This article was first published on Forbes.com on June 3, 2024.

CIOs and CDOs everywhere are busy chasing after AI, but they’re getting ahead of themselves. They’ve lost sight of a very important prerequisite: data strategy. Successful AI requires not just lots of data, but lots of high-quality data built on a well-executed data strategy.

The pressure to demonstrate capability with AI is so strong right now. In fact, according to new research by MIT Technology Review, 82% of C-suite leaders say that scaling AI or generative AI use cases to create business value is a top priority this year.

Executives and boards aren’t thinking about strategy — they are frantically looking for fast AI outcomes. One of our customers went to their board saying they needed some budget to build a data platform. The board said, “We have no money for data projects, but we have unlimited funds for AI innovation.” So the customer changed the proposal title, went back to the board and said, “I need to build an AI platform.” 

Not surprisingly — the board funded his project.

We can laugh about these types of stories, but they illustrate a difficult truth: executives don’t understand what makes AI work. For years, they’ve made the mistake of separating AI from data strategy and it’s coming back to bite them. AI is quickly showing executives that a successful data strategy is the foundation for AI strategy. 

Oversight in the C-suite: AI strategy is not separate from data strategy

For years, executive teams have funded and supported two different data stacks: one for analytics, one for product. Each stack had its own datasets, methodologies and mission. Typically, the analytics stack was owned by the CIO and the product stack, which included AI/ML, was owned by the CTO. One used Databricks and the other used Snowflake, leaving data teams scratching their heads trying to “make AI happen” across two platforms.

But AI doesn’t operate across organizational silos and independent strategies. It needs governed access to high-quality data from every part of the organization. C-suite execs are now realizing they’ve needed one platform, not multiple integrated platforms built for unique purposes. They should have been building around one strategy — not integrating multiple projects.

The most advanced CIOs and CTOs are the ones who’ve always had one big data strategy. Mike Hite, CTO at Saks, is seeing big wins with his unified strategy and data platform. He’s completely rebuilt Saks’ data ecosystem on Fivetran and Snowflake, giving Saks a major advantage over their luxury retail competitors. They’re using AI and LLMs to deliver personalized recommendations to customers via call center agents, strengthening relationships and creating unparalleled shopping experiences.

Hite’s new data ecosystem gives Saks flexibility and scalability they’ve never had before. He recently shared with us: “This is the beauty of the data ecosystem we’ve built with Fivetran as a core piece: We can think about data in a fundamentally different way, less consumed with the cost of getting it in and instead focusing on the value it can bring to our customers and brand partners.” 

To get AI and ML, you have to excel at the basics

Everyone wants the latest and greatest AI/ML tech, but many organizations still struggle with everyday financial reporting. You can’t do advanced, industry-leading work if you haven’t gotten the basics down yet. 

AI and ML are synergistic results of well-executed data strategy where governance, lineage, access and data integration all work in harmony. They’re not independent strategies that come together at a later date. If you’ve got two different data stacks, you’re going to be building governance twice. This is too complicated and you’re probably working twice as hard to make it work. Worse, you’ll likely need to rebuild it all before you could jump ahead to AI. And if you don’t get it right, it will be a major barrier for moving forward.

I recently met with a room full of CIOs, all at different levels of data and tech maturity. The only ones who are able to do anything with AI are the ones who’ve recognized the importance of data readiness — collecting, classifying, cleaning, and making meaning out of an organization’s data. Just like BI where they say "bad data in, bad data out" — it’s even more true for AI. If your AI models, or LLMs are working on top of bad data, you get bad AI results as well.

The less data-mature organizations are still dealing with legacy problems, fixing clunky reporting and haven’t yet migrated to a modern data stack. They have no shot at competing on AI capabilities, especially with yesterday’s tech. 

AI success comes from a great data strategy, not the other way around

There are no shortcuts. You can’t start with AI and work your way back towards a strong data foundation. The result of a strong data foundation is high-quality, governed data. It doesn’t happen by accident, nor is it a side effect of a thriving company. It’s built on reliable data pipelines, a modern cloud data platform and a collaborative culture in governance. Without all of that in place — good luck.

Executives need to set the stage for technological excellence and market innovation. The winners won’t be those who sidestep the prerequisites, but the ones who truly understand that data strategy is now a key part of business strategy.

Start there and everything else will follow.

[CTA_MODULE]