customer’s behavior. There are two methods for delivering recommendations to users. User-centric recommendations take a user’s behavior into account. Product-centric recommendations are based solely on a product.   A sampling of native recommendation analysis algorithms available in SAS include regularized & non-negative matrix factorization, k-nearest neighbor, bayesian personalized ranking, factorization machines, data translation w/ optimal step-size, slope one, market basket, link analysis, and the list goes on. Additionally, SAS supports usage of open source (Python/R) recommender packages. Given the high volume of algorithms to select from as an analyst, SAS enables champion-challenger recommender modeling prior to deployment.   Let's articulate two scenarios for designing and deploying a recommendation system:   Offline Training & Online Scoring   Offline training is performed on a snapshot of data at rest to produce scoring code. That scoring code can then be deployed into an online scoring system on forthcoming (new) streaming data produced by customers and prospects interacting with your brand. Keep in mind, the scoring on new data is based on a modeling solution that was trained offline (or in batch). Thus, the solution would need to be retrained at signs of performance decay. For a detailed demo video on this scenario (within the context of financial services), please check out this article. For SAS user documentation on developing offline recommenders, go here.   Online Training & Scoring   The demonstration video below (in the context of the retail industry) is based on a true online training and scoring solution for recommendation systems.  The incremental value proposition here is represented by a recommender modeling solution that trained once on a snapshot of data-at-rest in SAS (to mitigate issues like cold start problems), optimize using three algorithmic approaches leveraging factorization machines (FMs), bayesian personalized ranking (BPR) & data translation w/ optimal step-size (DTOS), deploy, and online training and scoring handles the re-training of the champion model and associated scoring on a 1:1 basis. For SAS user documentation on developing online recommenders, go here.   SAS Technology Demo 7: Real-Time Customer Recommendation Systems For Data-In-Motion   Using a SAS for Retail website, let's walkthrough a technology demonstration where SAS Customer Intelligence 360 is leveraged for an AIoT use case in the context of recommenders. We will cover:   The capture, contextualization and input streaming of digital interactions between a prospective customer and the brand's digital property The event monitoring and triggering of a champion recommender model Training and scoring on data-in-motion The output of the recommender scoring will drive immediate actioning, targeting and personalization back into the customer's digital experience     Brands today are complex ecosystems of decisions that must be executed with increasing levels of automation - due to their competitors digitally transforming and influencing customer expectations. In response, there is a need to change how decisions are made.  Organizations have the opportunity to increase their capability to perform augmented decision making - where a human takes analytically driven insight to make a decision (such as within a call center, website or mobile app). Automation within decision making is when an algorithm (or algorithms) blended with business rules make the decisions without human intervention (such as next best offers/actions/experiences). With each passing year, the acceleration of the scale, speed and complexity of customer 1:1 decisions is increasing.   A customer journey in its purest form represents a series of brand-orchestrated connected experiences addressing an individual's desires and needs — whether that be completing a desired task or traversing the end-to-end journey from prospect to customer to loyal advocate. When you reflect on this, the customer experience is the totality of cognitive, affective, sensory, and behavioral consumer responses during all stages of the consumption process including pre-purchase, consumption, and post-purchase stages.   Customer decisioning is best used to drive real-time actions in three contexts.   To drive the ideal next journey-based interaction that a customer or prospect should have with your brand. As part of a cross-channel marketing initiative that unifies an experience across customer-facing channels. To enable personalization that delivers customized messages based on an individual's profile and observed behaviors while respecting experiential privacy.   SAS blends DataOps, ModelOps, DecisionOps & marketing orchestration to support offer treatment prioritization requirements for a variety of journey-based use cases. To enable data-driven decisions at scale, the analytics life cycle must be highly operational, automated and streamlined. By connecting all aspects of the analytics life cycle – brands can turn critical questions into trusted decisions.   SAS Technology Demo 8: Real-Time Customer Offer Treatment Prioritization   The demo video below will feature a fictional financial services company comprised of multiple business units (savings, lending, wealth management, etc.) and operating in numerous geographies. The primary objective will be to showcase a mutual value exchange across the full customer journey by optimizing orchestration capabilities and using customer-directed engagement models to guide personalization strategies.   Secondarily, SAS recognizes brands must adapt to a mix of cross-functional and cross-brand goals. The use of machine learning and prescriptive analytics will be shown in support of how marketing teams can generate and prioritize single and cross-brand journeys. Data monitoring, ML and AI help surface alerts & address needed optimizations that govern which inbound and outbound interactions a customer should receive in a given time period. The intention is for brands to prove the value of marketing in a volatile business environment, connecting strategies to marketing and customer outcomes.     It's no secret who the two biggest players in digital ad media are. Google and Meta (Facebook) quickly come to mind. For those who aren't familiar, Google Ads and Meta (Facebook) Ads are online advertising platforms enabling marketers (and brands) to find customers.    When we consider Google Ads, brands have the opportunity to connect with current and prospective customers across Search, Display, Shopping, Video and App. Pivoting to Meta Ads, marketers can reach new/existing customers as well as their networked communities on Facebook, Instagram, Messenger and WhatsApp.    With this stated, reflect for a moment on the business opportunity to:   Maximize qualified leads and conversions Increase online sales Drive in-store foot traffic (or send more users to your website) Show your brand to more people to increase awareness, reach and engagement Market your app to new users (or increase app installs)   Now consider how many human beings on Planet Earth use Google and Meta (Facebook). It's not thousands or millions. It's billions. Hopefully we have your attention, and let's proceed to connect the dots to SAS. Brands can use connectors to retrieve or transfer data between SAS Customer Intelligence 360 and on-premises or cloud-based applications. Our design intent with these OOTB connectors is to offer brands time-to-value acceleration in activating commonly used data integration flows.   For example, if users of SAS Customer Intelligence 360 want to retrieve GA4 data residing on the Google Cloud Platform (GCP) for segmentation and audience targeting purposes on their owned digital properties, they can use the recently enhanced Audience Management capabilities. If the targeting criteria is met, SAS can then be used for testing, targeting and/or personalization benefitting from the availability of GA4 data. This is an important gap for SAS to fill for our customers because Google sunset (or retired) their Optimize software product in September of 2023.   Image 19: Cloud-based Audience Management in SAS Customer Intelligence 360   The fun doesn't stop with the ability to retrieve external cloud-based data for segmentation and targeting. Users of SAS Customer Intelligence 360 can utilize their brand's 1st-party data assets (including the contextualized digital interaction data SAS makes available) to form and transfer audiences/segments to media platforms like Google & Meta (Facebook) Ads. Given the amount of 3rd party ad targeting data in these media platforms, the opportunity to use 1st party data will benefit ad campaign performance metrics.   SAS Technology Demo 9: Customer Closed Loop Campaign Management On Google & Meta Ads   To bring this to life, provided below is an introductory demo video on how SAS guerrilla-like data science powers can be applied to this use case. It will exemplify how closing the loop on campaign management is much more than just getting clicks. SAS takes the baton from Google and Meta within a customer journey (websites, apps & other brand-owned properties) using analytical muscle (testing, propensity-driven retargeting, algorithmic recommenders, etc.) to improve conversion metrics and monetary-driven objectives.       SAS is frequently requested by our customers/prospects and challenged by the analyst community to showcase how we help marketers design and manage journeys. This tends to involve:   Audience segmentation. Creation, management and planning of the timing/sequencing of a diverse set of channels/touchpoints. Accommodating both new and in-progress campaigns.   Objectively, SAS strives to enable marketers to provide a mutual value exchange on digital channels across the full customer journey by optimizing orchestration capabilities and using customer-directed engagement models to guide personalization strategies. The intent is to provide increasing customer lifetime value (LTV), progressive returns on engagement, more personalized interactions, and sophisticated orchestration across the customer's end-to-end journey.    SAS Technology Demo 10: Customer Journey Building and Design   Let's get to the last demo video below. Here is a quick preview:   SAS will be used to demonstrate customer journey design capabilities, showcasing support for multiple customer-directed engagement strategies. The exemplified journey incorporates a variety of channels/touchpoints, including email, web, mobile, social and external CRM systems. Key features which will be showcased include libraries of trigger-based journeys, node controls, content and offers, testing/experimentation, and journey versioning.       Google Analytics 4 & SAS Customer Intelligence 360: Together Is Better   Although this article contains a tremendous amount of information, we will draw a few conclusions in our research contrasting GA4 and SAS. Let's take this in steps. Every idea, hypothesis and project born within a company begins and ends with a question and decision.   Image 20: Questions & Decisions   Borrowing from software development practices, DataOps provides an agile approach to data access, quality, preparation, and governance. It enables greater reliability, adaptability, speed and collaboration in your efforts to operationalize data and analytic workflows. Both GA4 and SAS acutely understand and enable a tremendous amount of features and capabilities in regard to instrumentation and implementation of data collection within digital properties. However, after these initial steps, the questions orbiting within a brand's business model will drive what is necessary from a data management perspective.   GA4 analysts primarily focus on reporting, measurement and revealing descriptive and diagnostic insights. The topics of data quality, non-normal distributions, missingness and feature engineering tend to be outside the boundaries of typical GA4 analysts use cases. If such data problems are identified, GA4 administrators adjust implementation settings to amend. SAS analysts embrace these data management topics and use technology solutions to improve their ability to identify predictive and prescriptive insights. Solving data problems can equate to a diversity of data engineering solutions, typically handled by the individual themselves as opposed to depending on administrators. Keep in mind, SAS provides a data connector into the Google Cloud Platform for scenarios where brands prefer their analysts to use GA data for reporting/measurement, but leverage SAS data engineering capabilities using this information for use cases that involve customization, sensitive data or advanced analytical approaches not currently supported in GA4.   Image 21: Perspectives On Managing Data Between GA4 and SAS   Conclusion: Although brands historically felt one digital analytics solution sufficed, as we progress through 2024, it is now a frequent observation to see brands investing into multiple solutions to improve their state of data readiness.   Moving on, analysts and data scientists use a combination of techniques to understand the data, visualize and build predictive models. They use statistics, machine learning, deep learning, natural language processing, computer vision, forecasting, optimization and other techniques to answer real-world questions. There are distinctly different flavors of analytical enablement between GA4 and SAS.   GA4's strengths lie in its focus on pre-made reports applied to common marketing questions. Once the solution is implemented, digital data flows into a variety of templated analysis views summarized earlier in this article. With a skew towards marketers and no-code analysts, it's easy to appreciate the simplicity of the user experience. Alternatively, it is challenging to accommodate more sophisticated business questions that require custom-authored solutions. Some will debate that GA4's improvements from Universal Analytics (UA) added a "light" version of a modern data visualization platform that addresses these exploratory concerns. To a certain degree, this is true yet lacks many of the graphing benefits a user of a Power BI, Tableau or SAS Visual Analytics is accustomed to. New features like predictive audiences and anomaly detection are steps in the right direction, but the lack of transparency or customizable experimentation limits the massive potential from an advanced analytical perspective. SAS provides a robust number of analysis and modeling capabilities, from no-code visualization to low-code model pipelining and high-code programming windows using the proprietary SAS language. Trends of DIFM auto-insights through DIY-derived insights heavily influence the current state and future of SAS software. With that said, SAS never intended to build a solution exactly like GA4, rather it provides complementary solutions when a brand is ready to mature their analytical usage and actionability of digital customer data.   Image 22: Building Analyses & Models   Conclusion: Anticipating an unprecedented year of Marketing AI hype, the opportunity of deriving incremental insight from 1st party digital customer data will be in focus. As consumers continue to spend more time interacting with brands digitally, the innovation within data science is begging every company to evaluate their technology stacks, analyst team skills and ROI on customer targeting initiatives. Our viewpoint is simple when comparing GA4 and SAS, and if a brand's investment budget allows, together is better. Whether through integration use cases, or leveraging the solutions individually to tackle specific use cases where one is better suited than the other, the value proposition is simple. To grow incrementally, brands need more business opportunities to take advantage of.   