Introduction to SAS 360 Marketing AI (Part 3: Configuring Recipes)
Over the past few months, we have been rolling out exciting updates for SAS 360 Marketing AI. Previous thought leadership articles (Part 1 and Part 2) summarized SAS development efforts to release a solution-oriented software application offering prescriptive use case or recipe-oriented experiences to address trending use cases for B2C (and B2B) brands. For readers unfamiliar with the term "recipe", the concept lives at the center of SAS 360 Marketing AI's design principles, which 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.
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 50 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?
Before proceeding, we need to address the market perception (or possible misperception) on the innovation and fireworks currently going off related to Generative AI (GenAI), which is a branch of machine learning. Like all machine learning models, GenAI systems are trained on large datasets to recognize patterns — in this case, patterns in language, images and behavior that allow them to produce new content rather than simply classify or predict. In marketing, these models apply learned patterns to specific tasks: drafting email subject lines from historical campaign performance, generating product descriptions from catalog data or summarizing customer feedback into actionable themes.
A modern marketing stack typically relies on two distinct machine learning capabilities working in tandem. Predictive and customer analytical models analyze data to guide targeting, segmentation, timing and optimization. GenAI models serve as the creative engine, producing assets such as ad copy, visuals, summaries and content variations. Both are forms of machine learning, but they play fundamentally different roles.
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 absorbing innovation from all branches of data science.
Our 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. 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. In other words, SAS is introducing AI and advanced analytic capabilities FOR marketing use cases acutely. For a moment, reflect on the idea of a software application that is:
- Designed for the domain space and themes of martech.
- Focuses on use cases (i.e. recipes) while minimizing adoption friction related to statistical jargon frequently misunderstood by anyone outside of the data science profession.
- Uses the best of both worlds - GenAI blended with best-practice machine learning, predictive & segmentation capabilities in a no-code rapid-scoring mechanism that seamlessly integrates with the broader SAS Customer Intelligence 360 solution, or external 3rd party martech tools.
As we continue to partner, guide and help find incremental value with our customer partners, SAS realizes the time in NOW to release a solution that bridges all of the power and innovation of prediction, machine learning, and GenAI together. Let's transition to outline how SAS 360 Marketing AI will benefit and be used by team members who have experience working with data (i.e. data engineers, data scientists, data analysts, IT, etc.) to craft recipes to simplify the project user's (or marketer's) experience to generate actionable customer scores for timely opportunities.
SAS 360 Marketing AI: The User Experience of Configuring Recipes
Configuring recipes are the starting point to addressing any use customer use case that would benefit from the presence of AI-generated scores being available. Once a recipe is configured, it is meant to unleash the opportunity for marketing teams to take action and activate on contextual, data-driven customer scores. Our philosophy is for SAS to reduce the complexity and automate as much of the analytical marketing process as possible for users. For this section, we will utilize both screenshots and a demo video at the end to outline the user's process of configuring recipes.
The first screen we will observe is the starting point which enables users to select a recipe template. It summarizes the template options available, and users simply need select one to begin.
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. A key area SAS developed in this software for users was to remove as much friction as possible to contextualize templates and simplify data engineering steps as they are applied to a brand's use case. For example, if users do not know the meaning of one of the templates, the software's co-pilot provides the option to ask questions or interact with an agent for assistance.
How do use cases and requirement meetings begin for leveraging recipes? 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 propensity to purchase.
Recipes can be given custom names, codes, descriptions, business context and specificity on customer events of interest. For example, in this propensity to purchase customer use case, users can specify whether the analytical scoring should be derived on first-time purchases, or any purchase behaviors.
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 a recipe's specificity features within the software. Moving on, 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 data scientists & analysts 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. So, we set out to challenge ourselves to redesign and remove challenges related to data access and sourcing. Let's highlight that working with data from multiple origins can be improved and made easier.
The screenshot above exemplifies the following options:
- Does the user require one or more data sources?
- Does the user need to apply time-series data for the use case to be successful?
- Does the user desire to refine and/or filter any columns or rows before proceeding?
Take note, both natural language generated explainers & co-pilot support users to ensure successful usage of the software. Let's explore this further.
In the example above, the user is presented multiple data source options. No APIs or other forms of high-code obstacles will get in your way. Once the software is set up, users benefit from the software's out-of-the-box connectors to various cloud-based and/or on-prem data stores.
Once the user has made their selections, the screenshot above represents the output. It's a breeze adjusting time indexes and refining what data is required. Let's highlight an example of adjusting what data users prefer to work with (and not).
Refining data involves the decision to use (or not to use) specific data attributes in a recipe configuration. In addition, users can filter observations to determine which records to include in the recipe.
The data sources that users select provide the inputs for purchase prediction within this recipe. As a best practice, users should choose sources that can be joined and are relevant to a brand's strategic objective. The next step in the user's workflow takes us there.
Data table joins are one of many examples in the DataOps phase for users to begin the process of acceleration through improving data quality, accelerating time-to-value, and fostering collaboration between data engineers, data scientists, and business/marketing stakeholders.
Moving forward, 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.
