AI is all about data, right?
It wasn't true in 2006, and it won't be true in 2025. For years, the tech world has repeated one phrase with the persistence of a bad ringtone: "Data is the new oil." It resonated deeply. It made executives nod their heads. It sold a lot of CRM-like systems.

 

We were promised that if we collected everything – every click, every message, every customer signal – it would give us access to instant customer magic.

But in practice, something different happened. We ended up with CRMs filled with half-completed forms, duplicate customer data from fifteen other applications, fields no one remembers creating, and notes that resemble the personal diary entries of a confused intern.

Data mantra

The revolution never happened. Why? Because collecting data is easy, but using it effectively is the part everyone skipped. And yet, in AI, we repeat the same mantra: "AI is all about data."

I was in a meeting recently where someone proudly proclaimed this. It was a bit like a bluff, but the room nodded. Pens moved dramatically. Someone took a deep breath, as if wisdom had sunk in. But here's the inconvenient truth: if AI were truly all about data, every company with a twelve-year-old CRM application would be an AI superpower right now. They're not. Not even close.

Digital fossils

Why is that? Because most organizations don't actually have "data." They have historical artifacts. Digital fossils. Long-forgotten input that means everything to auditors and almost nothing to AI. AI doesn't want dusty records; AI wants context.

Context is the new differentiator. Data tells you what happened then, but context tells you what matters now. And as AI focuses on real-time decision-making, context trumps volume. What's happening with this customer right now? Which choice creates value now? Which signal is important now and in the future? These are the questions that truly matter: real-time trumps volume. Signals trump storage, and even more importantly, context trumps history.

And even with good data… reality gets strange. AI doesn't "work" just because your data warehouse looks impressive in a photo. You still need business-focused models. Orchestration that actually aligns with workflows, with guardrails to prevent AI from running wild, and humans who don't immediately revolt when bots behave strangely.

And while it feels like all the talk about guardrails is fueled by vendors who don't really know how to test AI, unconstrained bots will continue to behave strangely: "You were right, human. The mushroom was deadly poisonous. Let's try this again."

Data problem or design problem?

These problems don't just arise from bad data. They also stem from weak design, missing context, and workflows held together primarily by optimism.

AI has never been a data problem; it has always been a design problem. The winners won't be the companies that hoard the most data. Winners will be the companies that design systems that understand a moment and can act on it while it's still relevant. That requires the courage of innovation, not the columns.

So no, AI isn't "just about the data." It's about whether you're willing to redesign the way your business works. Because the real reason AI fails isn't a lack of data, but waiting for perfect data. And perfect data never arrives. Perfection is the excuse. Progress is the strategy.

(Ziptone/Merijn te Booij)

Merijn te Booij Merijn has 30 years of experience in all facets of customer contact. He was CMO at Genesys, Contact Center Manager at ANWB, and more. A true CX enthusiast, driven by endless curiosity and tech optimism, Merijn combines his past and future to create exceptional customer experience experiences. He will be writing a series of columns for Ziptone in the coming period.

Featured, Opinions, Technology