This naturally pivots to the subject of ModelOps, which focuses on getting AI models through validation, testing and deployment phases as quickly as possible, while ensuring quality results. It also focuses on ongoing monitoring, retraining and governance of models to ensure peak performance and that decisions are transparent. For readers who have spent more time in GA4 vs. SAS, this topic of ModelOps may (or may not) be familiar unless you have explored other modules within the Google Cloud Platform.   If topics like TensorFlow Extended, Vertex AI, BigQuery and Cloud Build don't come to mind, it's because Google requires it's users to navigate outside of GA4 to address a brand's objectives around ModelOps. From our perspective at SAS, Google offers powerful software to support ModelOps, but requires users to have high-code skills in Python, SQL, etc. to make use of their offerings. If this suits you or your team, then great. But if you want the benefits of ModelOps without the requirements of being high-code users, SAS provides an alternative solution to support no-, low- and high-code users in one platform.   Model management is not a one-time activity, but an essential business process. Models must be well developed and validated to demonstrate that they are working as expected. Outcome analysis is necessary to:   Ensure that the scores derived from applying the model to new data are accurate. Verify that model performance over time remains satisfactory. Other aspects include cataloging and tracking this growing inventory of analytical assets, while providing support for the governance of these models using version control through repeatable and traceable workflows.   Practical considerations for data science emerge when an analysis worthy of addressing your marketing team’s business problem pivots the work stream to taking action via model deployment.   Image 23: ModelOps & Deploying Insights   Conclusion: SAS enables brands to use first-party data to make better decisions using predictive analytics and machine learning in conjunction with business rules across a hub of channel touch points. As your brand's journey into analytical marketing use cases progresses, usage of modeling intellectual property cannot be under-exploited. It’s competitive differentiation awaiting to be deployed. If the ModelOps offerings on the Google Cloud Platform prove challenging to use because of the high-code user requirements, this provides another opportunity to leverage both GA4 and SAS together to address the upward trend of Marketing AI use cases.   All of this culminates into a brand's guiding light framed around scalable customer decisioning. Rooted in a variety of analytical approaches that can be leveraged within a wide set of marketing use cases,  it doesn't matter if one or multiple technology solutions serve as the bridge to the finish line. There is innovation being served from the software industry to appreciate, explore and experiment with. We (at SAS) simply want to help our customers through partnership and adoption, and whether it involves working with an implementation of GA4 to derive incremental value for specific use cases, or using SAS standalone to resolve challenges in orchestrating meaningful customer experiences, one theme is clear.   2024 is showing a strong propensity for how marketing divisions will fall in love with analytics, machine learning and AI for an array of new customer use cases.   Image 24: Scalable Customer Decisioning   Two examples of well known companies describing their experiences and transformation from direct to integrated marketing/analytical-driven brands can be viewed at the links below:   Example 1: The Nature Conservancy [LINK] Example 2: World Wildlife Fund [LINK]   Our vision at SAS is to serve as the market leader in advanced audience creation & targeting, independent of channel, for enterprise customers leveraging complex, disparate data sources and wishing to consistently deliver superior understanding into their customer journeys. In other words, we want to empower brands to practice responsible marketing.    Image 25: 2024 Marketing Technology Themes   Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here.                   ... View more

Helping customers of your brand's owned digital properties find items of interest is useful in almost any situation. In part one and part two of this Customer Recommendation Systems article series, we completed:   An introductory tour of Do-It-For-Me (DIFM) and Do-It-Yourself (DIY) recommendation analysis use cases applied in martech leveraging SAS Customer Intelligence 360 and SAS Visual Data Mining and Machine Learning on SAS Viya A walkthrough of how champion-challenger DIY recommendation analysis elevates support of personalized marketing An explanation of the details behind the DIY analytical techniques (algorithms) generally available in SAS that can be applied to customer and product recommendations A transparent demonstration video of how SAS users can perform DIY recommendation model training and scoring for data-at-rest and customer experience orchestration In part three of this article series, we will:   Evolve from offline training & scoring into online training & scoring for real-time use cases leveraging streaming data-in-motion  Showcase how SAS Customer Intelligence 360 and streaming analytics work together Image 1: Streaming Recommender Analytics   Gone are the days when a data scientist was solely responsible for analytics. Now, everyone within a brand needs to make decisions based on data. At SAS, we are on a mission to enable and empower everyone – from executives to IT to marketing specialists, all to contribute and build a true customer analytics culture. In the context of real-time recommendation systems, SAS is delivering ready-made solutions to help brands get incremental value in areas like Customer Intelligence, Fraud, Risk, and others.    If brands can make every single decision better using data, this creates a competitive advantage that will be unmatched. SAS allows users to innovate everywhere by enabling analytics across all analytical environments and paradigms. Imagine this: no matter how or where you want your analytics – on-premise, cloud, hybrid, through APIs – or where you need it – stream, database, server, edge – analytics everywhere means bringing analytics to wherever there is data. Why is this important and relevant?   1st party customer data has value in various stages. Some is only valuable right when it originates, other data gains value when combined with other data. For example, a superior customer experience requires the ability to analyze relevant data, score and activate 1:1 recommendations when the prospect is directly interacting with the brand. Customer moments matter!   SAS knows that technology needs to work together. Which is how we pioneered the SAS Analytics Life Cycle and continue to perfect the process of extracting intelligence and value from data. Through integrating SAS, open source, and other vendors directly into the life cycle, the intention is to help brands become more powerful and governed, ultimately building trust and inspiring collaboration that leads to innovation.    Image 2: Strategic Focus of SAS Customer Intelligence 360 with AIoT   Now, recommendation analysis leverages customer interaction data to uncover hidden patterns in order to identify related products, services or content to surface for targeting and personalization. This analysis easily extends to other types of use cases to build more customer relevance, especially with recent innovations of new AI/ML techniques.    If you’ve ever used Amazon, Netflix, or YouTube, you’ve experienced the value of recommendation systems firsthand. These sophisticated systems identify recommendations autonomously for individual users based on past purchases and searches, as well as other behaviors. Customers get algorithmic recommendations on additional offerings that are intended to be relevant, valued, and helpful. Consumers can use recommendations to:   Find things that are interesting or useful. Narrow a set of choices. Explore options. Discover new things.   Marketers can enhance offers that proactively build better customer relationships, retention and sales. For example, brands typically realize:   Stronger customer relationships by providing personalization. Higher engagement, click-through and conversion rates. New opportunities for promotion, persuasion, and profitability. Deeper knowledge about customers.     Image 3: Recommendation system use cases in martech   Across the user spectrum, SAS enables DataOps, ModelOps & Customer Experiences. Think of it like this:   SAS is the “always on” universal inferencing engine, in the cloud and at the edge, being the real-time, operationalization platform for recommender models produced by a brand. What exactly does this mean in the context of streaming recommendation systems?   Processing data continuously, on the move, in-memory with very high speed and low latency Flexible Publish and Subscribe framework Performing actions such as filtering, aggregation, pattern detection, correlations, machine learning, geofencing, image processing, etc. Adding historical context or enrichment data to what is being observed of customer behavior in real time Scoring events using trained analytical models from SAS or externally from open-source (Python, R) Design and test projects in the low code, graphical design environment (or high-code if the user prefers) Orchestrate, deploy and monitor Track and update when new recommender champion models are promoted Image 4: SAS for Real Time Customer Recommendation Systems (Detailed View)   In SAS Customer Intelligence 360, events enable users to enhance their ability to understand, target, and interact with customers in a meaningful way. Events are used to track user behavior and to provide input conditions for other items such as spots, segments, tasks, data views, and activities. For example, events are used in some of these ways:   Record when customers view a web page or in a mobile app Observe implicit or explicit customer behaviors Trigger a recommendation system to act   Tracking user interactions is one of the core features that enables your digital properties to respond in real time to customers based on their profile, origin, browsing behavior, and so on. When a visitor interacts with your website (for example, through a click event), that event is processed by the run-time environment and can be used to control many website features.    Image 5: Actionable Event Tracking & Recommendation System Triggering   Tasks coordinate the items/offers/calls-to-action that SAS Customer Intelligence 360 uses to deliver content to a target audience. As a task runs, it collects data that can be used to monitor performance or be used to refine how the task delivers content. Building upon this concept, an activity (or customer journey) is a coordinated series of tasks that are designed to meet the goals of a marketing campaign, such as one that is supported by a recommendation system.   An activity uses tasks and events. It charts the customer paths between tasks, such as recommending a particular message through a specific channel, and conditions, like the primary metric and evaluation periods. For example, users might leverage a mobile task to present a product offer on your brand's app. Then, users could select an email task to send a discount offer to all customers who interacted with the mobile recommendation within the last week (but haven't converted yet).   Image 6: Recommender Activity Map   In SAS Customer Intelligence 360, users can create tasks (Web, Mobile, etc.) that display different creatives based either on a product being viewed or a customer’s behavior. There are two methods for delivering recommendations to users. User-centric recommendations take a user’s behavior into account. Product-centric recommendations are based solely on a product.   Martech industry solutions frequently offer "easy-button" or automated analytical solutions that over-promise the potential of machine learning and AI. In the end, for readers who have used DIFM (Do-It-For-Me) features, they automate analytical model templates with limited abilities to accommodate customization. Our viewpoint at SAS is the availability of DIFM features in software is important, especially in the absence of any analytical enhancements to a brand's present-day use cases. However, the desire to incrementally improve on DIFM technology features allows us to pivot to DIY (Do-It-Yourself) approaches in recommendation analysis.   A sampling of native recommendation analysis algorithms available in SAS include regularized & non-negative matrix factorization, k-nearest neighbor, bayesian personalized ranking, factorization machines, data translation w/ optimal step-size, slope one, market basket, link analysis, and the list goes on. Additionally, SAS supports usage of open source (Python/R) recommender packages. Given the high volume of algorithms to select from as an analyst, SAS enables champion-challenger recommender modeling prior to deployment.   Recommendation Algorithm Candidates To Improve Offer Performance   Let's highlight three algorithmic approaches leveraging factorization machines (FMs), bayesian personalized ranking (BPR) & data translation w/ optimal step-size (DTOS). Given we have covered FMs in an earlier article, let's briefly describe the others:   Bayesian Personalized Ranking (BPR)   BPR as an algorithm can be applied to creating personalized recommendations of items for users on the basis of the users’ implicit feedback (such as web/mobile clicks or purchase history). BPR is a common method that is designed specifically to optimize recommendation ranking and has shown superior performance compared to other standard analysis techniques that are widely used to analyze explicit feedback. In the scenario of implicit feedback, an observation or event from an instrumented website or mobile app using SAS Customer Intelligence 360 captures the required input data which simply consists of a user (or customer) and an item (or product/service).   Among the available recommendation methods, collaborative filtering, matrix factorization and factorization machines  have shown to be effective approaches, and many researchers and brands have focused on these methods. For example, a factorization machine model is a general factorization model that considers both latent and auxiliary features, and it includes and mimics many basic collaborative filtering methods under various scenarios. Although factorization machines have shown good performance in both model prediction and computational complexity, the majority of factorization machine methods are designed for data that contains explicit feedback, whereas only a limited number of approaches have been proposed for data that contains implicit feedback. An alternative is to model the likelihood of ranking between items to utilize a new optimization criterion, Bayesian personalized ranking (BPR), for analyzing implicit feedback with item features which can have a significant influence on model performance.    