In essence, users can select which aggregations, transformations, and/or calculated columns they desire for the recipe. Alternatively, the software will provide the user smart default suggestions & AI-driven explanations to assist in this step of the workflow. To be clear, selecting transformations and aggregations from raw data improves the performance of machine learning models. We (SAS) choose the most often used aggregations and transformations by default so that users do not have to. For more control, users can customize aggregations, transformations, and create your own columns.
This screenshot above is an example of 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.
Another topic for those who would configure recipes relates to population filters. For this acquisition marketing use case example, users may desire to specify the subsets of customers to predict likelihood to purchase for. Users can leverage population filters to get the most relevant data for their use case. For example, to predict which consumers that previously bought from the apparel category are going to churn, users can limit their data to include only customers who previously bought apparel. In addition, a recipe can support different use cases when users define multiple filters. These filters, along with the purchase/churn definitions and time frames, are used as the basis for the tables that become the inputs into an analytical model. These selections impact what options a project user (or marketer) has when a recipe is published and available for usage.
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.
In an acquisition marketing use case, every brand will need to define or specify how to define likelihood to purchase. In other words, users define the customer purchase event definition to exactly match how the brand operates by adding criteria. The purchase definition serves as the target variable for the analytical model.
- The definition can contain one or more events. Multiple events can be evaluated based on operators (either separately or in groups).
- Each event contains one or more sets of criteria that defines likelihood to purchase.
Image 16: Define business objectives for conversion events
Image 17: Timeframes
Image 18: Project table generation
Image 19: Customizing data display names for downstream simplification
Image 20: Adding columns for project users to select from
image 21: Attribute options for personalization & segmenting customer journeysNow, there's one more thing to touch on, and it's extremely vital from our perspective that the themes of AI governance, transparency and explainability blossom for users within SAS 360 Marketing AI. Project(s) can be activated with the support of monitoring vital metrics to ensure the analytical solution is fresh (not stale), and predicting with optimal accuracy.
The reasons SAS built this functionality into a Marketing AI solution's user workflow 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 scores needs to be revisited/refreshed, as well as when fairness becomes an issue and should be considered for mitigation.
Image 22: Fairness metrics for bias detection & mitigation of sensitive dataAgain, we at SAS challenged ourselves on reimagining how to simplify how the user sets up this automated monitoring. For those unfamiliar with this family of metrics, we have defined them below.
- Performance metrics
- Accuracy: In essence, it represents how often our predictions are right. In other words, it monitors how often we correctly predict both event levels of a classification use case, such as why customers choose to buy vs. not buy.
- F1 Score: Monitors the balance between accurately identifying customers who purchase (or churn) while avoiding unnecessary marketing tactical actions. This helps brands limit wasting resources while keeping marketing strategies effective.
- Precision: Monitor how often the model correctly identifies your desired event and/or objective. For example, in a churn use case, this metric tracks how many customers that we predicted to take on this behavior, actually churn. The intention is to assist brands in saving time and resources by ensuring that your retention actions are targeted at actual churn risks.
- Recall: Ensures your brand captures as many acquisitions/upsells/churners as possible, even if some were not going to follow that desired behavior. This is sometimes described as "catching as many fish as possible", while raising our tolerance for a higher threshold of errors.
- Overfit: Compares the model's performance on new data with its performance on data that it learned from. This ensures that the model performs consistently across new and trained data so that predictions remain reliable.
- Fairness metrics
- Performance bias: Measures bias in performance among different groups (or segments). A higher value indicates stronger bias. If a fairness attribute has multiple groups (like small, medium, large or bronze, silver, gold), we measure bias as the largest difference in performance between all groups.
- Predictive Bias: Represents how much greater the model's probability to predict the event is for one group over another on average. A higher value indicates stronger bias between groups. In cases where a fairness attribute has multiple groups, we measure bias as the largest difference in average predictions between all groups.
Monitoring fairness prevents biased targeting of specific groups. This helps ensure that your brand's marketing efforts are trustworthy and effective for everyone.
Make Data Engineering Go Faster & Remove Complexity
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. SAS set out to examine and improve how brands within the domain of data-driven marketing will enable a data-savvy individual in crafting recipes for their marketing counter-parts to leverage. In short, that is the role of the recipe configuration user.
When a recipe is completed and published, project users and marketers simply migrate to the Projects section of SAS 360 Marketing AI to take the baton and get to work.
For more details on the project user experience, please go here.
The Middle Is Where Good Gets Great
The hardest part of marketing isn’t getting started. It isn’t crossing the finish line, either. It’s the messy middle – that critical space between ideas and outcomes. The place where data meets decisions, where strategies either connect or collapse. The middle is where audiences come together. Where channels align. Where stories stop being plans … and start becoming experiences. But the truth is, the middle doesn’t always get the attention it deserves.
It’s where good intentions get stuck. Where disconnected tools slow the momentum. Where silos, spreadsheets and scattered signals make the work harder than it has to be. The middle is not where things should fall apart. It’s where everything should come together.
That’s the focus of SAS Customer Intelligence 360 – built to power the middle. Designed to connect data, decisions and delivery. Engineered for marketers who are ready to move from messy to magic. Because the middle matters. And when the middle works, marketing works. SAS Customer Intelligence 360 – where the middle makes the difference.
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.
