Analytics maturity, data silos, the skills gap and why AI really isn’t the issue
What is the current priority among CIOs and other information professionals? One way to find out is through studies such as our recent work on the use of generative AI. However, another way is to talk to partners such as Performance Technologies S.A., a SAS partner based in Greece. Performance Technologies has four business arms, including one to support the move to cloud as well as a data and artificial intelligence (AI) unit, which focuses on data-driven decision-making. Many of its customers are in the process of migrating to the cloud.
Building data transformation
Stavros Stavrinoudakis is the Director of the Analytics Business Unit at Performance. He describes these cloud migration projects as ‘data platform transformation projects’ because of the focus on removing data silos. He adds that, interestingly, CIOs are driving this change. As the use of data and AI evolves and matures within companies, we often see changes in who is controlling or using these technologies, from a central small unit to data and AI being embedded within the business. Stavros suggests that data and AI platforms are essentially an IT project, and therefore owned by the CIO rather than any other C-level executive. He believes that this is probably a practical decision to avoid creating new data silos, but also notes that there may be other issues.
“For the data platform per se, IT will be the key stakeholder. In most of the projects that I am seeing right now, there were silos in marketing, in risk, and so on. Now with the data platform, because it’s a very technical thing, the IT team is in charge of the data. However, there’s another role that’s also important in big institutions, especially in regulated environments, and that’s the Data Governance Officer.”
He adds that this may create tension because the data governance role is usually outside the IT team. Data governance officers play a crucial role, because they are the people responsible for ensuring that the company remains compliant.
“Data governance is absolutely essential with a single data platform, because all the data is in one place, and it has to be managed properly. Banks and other financial institutions, and to a certain extent energy companies, are in very complex regulatory environments, and they can’t afford to have anything go wrong.”
Stavros is unclear where the situation will end up, but is alert to the potential for this tension to be a threat to data and AI adoption and maturity.
“The challenge that I see is that usually there are one or two key stakeholders in the company that are responsible for analytics, marketing, risk and compliance. We certainly cannot assume that AI will be totally decentralised or that the process of adoption will be smooth just because it’s nice to play with these new generative AI systems. For real production use cases, we still need Chief Analytics Officers in place, and IT is still playing an important role.”
Small steps into the future
Stavros has also observed a key change when it comes to customer priorities—and one that chimes with what we have seen at SAS. There is now a clear push for the use of AI.
“More or less every customer right now has artificial intelligence as a priority. Even the CIOs who have not been part of the advanced analytics era over the last 20 years, and were focused on traditional data consolidation, now all want to use generative AI. There is a whole range of different purposes and use cases, from risk analytics to fraud. Customers are also becoming very knowledgeable about AI and the drivers of narrative AI.”
He notes that the main focus now is on generative AI, and particularly in customer service. However, there is a recognition that the move may need to be gradual. Stavros has also noted an important trend: people working together with machines, rather than being replaced.
“Customer service usually has unstructured data and this is often not well processed at the moment, even in companies that are doing some traditional text analytics. The first priority is therefore to build a use case often around helping the customer service team. Generative AI is not often being used directly with customers, but instead as a way to support human agents to review requests. These systems can process the information very fast, build content and accelerate the communication with the customer.”
A need for practical skills
It may be ten years since Harvard Business Review branded data science as “sexy”, but we’re still seeing problems with getting the right skills into businesses. Stavros agrees—but suggests that data science may not be the problem.
“The skills question is tough right now—but not in AI. Universities have a lot of programmes covering that, and there’s plenty of research into what we might term pure AI. However, universities are not focused on data engineering. We find that many of our customers have the knowledge of AI, they know what a neural network is, say, but they have very few skills in data engineering.”
This causes huge problems in migrating to cloud.
“It’s nearly impossible to discuss AI without talking about cloud or hybrid cloud, or consolidate data to a single platform. You cannot automate the use cases, or move them into production. I believe there is a definite skill gap in data engineering and it’s getting worse. Everybody is trying to find people, but it's very tough to find the proper resources.”
One answer to this skills gap is to adopt a data and AI platform that enables no/low/yes-code approaches. This enables more people to access this technology, embedding a data-driven approach within the organisation. It also becomes possible to increase collaboration, automate repeatable tasks and develop reusable assets. Another answer may lie in opportunities like the SAS Hackathon. This exposes people to new ways of working, including developing and managing data pipelines and working with APIs: key data engineering skills. Stavros agrees that this may be suitable.
“We have done this kind of thing at Performance: not hackathons but an Academy to train people. I think this is important. Hackathons really put people’s brains to work, and provide really good results.”