Data Translation With Optimal Step Size (DTOS)   From the perspective of customers, a recommender provides personalized recommendation by helping users to find interesting items (products, movies, music, etc). From the perspective of products, a recommender performs targeted advertising by identifying potential users that would be interested in a particular item. The information about users, items, and user-item interactions constitute the data that are used to achieve the goal of recommenders. Among the three types of information, user-item interactions are essential. Recommenders employing user-item interactions alone, without requiring the information of users or items, is based on collaborative filtering.  Typically, each user rates only a fraction of items and each item receive ratings from only a fraction of users, making an incomplete data matrix with only a fraction of entries observed. In this matrix formulation, the goal of recommenders, specifically collaborative filtering, becomes predicting the missing entries so as to locate the interesting items or potential users. A major bottleneck  is the reliance on singular value decomposition (SVD), limiting its use in large-scale applications.   An alternative approach to collaborative filtering is matrix factorization (MF), which models the user-item interactions as a product of two factor matrices. Each user or item is represented by a vector, and a rating entry is represented by the inner product of two vectors. These vectors can be considered as a feature representation of the users and items. As they are not observed, but rather are inferred from user-item interactions, these vectors are commonly referred to as latent features or factors. Moreover, the latent features of all users and all items may be inferred simultaneously, making it possible to incorporate the benefit of multitask learning (MTL). By the principle of MTL, the feature vector of each user is not only influenced by its own rating history, but also by the rating histories of other users, with the extent of influence dictated by the similarity between users. For this reason, a user may discover new interesting items from the rating histories of its peers who share similar interests, with the similarity identified from all users’ rating histories.   A widely adopted algorithm for learning MF models is Alternating Least Squares (ALS), which updates the two factor matrices alternately, keeping one fixed while updating the other. Given one matrix, ALS optimizes the other by solving a least squares (LS) problem for each user or item. As the LS solution is optimal, ALS can improve the learning objective aggressively in each iteration, leading to convergence in a small number of iterations. However, different users may have rated different items and, similarly, different items may have been rated by different users; thus, this leads to high computational cost in each iteration of ALS.   This issue can addressed with a softImpute-ALS algorithm, but sub-optimal results in applying this method has led SAS to introduce a new algorithm, termed Data Translation with Optimal Step-size (DTOS), to alleviate these drawbacks. As the name indicates, DTOS first performs data augmentation (or translation), an equivalent to the imputation step of softImpute-ALS. However, DTOS goes one step further to construct a set of solutions, with the softImpute-ALS solution included in the set as special element. The solutions are parameterized by a scalar that plays the role of step-size in gradient descent. The step-size is optimized by DTOS to find the solution that maximizes the original objective. The optimization guarantees a larger improvement of the original objective compared to the improvement achieved by softImpute-ALS, with this helping to alleviate the issue of slow progress per iteration and thus to speed up convergence. Thanks to the quadratic objective, the optimal step-size can be obtained in closed-form and its calculation does not introduce significant additional cost of computation; thus, DTOS has almost the same per-iteration computational complexity as softImpute-ALS. With the low cost per iteration and more aggressive improvement of the learning objective, DTOS blends the advantage of softImpute-ALS into that of ALS, and is expected to achieve a high performance-to-cost ratio.   In other words, DTOS is a fast algorithm for training recommender systems on implicit feedback. Users can leverage the DTOS action in SAS as a distributed, multithreaded implementation. The recommender model that the DTOS algorithm trains is represented by matrix factorization with partially defined factors (MF-PDF), a model that generalizes matrix factorization (MF) to include predefined factors (PDF) of users and/or items.   Image 6: SAS for Champion-Challenger Recommender Modeling   Demystifying Champion-Challenger Do-It-Yourself (DIY) Recommendation Analysis For Martech   In general, it is a good practice to develop multiple AI models that support the same task. The reason for this is simple: if one model fails or the performance of that model degrades over time, there is always another model that can take over. For those unfamiliar with this approach, the champion model is the best model that is chosen from a pool of candidate models. In the machine learning ecosystem, this approach is often referred to as the champion-challenger approach, where the champion model is the model that currently has the best performance for the AI task at hand. Before users identify the champion model, they can evaluate the structure, performance, and resilience of candidate models. Users leverage challenger models to test the strength of champion models.    The champion model is the model that typically runs in production and is continuously challenged by the challenger models. As soon as the champion model fails or one of the challenger models defeats the champion model, the current champion model can be quickly replaced, and the continuity of the AI system can be guaranteed.   Image 7: Event Stream Processing Engine   What does the difference between various types of recommender algorithms look like when it comes to metrics? There are several metrics to evaluate the performance of models in a champion-challenger context. In a forthcoming demo video available below, we will use the following metrics to make comparisons:   Area Under The Curve (AUC): AUC measures the likelihood that a random relevant item is ranked higher than a random irrelevant item. Higher the likelihood of this happening implies a higher AUC score meaning a better recommendation system.  Hit Rate (HR): The hit ratio is simply the fraction of users for which the correct answer is included in the generated recommendation list (top 10 for example) extracted from all users in the test (or validation) modeling data. Mean Reciprocal Rank (MRR): MRR calculates an average of reciprocal of ranks given to the relevant items. So if the relevant items are ranked higher, the reciprocal of the ranks would be lower leading to a lower metric score, as desired. Essentially, the idea behind evaluating a recommendation system is to make use of ranks given to the relevant items and translate into a single number indicating how good or bad the ranks are.   Image 78 Champion-Challenger Recommender Modeling Assessment Report   Please keep in mind, comparing the performance of recommendation models is not limited to these three metrics only.   Recommendation System Scenarios   Let's articulate two scenarios for designing and deploying a recommendation system:   Offline Training & Online Scoring   Offline training is performed on a snapshot of data at rest to produce scoring code. That scoring code can then be deployed into an online scoring system on forthcoming (new) streaming data produced by customers and prospects interacting with your brand. Keep in mind, the scoring on new data is based on a modeling solution that was trained offline (or in batch). Thus, the solution would need to be retrained at signs of performance decay. For a detailed demo video on this scenario (within the context of financial services), please check out this article. For SAS user documentation on developing offline recommenders, go here.   Image 9: Offline Training & Online Scoring   Online Training & Scoring   The demonstration video below (in the context of the retail industry) is based on a true online training and scoring solution for recommendation systems.  The incremental value proposition here is represented by a recommender modeling solution that trained once on a snapshot of data-at-rest in SAS (to mitigate issues like cold start problems), deploy, and now online training and scoring handles the re-training of the model and associated scoring on a 1:1 basis. For SAS user documentation on developing online recommenders, go here.   Image 10: Online Training & Scoring   Real-Time Customer Recommendation Systems For Data-In-Motion (Demo Video)   Using a SAS for Retail website, let's walkthrough a technology demonstration where SAS Customer Intelligence 360 is leveraged for an AIoT use case in the context of recommenders.   Readers can scroll or swipe up to revisit Image 6 and the customer journey activity map that will entail the following:   The capture, contextualization and input streaming of digital interactions between a prospective customer and the brand's digital property The event monitoring and triggering of a champion recommender model   Image 11: Demo Flow - Part 1   Training and scoring on data-in-motion The output of the recommender scoring will drive immediate actioning, targeting and personalization back into the customer's digital experience Image 12: Demo Flow - Part 2   Now that we have summarized the demo flow, view the presentation/demo video below for a summary of this use case, as well as live technology at work.     When we consider a customer experience with a brand, moments matter. Consumer are empowered by their digital devices, and can shift their attention from one brand to the next in a matter of seconds. The use cases for recommendation systems are expanding every day, across the entire martech industry. We look forward to what the future brings in our development process – as we enable technology users to access all of the most recent SAS analytical developments. In the near future, watch for SAS to release:   Deep learning based recommenders (multi-headed attention) Reinforcement learning based recommenders Graph convolution based recommenders    Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here.     ... View more

As we approach the holiday season of 2023, let's explore the challenge of addressing the empowered consumer. It's readily recognized customers are digitally savvy, discerning and motivated to get the best deal. This has made it increasingly difficult for brands to develop pricing strategies that optimize net revenue.   Customer-centric pricing can be a game-changer. Whether an online or brick-and-mortar shop, pricing personalization aims to use data and analytical insight to influence what is (or isn't) offered to each prospective buyer. Is the price a part of the shopping experience? Absolutely. Price is one of the major factors that forge a consumer's buying decision and loyalty.   The direct interaction between brands and customers (think web, mobile app, email, etc.) enable the opportunity to implement personalized pricing more effectively. Brands collect 1st party data on the consumer's engagement with their  product offerings, and can use this information to develop offer strategies. Over time, our consultative engagements here at SAS with various brands have shown common challenges:   Storage and effective use of high quality, un-sampled 1st party data on user behavior, purchase history, etc. Optimizing net revenue with pricing tactics   Pricing personalization can be a useful tool to help brands reach different objectives, depending on their business model, market conditions, and customer segments. For instance, brands can increase sales volume by targeting price-sensitive customers with lower prices while maintaining higher margins from less price-sensitive consumers. Sounds logical, rationale and intelligent, right?  While pricing is often described as a science by practitioners, it is a key factor to drive conversions. And we want to do this while optimizing profit, which points to a secondary challenge in regard to which customers should receive a marketing intervention (or stimuli), and which ones shouldn't.   Image 1: Pricing Personalization   Setting the right price for a good or service is an old problem in economic theory. Keep in mind, there are a vast amount of pricing strategies in existence that depend on the objective sought. One brand may seek to maximize profitability on each unit sold or on the overall market share, while another company needs to access a new market or to protect an existing one. Moreover, different scenarios can coexist in the same company for different goods or customer segments. Although strategies like premium and penetration pricing have existed for many years, let's focus on the use of algorithms to address this challenge.   Algorithmic personalized pricing is a process of setting optimal offers using the power of machine learning and artificial intelligence to maximize revenue, increase profit or address other business goals set by brands. Algorithmic personalized pricing is one of the most powerful means of gaining a competitive advantage.   Introducing SAS PROC DEEPPRICE to personalize prices and optimize revenue   One challenge brands will always face relates to the heterogeneous characteristics of their customers. Such variability can affect how customers choose to (or not to) convert from a targeted marketing intervention (such as a price increase or decrease). Understanding customer response behavior to targeted offerings is crucial for informing individualized pricing decisions. PROC DEEPPRICE offers a flexible framework for specifying and estimating customer responses to marketing treatments.   PROC DEEPPRICE enables SAS users to specify customer treatment effects (or marketing tactics) as unknown functions of observed customer behaviors. For readers who are finding this technical, let's break this down.    A ‘customer treatment effect’ is the causal effect of a variable (like price) on a target outcome (like purchase conversion). The term ‘treatment effect’ originates in medical literature concerned with the causal effects of yes-or-no ‘treatments’, such as an experimental drug or a new surgical procedure. The term is now used much more generally, and can be applied in marketing and customer experience. 'Unknown functions of observed customer behaviors' originates from data science and specifically nonparametric regression which is a category of regression analysis in which a predictor (such as a customer behavior signal) does not take a predetermined form (like linear or quadratic regression) but is constructed according to information derived from the data itself. That is, no parametric form is assumed for the relationship between predictors and target outcome variable.   Moving on, PROC DEEPPRICE uses Deep Neural Networks to estimate unknown functions. In other words, SAS is providing users (analysts/data scientists who support customer experience use cases) native Marketing AI capabilities through deep learning to overcome obstacles in improving revenue-centric KPIs. Although other software vendors like to broadcast how they have AI capabilities, there is much more to explore and benefit from the discipline beyond simply generative AI (which is a subset of what AI is capable of).   