Almost every stalled AI project traces back to the same place: the data underneath it. The good news is that “data readiness” isn't a mysterious technical bar. It's a handful of practical questions you can answer about your own business. Here's what that actually involves.
You don't need perfect data to start. You need to know what you have, where it is, and whether you're allowed to use it. Work through these five in order.
1. Does the data exist at all?
AI learns from examples. If you want it to draft quotes, it needs past quotes; to triage tickets, past tickets. Before anything else, confirm the relevant history is actually recorded somewhere, not just living in people's judgment. If it isn't captured, capturing it is step one, and that's fine to discover early.
2. Can you get it out?
Data trapped in a system you can't export from is, for practical purposes, not yet usable. You don't need a fancy pipeline, but you do need a way to pull the information out (an export, a report, an API). If a vendor holds your data hostage, that's worth knowing now.
3. Is it consistent enough to trust?
“Clean” is the wrong word. Real business data is never spotless. “Consistent enough” is the bar: are the same things recorded the same way often enough to spot a pattern? A few typos and gaps are normal and workable. The thing to watch for is data that means different things in different places.
4. Who's allowed to see it?
Some of your most valuable data is also your most sensitive: customer records, contracts, anything personal or regulated. Before it's used, you should be able to answer: who can access it, does using it this way fit our obligations, and would a third-party AI provider ever see it? Good projects settle this up front, with access controls and clear terms, not after the fact.
5. Is there enough of it?
“Enough” depends on the task, and it's usually less than people fear. A workflow that's happened a dozen times won't give AI much to learn from, while one that happens daily will. If volume is thin, that often points you toward a different, more frequent place to start.
The honest summary
Data readiness is mostly about knowing your own situation, not achieving perfection. Exists, exportable, consistent, permitted, and plentiful enough: answer those five and you're in good shape to start, gaps and all. Most businesses are closer than they think, and the few that aren't usually have one specific thing to fix rather than a mountain to climb.
Want a quick read on where your business stands across data and the other things that matter? The readiness scorecard takes about two minutes and points you to a sensible first step.