It happens in almost every company that wants to “do something with AI now”: step one is a tool list. A platform here, a copilot there, a data catalog on top. A year later there are plenty of licenses – and still no answers.
The fallacy is always the same: tools get mistaken for strategy. But a data strategy doesn’t answer a procurement question – it answers a leadership question: Which decisions do we want to make better – and what data do we need to do it?
First the decision, then the data, and only at the very end the tool
If you’re serious about this, you reverse the order. It starts with business goals and the specific decisions that are currently made too slowly, too expensively or too imprecisely. From those follow the use cases – and only from the use cases does it become clear which data is needed, and at what quality.
Then come the uncomfortable questions: Who owns the data internally? Who is accountable when it’s wrong? What rules govern access, protection and use? Governance is not a bureaucratic appendix – it’s the part where most strategies die, because nobody owns it.
And the architecture? It comes last. Once it’s clear what’s needed, selecting tools is suddenly easy – and considerably cheaper. The tool follows the task. Never the other way around.
“The tool comes last – not first.”
How to recognize a real data strategy
It fits on a few pages. It names accountable people by name. It contains measurable goals instead of buzzwords. And it has a rhythm: measure, learn, sharpen. Everything else is a shopping list with a cover page.
The complete roadmap
Our guide “Data Strategy in 9 Steps” – compact, as a PDF.