We believe the 2023 incremental excitement surrounding AI has flipped the marketing industry's  perspective upside down in how to benefit from this phenomenon.  The most challenging problems in marketing (and enterprises) can be addressed with analytical innovation and best practices across the entirety of AI capabilities.   Image 2: Welcome to the Marketing AI Party   Pivoting back, such a rich and flexible approach through PROC DEEPPRICE accounts for individual customer variability in treatment responses providing the advantage of extracting clear insights from complex forms of heterogeneous behaviors (even with large data sets). It enables brands to determine, for example, what types of consumers are price sensitive (and which aren't).    PROC DEEPPRICE enables users to perform policy analysis to compare outcomes under various hypothetical treatments. For those unfamiliar with this term, 'policy analysis' is the process of identifying potential options that could address a business problem and then comparing those options to choose the most effective, efficient, and feasible one. For example, it can answer questions such as:   What is the best pricing strategy to optimize revenue? Subsequently, marketing teams  can leverage these insights for pricing personalization and improving revenue metrics.   Use Case & Demo: Online Media Brand   Let’s look at an example of how an online media brand can offer targeted discounts through personalized pricing to optimize revenue. The  data set is provided by the Microsoft research project ALICE. For more details, see Example 15.1 in the SAS Help Center documentation. The data set has 10,000 simulated observations that represent user personal characteristics. Also taken into account is user online behavior history, including previous purchases and previous online times per week. The treatment variable, price, is the price the customer was exposed to during the discount season. The outcome variable, demand, is the number of songs that the customer purchased during the discount season. The image below summarizes the analysis table through showing the names of the variables that are used in the model, along with their type and definition.   Image 3: Analysis table   Now, let's summarize our intention with this data table. This use case example has three goals:   Estimate the parameters (or customer signals) of interest, such as average partial effect (or influence) of price on demand. In other words, we want to understand how much influence pricing has on purchase conversions. Estimate the price elasticity of demand for product offerings (songs), and assess its variation with customer characteristics. The translation here is what impact will pricing adjustments (increase on decrease) have on purchasing behavior. Conduct a policy analysis that aims at increasing revenue despite decreasing the purchase price for some customers. The analysis results below will show eight different pricing strategies across a customer universe and determine which policy (or scenario) generates the most revenue.   Next, the following Flow Swim Lane in SAS Studio (Image 4 below) contains initial steps which estimate the effect of price on the demand for songs purchased and the details of the estimation are used later for strategic pricing policy evaluation.   Image 4: SAS Studio Flow for estimates the effect of price on product demand   One output from the analysis Swim Lane shown below in Image 5 is the estimated average slope. It represents the average impact of price on purchase demand for all the customers in this data. It is negative (-9.78) and statistically significant. This is an expected result in accordance with economic theory as many readers will recognize that as the price of items increases, demand decreases.   Image 5: Average Partial Effect of Price on Purchase Demand   Another customer data signal of interest is the price elasticity of demand, which measures the degree of sensitivity of demand to price. In general, demand decreases when price increases for most products, but the amount by which demand decreases is greater for some products than for others.  PROC DEEPPRICE can be used to compute individual-specific elasticities. It is of particular interest to assess how price elasticity varies among customers with respect to their income level (as well as other personal attributes).    Image 6: Sales Price Elasticities By Customer Income Level   Figure 6 above shows a remarkable discovery among two sub-groups of homogeneous customers with respect to income and their response behavior related to a price increase. Customers whose income segment (portrayed on the x-axis) is less than 1 (representing ~51% of all customers in this data) are more sensitive to a price increase; if price goes up (for example, by 1%), their number of songs purchased falls by 1% to 5%, compared to ~ 0.15% decrease for higher-income customers. Right away, from a marketing and pricing perspective, the segmentation insights are profound. One group is exclusively price insensitive with elasticity close to zero where discounting would be wasteful. The second group has a higher variance and highlights the price-sensitive customers where discounting can be effective.    Analysts can explore the relationship with other customer attributes to collect additional insights.  For example, the variability of price elasticities with income and days_visited is shown in Figure 7. The variable days_visited represents the average number of days per week a customer engages with the brand's website. Low-income customers who visited the brand's website less often (days_visited <5) are even more price-sensitive than low-income customers who visited the website more often (days_visited >=5). The latter group has a price elasticity between –2 and –1, compared to a range of –5 to –1 for the former. This information is vitally important for targeting customers with (or without) a price discount.   Image 7: Sales Price Elasticities By Customer Income Level & Weekly Digital Engagement   Translating Noisy Heterogeneity From Price Effects Into Actionable, Personalized Prices   In policy optimization, the goal is to maximize the expected utility of a policy. For the definition of the expected utility function and how it is estimated, as well as details such as the definition of a policy rule, see the SAS Help Center page summarizing the function.   In our working example, the policy decisions are to choose customer-specific prices in order to maximize the revenue, which is defined as: Image 8: Revenue maximization formula This particular formula for pricing in Image 8 seeks to maximize revenue in a personalized manner, but can exceed the original prices which can be viewed as a risky customer approach by the brand. The variability observed in Images 6 and 7 above show price elasticities suggesting that this online media company should consider a price discounting strategy that targets a specific segment vs. all customers. Whether the company’s revenue will rise or fall depends on the price elasticity of product demand. Microeconomic theory predicts that lowering prices will increase revenue if demand is price-elastic (elasticity < –1) and decrease revenue if demand is price-inelastic (elasticity > –1). Image 9 below shows a report summary of SAS performing an evaluation of the revenue-maximizing policy s1, as well as six other discounting strategies for comparison with the observed pricing policy approach.   Image 9: Policy evaluation report To be clear, as an analyst, the input training data for this exercise contains the observed number of products (songs) purchased at the observed prices. Thus, it is straight forward to compute the corresponding revenue per customer, but we wish to learn what the revenue per customer would be if the online media company had set a different price. For example, consider the different price strategies described in Image 9. The objective is to compare each of the hypothetical pricing policies with the observed prices to determine which strategy generates the most revenue. The SCORE and INFER statements within PROC DEEPPRICE enable analysts to do that after estimating the price effects and saving the details of the estimation (Image 10).   Image 10: SAS Studio Flow Swim Lane for policy evaluation, comparison and optimized price scoring   The results of this robust policy comparison are plotted in Image 11 below. Each of the hypothetical policies are compared to the observed policy price, and each blue-filled marker (some might see a resemblance to a Star Wars TIE Fighter) represents the monetary difference between each policy and estimated revenue per customer. The green vertical dotted line represents a revenue difference of zero (useful as a benchmark comparison). Markers on the left of the green line represent policies that are worse (revenue loss), and those on the right represent policies that are better than the observed strategy (revenue gain).   The optimal personalized revenue-maximizing policy, s1, mentioned earlier is clearly the best in the lower-right area of the graph. However, because prices under s1 can exceed the original prices, this can contradict how a brand desires to facilitate customer experiences. This is where human-driven marketing strategy and AI come together.   Image 11 Pricing policy evaluation results   The discounted optimal personalized policy, s1d, is more meaningful as it performed a policy analysis with the constraint of never increasing the original product price for a customer. It resulted in an estimated net revenue improvement of $0.63 per customer. As a comparison, evaluated policies that offer every prospect a discount (s3 and s5) are worse than the observed approach (as they index below the baseline threshold).   PROC DEEPPRICE offers a unified framework for estimating heterogeneous causal effects of treatments and performing policy analysis. Its flexibility has enabled us to specify individual price effects as a function of individual customer characteristics. The estimated individual price effects were then translated into individualized prices. These were used, along with other hypothetical pricing strategies, to show how the online media company can set user-specific prices to achieve maximum revenue.   Now, let's activate on these insights to show how a brand can leverage the optimal discount policy to target a specific group of customers with the objective of increasing revenue.   SAS Customer Intelligence 360 - Take Us Away!   Brands today are complex ecosystems of decisions that must be executed with increasing levels of automation - due to their competitors digitally transforming and influencing customer expectations. In response, there is a need to change how decisions are made.  Organizations have the opportunity to increase their capability to perform augmented decision making - where a human takes analytically driven insight to support decisioning (such as within promotional campaigns through outbound emails, or inbound 1:1 personalized interactions via website or mobile app). With each passing year, the acceleration of the scale, speed and complexity of customer 1:1 decisions is increasing.   SAS Customer Intelligence 360 is a martech solution to support adaptive planning, journey activation, personalization and AI-elevated decisioning to help users create appealing, moments-based customer experiences that boost profitability and strengthen brand loyalty. Within it, marketers are provided out-of-the-box (OOTB) connectors that are preconfigured for data integration with external applications (such as SAS Viya). Brands can use connectors to retrieve or transfer data between SAS Customer Intelligence 360 and on-premises or cloud-based applications. Our design intent with these connectors is to offer brands time-to-value acceleration in activating commonly used data management and analytical flows, such as the actionable output of sophisticated AI models that get created from procedures like PROC DEEPPRICE.   Now what would a software technology company like SAS with guerrilla-like data science powers as well as journey orchestration and marketing activation capabilities be up to here? Let's bring it home...   Here's the situation. The SAS Store is a retail brand with a digital presence.    Image 12: SAS Store website   Before proceeding, this website is instrumented with SAS Customer Intelligence 360 for digital interaction collection, contextualization, targeting and personalization. With that stated, the story line begins with customer engagement. An activity within SAS is a coordinated series of interaction tasks and events that are designed to meet the goals of a marketing objective. An activity map charts the customer paths between tasks, such as sending a particular message through a particular channel, and conditions, such as the primary measurement metric and evaluation periods. For example, users might use a variety of customer engagement tasks to raise awareness and present a product discount from their brand's website.   Image 13: Customer journey for pricing personalization   The tasks on the left side of Image 13 represent four mechanisms to qualify a prospective customer for this journey.    Facebook (Meta) Ads task to target customers with a personalized pricing advertisement related to a product category Google Ads task to target customers with a personalized pricing advertisement related to a product category Web event monitoring for 1st party customer behavior that would qualify prospective interest in a product category Email task to target customers with a personalized pricing advertisement related to a product category   Each of these targeting tactics carry the same objective to provide a 1:1 model-driven pricing offer to entice the customer to interact with the SAS Store brand digitally and carry out further personalization to increase the efficiency of purchase conversion. Let's walk through the email task for transparency.   Image 14: Email content with dynamic merge tags and variables   The use of merge tags and variables enables a marketer to personalize the email content in a 1:1 manner.  Users can use merge tags to personalize the text of the creative for each recipient. When users enter a merge tag in a creative, the available merge tags are retrieved from uploaded customer data and are displayed consistently across tasks. For example, let's look up my Customer ID profile:   Image 15: Customer identity profile   Image 15 highlights the fragmentation of identity management. Across a variety of cookie, browser, device and interaction keys, one master identity  profile is deterministically linked with one customer. Digging deeper into the customer state service, we can observe available merge tags available for personalization. The property 'S1_Discount' is an example of preloaded 1:1 scoring extracted from the PROC DEEPPRICE exercise to produce the discounted optimal personalized policy. Recall, the 's1d' policy discussed earlier in this article resulted in the most meaningful outcome as it performed a policy analysis with the constraint of never increasing the original product price for a customer while maximizing estimated net revenue.   Image 16: Customer state service and personalization properties   So, for my customer profile associated with the email address displayed in Image 16, users can perform responsive previews of the personalized email content to test the targeting results.   Image 17: Responsive preview of personalized pricing email    Tracking user interactions is one of the core features that enables a brand to respond in real time to users based on their profile, origin, browsing behavior, and so on. The interactions that are monitored are the driving force behind many of the features that SAS Customer Intelligence 360 offers.  Assuming this, now I receive an email from the SAS Store brand.   Image 18: Delivered email with 1:1 pricing personalization   As many of us can anticipate, I click-through on the email and I'm redirected to the brand's digital property. Based on the activity map highlighted above in Image 13, I will now receive a 1:1 personalized pricing treatment within the lower middle spot of the site's homepage.   Image 19: Website pricing personalization and retargeting       Although I could go on with this use case, the transparency and value proposition of activating the AI-insights of PROC DEEPPRICE through SAS Customer Intelligence 360 has been demonstrated through screenshots. If readers prefer to view a demo, please enjoy the short video below:     For readers who have a desire to learn more, interactive demo videos for active users of SAS Customer Intelligence 360 are available here. Finally, go here to gain incremental awareness about how SAS can be applied for customer analytics, journey personalization, AI decisioning and integrated marketing.     ... View more

