Your AI Strategy Has a Data Problem — And Most Leaders Don’t See It Coming

Here’s a number worth sitting with: 42% of companies abandoned most of their AI initiatives in 2025. Not paused. Not restructured. Abandoned. Meanwhile, over 80% of AI projects fail outright — more than double the failure rate of traditional IT projects.

We’ve spent billions chasing AI’s promise. So why are results so elusive?

The answer isn’t the AI. The issues lie in the data we are feeding it..

You Can’t Bolt Intelligence onto Chaos

Gartner predicts that through 2026, organizations will abandon 60% of AI projects unsupported by AI-ready data. Not because the use cases aren’t compelling but because the foundation isn’t there.

For companies that make physical products, that foundation comes down to one thing: specification data. This includes formulations, ingredients, packaging materials, supplier compliance records, regulatory attributes, and more. 

To add to the complexity, there is no one-size-fits-all data structure for managing this information, making it even more challenging for companies to organize and maintain effectively.

As a result, this critical data often ends up scattered across spreadsheets, buried in PDFs, trapped in ERP systems that weren’t designed to hold it, and stored in the institutional memory of people who won’t be around forever.

AI amplifies whatever it’s built on. When built on fragmented, outdated data, confusion becomes amplified. Built on connected, accurate, spec-first data, intelligence will dominate.

This distinction matters enormously in product development, where a bad AI recommendation isn’t just inefficient — it can trigger a regulatory issue, a supplier dispute, or a product recall.

The Question That Changes Everything

In conversations with product leaders across CPG, food and beverage, and packaging, one frustration surfaces more than any other — and it’s not about AI at all.

“We can’t answer basic questions about our own products.”

Not advanced analytics. Just: how many products use this ingredient? Which suppliers provide this material? What’s affected if this regulation changes tomorrow?

These should take seconds. Instead, they take days of emails, spreadsheet archaeology, and educated guesses.

The moment leaders realize it doesn’t have to work this way — that you can query your specs, trace relationships, and get answers in real time — that’s when the conversation shifts from should we use AI to how do we build the foundation that makes AI actually work.

The Advantage You Can’t Outsource

Every company now has access to capable AI — the same foundational models, comparable tooling. It has become a commodity.

Competitive advantage will come from your data. The value lies in our specifications,formulations,compliance history, and your supplier relationships. These are  the only assets a competitor can’t license from the same vendor you use.

But if that data isn’t structured and connected, you don’t truly own it. And the window to build this foundation is shorter than it looks. AI capabilities are advancing faster than most companies’ data maturity. The gap between organizations that have their specifications structured and those that don’t is already widening — and it will only accelerate.

The companies making progress aren’t waiting for their data to be perfect. They’re starting now, with one product line or one category, and compounding from there.

If you’d like to read more of our perspective on AI, download our latest Executive Brief: The AI Data Crisis.