It's no secret who the two biggest players in digital ad media are. Google and Meta (Facebook) quickly come to mind. For those who aren't familiar, Google Ads and Meta (Facebook) Ads are online advertising platforms enabling marketers (and brands) to find customers.    When we consider Google Ads, brands have the opportunity to connect with current and prospective customers across Search, Display, Shopping, Video and App. Pivoting to Meta Ads, marketers can reach new/existing customers as well as their networked communities on Facebook, Instagram, Messenger and WhatsApp.    With this stated, reflect for a moment on the business opportunity to:   Maximize qualified leads and conversions Increase online sales Drive in-store foot traffic (or send more users to your website) Show your brand to more people to increase awareness, reach and engagement Market your app to new users (or increase app installs) Now consider how many human beings on Planet Earth use Google and Meta (Facebook). It's not thousands or millions. It's billions. Hopefully we have your attention, and let's proceed to connect the dots to SAS.   SAS Customer Intelligence 360 is a martech solution to support adaptive planning, journey activation, personalization and AI-elevated decisioning to help users create appealing, moments-based customer experiences that boost profitability and strengthen brand loyalty. Within it, marketers are provided out-of-the-box (OOTB) connectors that are preconfigured for data integration with external applications. Brands can use connectors to retrieve or transfer data between SAS Customer Intelligence 360 and on-premises or cloud-based applications. Our design intent with these OOTB connectors is to offer brands time-to-value acceleration in activating commonly used data integration flows.   Now what would a software technology company like SAS with guerrilla-like data science powers that can be applied to customer analytic topics like segmentation, propensities and optimization, as well as journey orchestration and marketing activation capabilities be up to here?   Image 1: Together Is Better   The Google Ads Connector   SAS Customer Intelligence 360 provides a Google Ads connector that marketers can use to access the Google Ads API (no coding required) and to set up data integration with the ad platform. By using the connector, users can select a segment of their 1st party customer data that is stored in SAS Customer Intelligence 360 and upload it to Google Ads. The Customer Match feature from Google Ads enables SAS users to target a rule-driven or algorithmically-derived segment of your existing customers and deliver personalized messages via Google Ads.   For emphasis, when SAS states algorithmically-defined segment using your brand's 1st party data, this is where together is better. Do-it-for-me (DIFM) or do-it-yourself (DIY) segmentation specialists, everyone is welcome to the paid search and display media party!   Google Ads uses the uploaded data from SAS to build a Customer Match segment of your selected customers (who are also Google users), and enables brands to engage with them through Google applications such as Search, Display, Shopping, Video and App. Users can define the customer segment that you want to target by using any of these identifiers:   Email address Customer Identifier Phone number Name Location Marketers can use the configurable templates that are provided by SAS Customer Intelligence 360 to configure multiple out-of-the-box (OOTB) connectors that are associated with their Google Ads account(s). To use Google Ads and the Google Ads APIs, brands need to create a Google Ads business account and a Google manager account. After creating and configuring the Google Ads connector, to synchronize your selected customer data with Google Ads, users simply create a Google Ads task in SAS Customer Intelligence 360.   Image 2: SAS Customer Intelligence 360 Connector For Google Ads   The Meta (Facebook) Ads Connector   SAS Customer Intelligence 360 provides a Meta (Facebook) Audience connector that brands can use to access Meta’s marketing APIs (once again, no coding required) and to set up data integration with the ad platform. Similar to the Google Ads connector, users can select a segment of their brand's 1st party customer data that is stored in SAS Customer Intelligence 360 and upload it to Meta Ads.   Another use case for AI-driven audience segmentation points to Supervised Learning and Profit Matrices prior to activating customers in Meta Ads. Determining an appropriate cutoff as to who to target (and who shouldn't) with paid media advertising is critical from a revenue optimization perspective. Leveraging an optimization approach using a Profit Matrix, which is a formal approach to determining the optimal cutoff of audience segmentation, meets this objective. The approach starts by assigning profit margins to true positives and loss margins to false positives. The optimal decision rule maximizes the total expected profit. As an analyst/data scientist, when one can communicate the value of modeling efforts in monetary terms, every executive paying attention is going to lean in and focus. Passing these insights to influence our marketing teammates will directly impact their segmentation strategies and KPIs. Remember friends, data science and martech together is better!   Returning back to the Meta Ads connector, the uploaded data from SAS is used to build a custom audience segment of Facebook, Instagram, Messenger or WhatsApp users whom marketers can target with personalized content. Users can define custom audiences by using any of these identifiers:   Email address Customer Identifier Phone number Name Location Matching the user experience with the Google Ads connector, marketers can use the configurable templates that are provided by SAS Customer Intelligence 360 to configure connections to their Meta accounts by accessing the relevant applications. To access the Meta marketing APIs, users need to set up a Meta developer account. To upload custom audiences to Meta Ads, brands also need to set up a Meta Business Manager account and associate the Meta ad accounts to the Meta Business Manager account. This will allow SAS Meta (Facebook) Audience connector to use the Meta Graph API to read from and write data into the Meta Ads platform. After creating and configuring the Meta Ads connector, to synchronize your selected customer data with Meta Ads, users simply create a Meta (Facebook) Ads task in SAS Customer Intelligence 360.   Image 3: SAS Customer Intelligence 360 Connector For Meta (Facebook) Ads   Customer Match   In the context of Google Ads, Customer Match lets marketers use online and offline data to reach and re-engage with customers. Using information that your brand’s customers have shared, Customer Match will target ads to those customers and other customers like them. Customer Match is useful for many business goals, from increasing brand awareness to driving conversions. Here are a few examples of different audiences marketers can target with Google Ads and Customer Match:   On the Google Search Network and the Shopping tab, users can optimize campaigns by adjusting bid amounts based on customer activities. On Gmail, brands can reach customers or new potential customers with similar interests using personalized ads at the top of inbox tabs. On YouTube, marketers can reach new segments, by targeting segments similar to your brand’s most valuable customers. On Display, reach customers or new potential customers with similar interests using personalized ads on the Google Display Network. Your brand’s match rate using the Customer Match feature is the percentage of the uploaded customer list that could be connected to Google users. In other words, how much of your first party customer data can actually be leveraged for campaign use cases.   Image 4: Google Ads - Audience Manager Interface   Users can create a customer data file based on 1 st party information (for example, customers who have signed up for your business’s newsletter, customers who have bought from you before that have churned, etc.). Your list will be uploaded by a secure hashing algorithm, such as turning the emails into codes that cannot be unencrypted. Google does not receive the actual email addresses, nor can Google see the contents of the list. Google accounts are also in hashed codes, and the codes are compared to determine if there’s a match.   After this process is done, your data file is deleted. Whether or not there’s a match, Google doesn’t keep this data or use it for any other purposes. In accordance with Google Ads user best practices, to improve your match rate, SAS recommends adding as many match keys (i.e., customer identifier, email, etc.) as possible.   Image 5: SAS Customer Intelligence 360 for Customer Data Matching   In the context of Meta (Facebook) Ads, it’s similar. A custom audience made from a customer list is a type of audience marketers can create to connect with people who have already shown an interest in your brand’s business or product. Prior to use, Meta hashes shared customer information. Then, Meta will use a process called Matching to connect the hashed information with Meta technologies profiles so that users can advertise to their customers on Facebook, Instagram and the Meta Audience Network. The more information brands can provide, the better the match rate. Additionally, Meta doesn't learn any new identifying information about your customers.   Moving on, most digital ad targeting use cases tend to NOT include 1st party customer data. Frequently, when we (SAS) work with our customers, it is a common observation to view search and social ad teams primarily using the behavior of anonymous visitors, and here is where SAS comes into play!   Because SAS Customer Intelligence 360 can automate the process of segmentation, activation and auto-scheduling, the process of creating and uploading analytically-driven 1st party segments for Google & Meta Ads can avoid manual and infrequent audience updates.   Image 6: SAS Customer Intelligence 360 For Automation   If you consider yourself an analyst or data scientist who helps your organization’s marketing efforts, aligning yourself with teammates who manage Google & Meta Ad campaigns (and their associated budgets) is the opportunity being advocated. For example, a key value proposition is the ability to scale with automation. However, it gets better when analytical segmentation insights layer on a second value proposition to motivate the union. In other words, think bigger about the variety of use cases that can be addressed.   Image 7: SAS For 1st Party Segment Use Cases In Paid Media   SAS advocates the topic of Customer Match as an important consideration in support of brands desiring the maximum benefit of investing in paid media campaigns. In support of this statement, let me share a short story.   A best practice conducted internally prior to the release of new user features in SAS Customer Intelligence 360 is working with our current customers as early adopters. A non-profit globally recognized brand recently presented at SAS Explore 2023 on the topic of constituent journey management and digital personalization. A portion of their presentation story highlighted their efforts to use tasks within SAS for Google and Facebook Ad campaigns to target and re-engage inactive donors. A relevant and important milestone in their experience centered on challenges with Customer Matching.   Image 8: Early Adopter Insights   This challenge exists for both Google and Meta (Facebook) Ads across any industry. So, what did this non-profit brand do to address the obstacle? It was a collaborative effort between the organization’s analytics & digital marketing teams. Supervised learning, champion-challenger modeling and the design of a customer match rate estimation model was an innovative solution to help the company maximize the match rates between Meta & Google Ads (above and beyond the traditional best practices of using multiple identifiers). The beauty of this was its simplicity!   Simply adjust your thinking of predicting who is likely to convert and pivot to who is likely to be matched. Now you get it…   Image 9: Supervised Learning For Customer Match Rate Estimation   The results of these match rate models provided improved clarity and expanded the marketable universe. In addition, the brand knew which customers should be prioritized for Meta Ads vs. Google Ads driven by predictive match rate propensities. In other words, the brand optimized (or maximized) the match rates using a data-driven approach. Nicely done!   Closed Loop Campaign Management   A customer journey in its purest form represents a series of brand-orchestrated connected experiences addressing customer desires and needs — whether that be completing a desired task or traversing the end-to-end journey from prospect to customer to loyal advocate. When you reflect on this, the customer experience is the totality of cognitive, affective, sensory, and behavioral consumer responses during all stages of the consumption process including pre-purchase, consumption, and post-purchase stages.   SAS is frequently requested by our customers/prospects and challenged by the analyst community to showcase how we help marketers design and manage journeys. This tends to involve:   Audience segmentation Creation, management and planning of the timing/sequencing of a diverse set of channels/touchpoints Journey-based optimization and conversion attribution   Objectively, SAS strives to enable marketers to provide a mutual value exchange on digital channels across the full customer journey by optimizing orchestration capabilities and using customer-directed engagement models to guide personalization strategies. The intent is to provide increasing customer lifetime value (LTV), progressive returns on engagement, more personalized interactions, and sophisticated orchestration across the customer's end-to-end journey. It is well recognized every brand is striving to prove the value of martech in a volatile business environment, connecting marketing and customer outcomes, implemented through increased usage of journey orchestration technologies.    This is why SAS built the Google and Meta Ads connectors for SAS Customer Intelligence 360 users. The vast number of use case opportunities to bring 1st party data, best-in-class AI/ML within customer analytics and the activation layer to optimize customer journeys is here. This is much bigger than just serving an ad on a search engine or social platform driven on 3rd party data targeting.    In SAS Customer Intelligence 360, events enable users to enhance their ability to understand, target, and interact with customers in a meaningful way. Events are used to track user behavior and to provide input conditions for other items such as spots, segments, tasks, data views, and activities. For example, events are used in some of these ways:   Record when customers click a spot on a web page or in a mobile app Observe when products are left  by a customer in a shopping cart Trigger the inclusion of an uniquely behaving customer for a Google or Meta Ads campaign   Tracking user interactions is one of the core features that enables a brand to respond in real time to users based on their profile, origin, browsing behavior, and so on. The interactions that are monitored are the driving force behind many of the features that SAS Customer Intelligence 360 offers. This includes the ability to play a meaningful role in paid media campaigns for:   The inclusion (or exclusion) of customers for paid media campaigns The journey-based personalization and optimization after served impressions and customer media clicks   Image 4: Closed Loop Campaign Management For Customer Journeys      Show Me The Demo   To bring this to life, please view the short introductory demo video. It exemplifies closing the loop on campaign management is more than getting clicks. SAS takes the baton from Google and Meta within a customer journey (websites, apps & other brand-owned properties) using data science muscle (testing, propensity-driven retargeting, algorithmic recommenders, etc.) to improve conversion metrics and monetary-driven objectives     For readers who have a desire to learn more, interactive demo videos for active users of SAS Customer Intelligence 360 are available below:   Google Ads - Task Demo Google Ads - Connector Set Up  Meta (Facebook) Ads - Task Demo Meta (Facebook) Ads - Connector Set Up    Finally, go here to gain incremental awareness about how SAS can be applied for customer analytics, journey personalization, AI decisioning and integrated marketing. ... View more

Brands today are complex ecosystems of decisions that must be executed with increasing levels of automation - due to their competitors digitally transforming and influencing customer expectations. In response, there is a need to change how decisions are made.  Organizations have the opportunity to increase their capability to perform augmented decision making - where a human takes analytically driven insight to make a decision (such as within a call center, website or mobile app). Automation within decision making is when an algorithm (or algorithms) with business rules make the decisions without human intervention (such as next best offers/actions/experiences). With each passing year, the acceleration of the scale, speed and complexity of customer 1:1 decisions is increasing.   Image 1: Complex ecosystems of customer decisioning   Decision intelligence is a practical discipline used to improve decision making. This is performed by explicitly understanding and engineering how decisions are made based on how outcomes are evaluated, managed, and improved via customer feedback. Decision intelligence helps brands reduce technical debt, increase visibility, and improve the impact of business processes. Further, it enhances the sustainability of decision models through relevance and transparency.   Organizations are allocating time and money into DataOps and ModelOps, yet do not receive the full value of these investment. AI and machine learning help brands make better decisions, whether that is deciding to accept a loan application or what pricing discount to offer. DecisionOps help organizations get the most value from their data engineering and data science investments by creating streamlined, efficient processes around model management and decisioning.   Image 2: DecisionOps   A customer journey in its purest form represents a series of brand-orchestrated connected experiences addressing an individual's desires and needs — whether that be completing a desired task or traversing the end-to-end journey from prospect to customer to loyal advocate. When you reflect on this, the customer experience is the totality of cognitive, affective, sensory, and behavioral consumer responses during all stages of the consumption process including pre-purchase, consumption, and post-purchase stages.   Customer decisioning is best used to drive real-time actions in three contexts.   To drive the ideal next journey-based interaction that a customer or prospect should have with your brand. As part of a cross-channel marketing initiative that unifies an experience across customer-facing channels. To enable personalization that delivers customized messages based on an individual's profile and observed behaviors while respecting experiential privacy.     Image 3: Decisioning & offer treatment prioritization   These capabilities come together to enable your brand to create a robust decisioning cycle that deliver analytically driven customer-focused decisions. Users create, manage and deploy decisions via an optimization engine which are made available to a wide variety of endpoints for incorporation into operational systems and processes. A continuous learning cycle ensures the best decisions are being made and governance workflows ensure the orchestrated treatments can be trusted and understood.   Image 4: The decisioning cycle   SAS blends DataOps, ModelOps, DecisionOps & marketing orchestration to support offer treatment prioritization requirements for a wide variety of journey-based use cases. To enable data-driven decisions at scale, the analytics life cycle must be highly operational, automated and streamlined. By connecting all aspects of the analytics life cycle – brands can turn critical questions into trusted decisions.   Image 5: Data-driven decisioning at scale   Chapter 1: Customer Offer Treatment Prioritization   The first demo video below will feature a fictional financial services company comprised of multiple business units (savings, lending, wealth management, etc.) and operating in numerous geographies. The primary objective will be to showcase a mutual value exchange across the full customer journey by optimizing orchestration capabilities and using customer-directed engagement models to guide personalization strategies.   Secondarily, SAS recognizes brands must adapt to a mix of cross-functional and cross-brand goals. The use of machine learning and prescriptive analytics will be shown in support of how marketing teams can generate and prioritize single and cross-brand journeys. Data monitoring, machine learning and AI help surface alerts & address needed optimizations that govern which inbound and outbound interactions a customer should receive in a given time period. The intention is for brands to prove the value of marketing in a volatile business environment, connecting strategies to marketing and customer outcomes.     Chapter 2: Customer Offer Treatment & Risk Assessment   When it comes to use cases, SAS provides brands the ability to make decisions across the entire customer lifecycle and within each discrete customer journey.  The result is a diversity of use case applications across credit services, fraud prevention, claims processing, next best action, and personalized marketing.   Image 6: Customer decisioning scenarios   For our second demonstration example, we will cross-pollinate traditional customer engagement with risk & fraud detection.  This will begin by transparently highlighting how SAS enables decisioning, orchestration and channel delivery services in support of a customer digital loan application experience. It will pivot and conclude on exemplifying erroneous customer behavior that will trigger the risk assessment value proposition.     Keep in mind, when thinking about risk, we are not working on a happy path problem, but trying to protect against exception cases, making security as tolerable as possible for good customers. However, we are not solving a fraud problem; it is really a customer experience problem and the goal is to minimize inconvenience for good customers. Digital identity is less about who we physically are, and much more about what, where, when, and why we do those things.   Image 7: Customer Decisioning, Offer Treatment & Risk Assessment We look forward to what the future brings in our development process – as we enable marketing technology users to access all of the most recent SAS analytical developments. Learn more about how SAS can be applied for customer analytics, decisioning, journey personalization and integrated marketing here.   ... View more

A customer journey in its purest form represents a series of brand-orchestrated connected experiences addressing customer desires and needs — whether that be completing a desired task or traversing the end-to-end journey from prospect to customer to loyal advocate. When you reflect on this, the customer experience is the totality of cognitive, affective, sensory, and behavioral consumer responses during all stages of the consumption process including pre-purchase, consumption, and post-purchase stages.   SAS is frequently requested by our customers/prospects and challenged by the analyst community to showcase how we help marketers design and manage journeys. This tends to involve:   Audience segmentation. Creation, management and planning of the timing/sequencing of a diverse set of channels/touchpoints. Accommodating both new and in-progress campaigns.   Image 1: Personalization Strategies Objectively, SAS strives to enable marketers to provide a mutual value exchange on digital channels across the full customer journey by optimizing orchestration capabilities and using customer-directed engagement models to guide personalization strategies. The intent is to provide increasing customer lifetime value (LTV), progressive returns on engagement, more personalized interactions, and sophisticated orchestration across the customer's end-to-end journey. It is well recognized every brand is striving to prove the value of martech in a volatile business environment, connecting marketing and customer outcomes, implemented through increased usage of journey orchestration technologies.    With that said, let's keep the reading to a minimum, and introduce the demo video below. Here is a quick preview:   SAS Customer Intelligence 360 will be used to demonstrate customer journey design capabilities, showcasing support for multiple customer-directed engagement strategies. The exemplified journey incorporates a variety of channels/touchpoints, including email, web, mobile, social and external CRM systems. Key features which will be showcased include libraries of trigger-based journeys, node controls, content and offers, testing/experimentation, and journey versioning.     We look forward to what the future brings in our development process – as we enable marketing technology users to access all of the most recent SAS customer data management and analytical developments. Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here. ... View more

As marketers and advertisers utilize artificial intelligence (AI) and machine learning (ML) to elevate their brands, understanding the potential for detecting and mitigating bias within predictive analytic procedures is crucial.  Today, AI plays a progressive role in advancing the marketing and advertising mandate. Data is generated constantly in the digital ecosystem, associated with real-time customer engagement behavior, web/mobile interactions and incremental streams of desired revenue year-after-year. This gives brands the chance to apply a variety of analytical techniques to draw "ah-ha" insights that can be used in segmentation, targeting, recommendations, next-best-experiences and other forms of customer interactions.   At a fundamental level, AI/ML helps brands improve their viewpoints on customers and subsequently, running their business. When using data science for these intentions, brands are placing immense trust that their modeling IP is providing useful insights that subsequently influence marketing treatments. Keep in mind, not every single human being on planet Earth has the same level of data/analytical literacy, and some may consider applications of AI/ML as opportunistic innovation, free from the mistakes commonly attributed in human-driven decisions.   Image 1 - Fact or Fake   I hate to spoil the fun, but we need to be more prudent as practitioners. AI/ML can contain bias. Instead of ushering in a utopian era of fair decisions, AI/ML have the potential to exacerbate the impact of biases. As innovations help with everything from the identification of attractive prospects to predicting who should receive a marketing stimuli, it is important to understand every modeling application has the potential to affect separate segments of a customer population differently. When applied in martech, biased AI/ML can negate efforts to learn, understand and anticipate consumer behavior. Brands should improve their understanding of how AI bias impacts them, how to detect it and ultimately, mitigate.   I'm a Marketer. Not a Data Scientist. Why Should I Care?   There are numerous ways in which bias can slip into customer data. Although marketers themselves may not build analytical models, it's hard to find a use case these days that doesn't benefit from propensities or probabilities. There are many perspectives to take into account when describing bias in data science. Bias can happen during data collection, data processing, sampling, model building, and so on. However, when AI/ML is applied to data that is inaccurate, it has the potential to magnify the errors and cause unintended bias in campaigns, personalization or testing. If that wasn't enough, bias can lead to irrelevant results from severely impacted KPIs, such as failing to reach the correct audience or serving up the wrong offers to a particular demographic. Ultimately, this means wasted money and resources, failure to reach relevant customers, and potential harming the reputation of a brand.   Perhaps I have your attention now? While most brands readily promote fairness in AI/ML as a principle, putting the processes in place to execute it consistently is an ongoing obstacle. There are multiple dimensions for evaluating the fairness of AI/ML, and determining the correct approach depends on the use case.   As a practitioner in the application of data-driven methods in martech, it's astounding that in 2023, numerous brands are acknowledging a trust problem with AI/ML. As a result, brands will lose the benefit from the innovation and derived insights of advanced analytical methods to make better decisions.  In short, I see two parallels at the moment:   Data scientists and analysts need to continue their focus on translating the output of AI/ML in business language and storytelling to reduce stakeholder intimidation. Remember, if your models are not put into action, what was the point of your effort?  Marketing and CX decision makers may not be passionate about statistics, but nearly every use case can elevate through the usage of propensities and probabilities. While AI/ML is advertised as next-level precision, it is not 100% perfect. Therefore, the translation of propensities and probabilities into business context must be interrogated, transparent and understood.   Detecting Bias for Fair AI/ML   In predictive modeling, bias occurs when a model’s prediction or performance differs for unique values of a given variable, referred to here as the "sensitive" variable.  Prediction bias is measured by calculating the difference in average model prediction for values of the sensitive variable. Performance bias is measured by calculating the difference in model performance for values of the sensitive variable.   SAS provides users the assessBias action which calculates performance statistics and average model predictions for each level of a nominal (categorical) variable. The difference in these statistics is then calculated either by reporting the maximum pairwise difference among all levels or by comparing each level to a reference group in order to report a single number that represents model bias.   Turning this feature on is straightforward. As an analyst prepares to build a model in SAS, they simply need to indicate which variable(s) they would like assessed for bias. This can be done in a single-click.   Image 2 - Detect Bias Feature in SAS   For the purposes of this example, we’ll be creating a simple pipeline with two Gradient Boosting models. The model node on the left will be used for assessing bias detection.   Image 3 - Model Pipelining Example - Assessing for Bias   Once the pipeline has been trained, the auto-generated Bias Assessment report can be viewed by opening the model’s results and selecting “Fairness and Bias."    Image 4 - Fairness & Bias Report   There are multiple plots in Image 4, and we’ll review each of them one at a time. Before we interpret, let's summarize this modeled data. It represents a financial services brand concerned with extending loans to it's customers, and potential default. The focus of bias assessment will be centered on a variable entitled Reason, which indicates why the customer applied for a loan. The values for the variable are Debt Consolidation, Home Improvement or Missing.    Image 5 - Prediction Bias Assessment   Prediction bias represents how much greater the model's probability to predict the event is for one group over another on average. In this modeling project, the target event level is “1” indicating when a customer defaulted on their loan. The bars in this plot represent the target event's average predicted probability for each level of the variable REASON. Large differences in bar size indicate that the model predicts the event at considerably different rates for unique levels of REASON, and analysts should be aware of this before using the model.    This prediction bias plot in Image 5 shows that the loans with a REASON code set to Home Improvement are higher than that of Debt Consolidation, meaning that the model predicts a higher default probability for loans related to Home Improvement.    Image 6 - Prediction Bias Parity   The bar in the plot of Image 6 represents the maximum pairwise difference in the target event's average predicted probability between levels of the REASON variable. It shows that the maximum prediction difference is 7.6% which occurs between the levels of Home Improvement and Missing.    Image 7 - Performance Bias Plot   Performance bias is a representation of how accurate the model is for one group over another. The bars in this plot, displayed in Figure 7, represent the values of performance (or accuracy) metrics for each level of the sensitive variable REASON. The performance metrics included here are Multi-Class Log Loss (MCLL), True Positive Rate (TPR), and Best Kolmogorov-Smirnov Along ROC (maxKS). Higher values of TPR and maxKS, and lower values of MCLL indicate a better fit. Large differences in bar size for any specific metric indicates that the model performance (accuracy) is not consistent across all levels of the variable REASON, and analysts should be aware of this before using the model.    Image 8 - Performance Bias Parity   The bars in this plot of Image 8 represent the maximum pairwise difference in performance metrics between levels of the REASON variable. For example, the bar for maxKS shows that the greatest difference in maxKS is 11.3%, which occurs between the levels Missing and Debt Consolidation. The maximum pairwise difference in True Positive Rate (TPR) is also known as "Equal Opportunity".   Mitigating Bias for Fair AI/ML   Now that we have taken an introductory tour of assessing bias for sensitive variables, let's pivot and discuss mitigation.  With the widespread use of AI/ML models in decision-making processes, there is increasing interest about the impact of these models. Bias can be embedded in both the training data and the trained model, either of which can lead to unfair or biased predictions for certain groups or populations. To reduce the bias in the data and models, there are three  categories of bias mitigation methods that can be considered.   Preprocess methods, which transform the data prior to model training. These methods try to eliminate the correlations between the input features and the sensitive feature while retaining as much information from the input data as possible. In-process methods, which consider fairness constraints during the model training process. They actively adjust the model parameters during training in order to create a model that produces fair predictions and classifications. Postprocess methods, which do not directly alter the input data or the model training process. They adjust the predictive outputs from the learned models to compensate for the bias in the model.   SAS provides users the mitigateBias action which implements the exponentiated gradient reduction (EGR) algorithm, which uses an in-process method of mitigating bias. It considers fairness constraints during the model training process. The EGR algorithm has three main advantages.   First, it is model-agnostic, which means that analysts can apply it to a wide variety of ML models. The EGR algorithm works like a wrapper that is built around existing ML models. The wrapper iteratively reweights the data, and then it trains a new classifier on the basis of reweighted data in each iteration. By doing so, it reduces the complex optimization problem of training a ML model with fairness constraints to a series of standard model training problems without the fairness constraints. Second, the EGR algorithm supports various fairness measurements, such as demographic parity, equal opportunity, and equalized odds. Finally, the EGR algorithm has strong theoretical support in terms of its optimization trade-off between fairness and accuracy. In theory, the EGR algorithm returns solutions that satisfy all the fairness constraints and also achieve the optimal error rate of the best fair solution.   Training a fair ML model is in general a nonlinear programming problem. In nonlinear optimization tasks, it isn't uncommon to get suboptimal solutions because the constraints are infeasible or because the algorithm is stuck in a local minimum. As with many other ML algorithms, the effectiveness of the EGR algorithm depends on many factors, such as the quality of the input data, the selection of fairness constraints, the quality of the ML models that it wraps around, and the values of the hyperparameters.   The mitigateBias action supports multilevel nominal target variables and multilevel nominal sensitive variables. Note that when analysts include a large number of levels in the sensitive variable, the EGR algorithm searches more complex optimization scenarios, which can result in suboptimal solutions. Therefore, the solutions from the EGR algorithm might not always be desirable when the number of levels in the sensitive variable is high. You should review and validate your sensitive variable, training models, and fairness constraints before applying the EGR algorithm.   In Image 5 above, we observed the prediction bias plot showing that the loans with a REASON code set to Home Improvement are higher than that of Debt Consolidation. Let's apply the mitigateBias action and review the subsequent results.   Image 9 - Model Pipelining Example - Mitigating for Bias   We are simply shifting our focus from the left modeling node to the right modeling node (which was run in parallel) to obtain comparative results. Just like we did in the previous example, once the pipeline has been trained, the auto-generated Bias Assessment report can be used to review the impact of the mitigateBias action through viewing the model’s results and selecting “Fairness and Bias."    Image 10 - Prediction Bias Assessment After Mitigation   Readers will now observe that Image 10 shows a prediction bias plot loans with a REASON code of Home Improvement or Debt Consolidation are producing similar results. Assuming as analysts we are satisfied with the results, users can pivot to topics like model interpretability reviewing PD, ICE, LIME Explanation and HyperSHAP Value plots.   Image 11 - Model Interpretability   ...as well as general model assessment for Lift, Gain, ROC, F1, Accuracy and other essential diagnostics.   Image 12 - Model Assessment   Effective risk management is increasingly being brought to the frontline rather than functioning in the back office. When using advanced analytics, it’s becoming increasingly important to understand and measure fairness risk to avoid exploiting vulnerable customers.   For those who prefer to see live demos, check this out:     Although our demo story in this article focused on one example of a sensitive variable, consider other candidate variables that a brand could (and should) assess. Creating frameworks and processes to mitigate bias and address fairness risk will mean that it can be expanded to other models with rigor in the future.   As stated earlier, if your models are not put into action across the touchpoints that are relevant to your brand's strategy in orchestrating customer journeys...   Image 13 - SAS Customer Intelligence 360 - Multichannel Orchestration   What was the point of your effort? Don't let misaligned perceptions and distrust of business stakeholders hold back the innovation that AI/ML can bring forth within martech.    We look forward to what the future brings in our development process – as we enable technology users to access all of the most recent SAS analytical developments. Learn more about how SAS can be applied for customer analytics, journey personalization and integrated marketing here.                 ... View more

Within the martech industry, there are several factors that contribute to the challenges surrounding brand decision-making. Obviously, customers and markets are more competitive and demanding. When you step back and reflect on this, it's a linear trend upward year-after-year when it comes to consumer expectations. This means, to satisfy that demand, it's well recognized that brands need to respond quicker, but it's often overlooked that accuracy holds an equal weight. Personalization, targeting, segmentation, relevance and other fun martech buzzwords all rely on this.   Data continues to flood every organization, both in size and in speed. Sometimes more data is better, but the challenge can be that critical decision-making information gets lost. Skilled analytical talent with application experience in the various domains of modern marketing is the key to move a brand from reactive to proactive. Thus, varying flavors of technology and automation are critically important to augment customer analysts in accelerating their delivery's time-to-value.    Machine learning is a branch of artificial intelligence that automates the building of systems that learn iteratively from data, identify patterns, and predict future results. And it does that with minimal human intervention. Machine learning shares many approaches with other related fields, but it focuses on predictive accuracy. Building representative machine learning models that generalize well on future data requires careful consideration of both the data at hand and assumptions about the various available training algorithms.   Supervised Learning   Supervised learning algorithms are trained using labeled examples (conversion vs. non-conversion), such as an input where the desired output is known. The learning algorithm receives a set of inputs along with the corresponding correct outputs, and the algorithm learns by comparing its actual output with correct outputs to find errors. It then modifies the model accordingly. Supervised learning is commonly used in applications where historical data predict likely future events. For example, supervised learning can anticipate when an insurance customer is likely to file a claim, or when a retail customer has a higher likelihood to be interested in an upsell recommendation.   Image 1 - Supervised Learning   SAS supports two types of supervised learning problems through natively-supported algorithms such as gradient boosting, forests, neural networks, support vector machines, Bayesian networks and more.   Classification – When the data are being used to predict a categorical target, supervised learning is called classification. This is the case when assigning a label or indicator (for example, labeling an image a dog or a cat). When there are only two labels, this is called binary classification. When there are more than two categories, the problems are called nominal classification. Regression – When the data are being used to predict interval targets, the problems are referred to as regression.   The reason supervised learning as a category contains a variety of algorithms is based on the notion that no model is uniformly the best, particularly when considering the deployment over time, when data changes. Analysts select a model primarily based on fit statistics and assessment graphics of performance.   Fit statistics transform model performance to numerical scores for easy comparison. Assessment graphics provide a global view of model performance. They facilitate model comparisons across a variety of deployment scenarios.   Image 2 - Comparing Algorithmic Models   SAS enables users to facilitate selection between models using fit statistics and assessment graphics. The purpose of predictive modeling is generalization, which is the performance of the model on new data (not used during the training process). To compare across several modes, SAS computes all assessment measures for each available data partition (train, validate, and test).    Image 3 - SAS for Model Pipelines, Assessment & Comparison   Classifier Performance   Supervised classification does not usually end with an estimate of the posterior probability. For example, in binary classification problems, the ultimate use of a predictive model is to allocate cases (customers) to classes (target / don't target). This is accomplished by appropriately choosing a posterior probability cutoff. The cutoff or threshold represents the probability that the prediction is true.   Image 4 - Probability Cutoff Challenge   Analysts often might want to choose probability that gives the maximum accuracy. However, care should be taken when analysts have a case where the response column is skewed. This is EXTREMELY common in martech and customer analytical use cases.   Have you ever heard of a brand with a 50% conversion rate?   Unless you're reading a fictional story, this never happens. For example, a bank wants to predict the loan defaulters, so model performance needs to be assessed considering the posterior probability cutoff. An allocation rule is merely an assignment of a cutoff probability, where cases above the cutoff are allocated to class 1 (when we predict a customer will convert) and cases below the cutoff are allocated to class 0 (when we predict a customer to not convert). For example, the standard logistic regression model separates the classes by a linear surface (hyperplane), shown on the left of the image below.   Image 5 - Classification Cutoff   The decision boundary is linear, and determining the best cutoff is a fundamental concern. An allocation rule corresponds to a threshold value (cutoff) of the posterior probability that affects the confusion matrix. For example, all cases with probabilities of default greater than 0.04 might be rejected for a loan. For a given cutoff, how well does the classifier perform? This is the question. The fundamental assessment tool is the confusion matrix. Don't you love the name? The confusion matrix is a crosstabulation of the actual and predicted classes. The confusion matrix contains true positives (events that are correctly classified/predicted), false positives (non-events that are classified/predicted as events), false negatives (events that are classified/predicted as non-events), and true negatives (non-events that are correctly classified/predicted as non-events). It quantifies the confusion of the classifier. Having fun yet?   The event of interest, whether it is unfavorable (like fraud, churn, or default) or favorable (like response, click, or purchase to an offer), is often called a positive, although this convention is arbitrary. Here are the simplest performance statistics:   Accuracy = (true positives and true negatives) / (total cases) Misclassification rate = (false positives and false negatives) / (total cases)   Image 6 - Sensitivity and Positive Predicted Value   Image 6 highlights two specialized measures of classifier performance focused on true positives. Large sensitivities do not necessarily correspond to large values of positive predicted value (PV+). Ideally, analysts would like large values of all these statistics. Sensitivity is also known as recall or true positive rate (TPR). Positive predicted value is also known as precision. Precision and recall are popular in applications such as information retrieval and anomaly detection.   Image 7 - Specificity and Negative Predicted Value   Image 7 focuses on the analogs to these measures referred to as true negatives. Specificity is also commonly known as true negative rate (TNR).   The context of the classifier performance problem determines which of these measures is the primary concern. For example, a marketer is likely most concerned with PV+ because it relates to targeted offers (guided by predictions) and the associated response or conversion rate impacting KPIs they measure against. There is a cost to marketing and efficiency matters.   Image 8 - The Use Case Matters   Analysts must balance their desire for true positive rate (and false positive rate). There is always a trade-off between them. If you want to increase TPR, your FPR will also increase. For example, if you want to increase positive (conversion) outcomes (higher TPR), you must also be willing to incur incremental error in terms of predicting non-conversions (higher FPR). Customer behavior can never be modeled perfectly (human beings can act irrationally or unexpectedly from time-to-time). This is the very core of the challenge in deciding the probability cutoff.   Image 9 - The Interactive Cutoff Plot in SAS   The Cutoff plot is an auto-generated interactive visualization in SAS that shows how different model assessment statistics change as the prediction cutoff value changes. The model assessment statistics are based on the selected event (conversions, responses, clicks, etc.) for the model compared to non-events (non-conversions, non-responses, no clicks, etc.). Analysts can interactively move the cutoff line to represent different prediction cutoff values. As users move the cutoff line, the model assessment statistics are updated. This allows analysts to to choose a cutoff that best represents, their particular problem (maximizing marketing offer/tactic efficiency) and business objective.    Image 10 - The Event Classification Graph in SAS   The Event Classification graph is another auto-generated visualization in SAS that displays the confusion matrix at various cutoff values for each partition. Recall, the confusion matrix contains true positives (events that are correctly classified as events), false positives (non-events that are classified as events), false negatives (events that are classified as non-events), and true negatives (non-events that are correctly classified as non-events). Each of these segments are displayed in blue or yellow within the corresponding bar associated with the model's classification event level (such as conversion/non-conversion).   Image 11 - The Confusion Matrix in SAS   Different cutoffs produce different allocations and different confusion matrices. To determine the optimal cutoff, a performance criterion needs to be defined. If the goal were to increase the sensitivity of the classifier, then the optimal classifier would allocate all cases to class 1. If the goal were to increase specificity, then the optimal classifier would be to allocate all cases to class 0. For realistic data, there is a trade-off between sensitivity and specificity. Higher cutoffs decrease sensitivity and increase specificity. Lower cutoffs decrease specificity and increase sensitivity.   Users in SAS have access to a binary classification cutoff feature that specifies the cutoff for determining the predicted value for a binary target based on the posterior probabilities. If this feature is not customized by an analyst, the default value is 0.5. This is the case not only in SAS, but any software package that enables users to build supervised learning models. When applying predictive models for marketing use cases, it is essential to consider posterior probabilities prior to finalizing a modeling exercise. Remember, when have you ever heard of a brand that has a 50% conversion rate for anything they offer to customers? Okay then, let's move on.   The Profit Matrix   Determining an appropriate cutoff is problem specific, and there are many ways of accomplishing this (Bayes' Rule, Central Cutoff, KS Cutoff, etc.). We will focus on one solution referred to as the Profit Matrix, which is a formal approach to determining the optimal cutoff using statistical decision theory (McLachlan 1992, Ripley 1996, Hand 1997). The decision-theoretic approach starts by assigning profit margins to true positives and loss margins to false positives. The optimal decision rule maximizes the total expected profit.   Image 12 - Profit Matrix Example   The profit matrix in Image 12 is meant to portray a simple example. It is based on a marketing effort that costs $1 for every impression (choose your favorite channel/touchpoint)and that, when successful (targeted customer conversion), garners revenue of $100. Hence, the profit (or loss) for targeting a non-responder is -$1, and the profit for targeting a responder is $100 - $1 = $99. Given that everyone in this population has a posterior probability, simple algebra can be used to find the optimum cutoff.   Here is a typical decision rule. Target a customer if the expected profit for making an offer, given the posterior probability, is higher than the expected profit for ignoring the customer. The optimized cutoff can be identified by calculating the expected profit. Goodbye confusion matrix, hello profit matrix!   When the desired target event is rare, which is common in martech and customer journeys, the cost of a false negative is usually greater than the cost of a false positive. In other words, the monetary cost (or missed opportunity) of not targeting a customer who would have resulted in a conversion is greater than the cost of offering a promotion to someone who does not convert.  Such considerations dictate cutoff rates that are less (often much less) than the default 0.5 value set in modeling software.   Image 13 - Envisioning a Profit Matrix for Subscription Business Model   To determine reasonable values for profit and loss information, consider the outcomes and the actions that your subscription-oriented brand would take given knowledge of these outcomes. In Image 13, there are two outcomes (churn and active) and two corresponding actions (offer discount and no action). Knowing that someone is a churner, analysts would naturally want to offer a discount to that person in hopes of preventing them from de-subscribing. Knowing that someone is a non-churner, you would naturally want to not offer any discount to that person. On the other hand, knowledge of an individual’s actual behavior is rarely perfect, so mistakes are made. For example, offering discount to non-churners (false positives) and not taking any action for churners (false negatives). Taken together, there are four outcome-and-action combinations shown. Each of these outcome-and-action combinations has a profit consequence (positive and negative).   Suppose from the description of the analysis problem that the variable AVG_ARPU_3M gives the customer's average revenue for the past three months. Also, there is a 15% decline in the average revenue of that customer when a discount is offered to retain them. From a statistical perspective, AVG_ARPU_3M is a random variable. Individuals who are identical on every input measurement might be associated with varying revenue amounts. To simplify the analysis, a summary statistic for AVG_ARPU_3M is plugged into the profit consequence matrix.   Image 14 - Profit Matrix - Outcome and Action Combo One - Successful Marketing Intervention   This is when the brand gives a 15% discount to customers that were predicted as churn, and in response, they no longer churn. The brand earn 3 months of average revenue minus the 15% discount. In other words, $60.30 – $9.04. The total profit is equal to $51.26.   Image 15 - Profit Matrix - Outcome and Action Combo Two - Unnecessary Marketing Intervention Wasting Budget   This is when the brand predicts that a segment of customers will churn but they actually stay indicating a marketing intervention was not necessary. The brand provided a discount to them, so it experiences a negative consequence monetarily. The amount lost is $9.04 per subscriber.   Image 16 - Profit Matrix - Outcome and Action Combo Three - Incorrect Churn Prediction and No Marketing Intervention   This is when the brand predicts that a segment of customers will not churn but they actually do indicating a marketing intervention could have mitigated this behavior. The brand did provide a discount, so it experiences a larger negative consequence monetarily. The amount lost is $60.30 per churned subscriber.   Image 17 - Profit Matrix - Outcome and Action Combo Four - Successful Non-Marketing Intervention   This is the rate when the brand correctly predicts that customers will not churn, so no discounts are given to them. The brand will earn profits as usual. If the customer does not churn, it has no effect on the model's influence on decisioning. So the value can immediately be set to 0.   Image 18 - Completed Profit Consequence Matrix   With the completed profit consequence matrix, analysts can calculate the expected profit associated with each decision. This is equal to the sum of the outcome and action profits multiplied by the outcome probabilities. The best decision for a case is the one that maximizes the expected profit for that observation. When the elements of the profit consequence matrix are constants, prediction decisions depend solely on the estimated probability of response and a constant decision threshold.   SAS enables users to leverage the HPDECIDE procedure which creates optimal decisions that are based on a decision matrix, on prior probabilities, and on output from a modeling project.    Image 19 - SAS HPDECIDE Procedure   Each model an analyst runs in a project can make a decision for each customer observation in a scoring data set, based on numerical consequences specified via a decision matrix and cost variables (or cost constants). The decision matrix can specify profit, loss, or revenue. The HPDECIDE procedure chooses the optimal decision for each observation, such as maximum expected/estimated profit or minimum expected/estimated loss. For the demonstration example in Image 20, the average revenue for the past three months minus the 15% discount cost is used as the (constant) profit associated with the churn outcome and the offer discount decision.    Image 20 - Using SAS HPDECIDE Procedure in Model Pipelining   Let's review the results of the Profit Matrix node.   Image 21 - SAS HPDECIDE Procedure Scored Customer Data   The table shows the scored data set, which displays the decision for each observation. • D_DecisionData is the label of the decision chosen by the model. In other words, take action or don't take action. • EP_ DecisionData is the expected profit for the decision chosen by the model. • CP_ DecisionData is the profit computed from the target value. The value 0 signifies no change in the usual profit.   For example, the predicted probability of churn for the first observation is 0.18162 (or approximately 18%). Expected profits for decisions that offer a discount and no action are $1.90 and $-10.95, respectively. Because the first value is larger, the decision is to offer a discount, reflected in the D_ column, and the expected profit for this observation would be $1.90, reflected in the EP_ column.   Further, here comes the the statement every executive leader wants to hear during the analyst presentation.    Image 22 - SAS HPDECIDE Procedure - Profit Summary   Average profit can be used to summarize the model's overall performance. For the profit matrix used in this example, average profit is computed by multiplying the number of cases by the corresponding profit in each outcome-and-decision combination, adding across all outcome-and-decision combinations, and dividing by the total number of cases in the assessment data. This example shows that the total profit is $23,432.33 and the average profit is $0.41431, based on the decisions scored from the HPDECIDE procedure.   As an analyst, when you can communicate the value of your modeling efforts in monetary terms, every executive paying attention is going to lean in and focus. Passing these insights to influence our marketing teammates will directly impact their segmentation strategies and touchpoint tactics.   Image 23 - Profit Matrix, Segments & SAS Customer Intelligence 360   Oh yeah, let's not forget everyone needs that easy-to-understand report too!   Image 24 - Profit Matrix Optimization - Summary Report   As a life-long student of business and marketing analytics for the last two decades, this concept of a profit matrix is one of the most industry-practical topics I have ever learned. I hope readers will consider learning more on how to use this amazing spell of data magic! For those who prefer to see live demos, check this out:     For readers who have a desire for more, go here to gain incremental awareness about how SAS can be applied for customer analytics, journey personalization and integrated marketing here.       ... View more