Is our data ready for AI?
Rarely all of it, and that is not an absolute prerequisite. What matters is the quality of the data the intended use depends on. Start there, rather than with an endless clean-up.
Updated on 24 October 2026
What is going on.
When a company considers AI, the question of data comes up quickly, often with a harsh verdict: it is scattered, incomplete, badly organised. The temptation is then to launch a big tidy-up first, and put AI off until later.
That project drags on, costs money and rarely serves the use that will matter. Not all data is equal: a given use depends on a few specific sources, with a level of quality of its own. It is better to start from the use, identify the data it needs, check its state and improve it where necessary.
What we observe.
- No company has perfect data.Those that succeed with AI do not have better data: they know which data matters.
- Quality depends on the use.Data that is good enough for a statistic may be unusable for an individual decision.
- Data problems often reveal flow problems.Information is often badly entered because the person entering it does not see what it is for.
What you can do right now.
- For your first use, list the data it depends on and go and look at its real state.
- Ask the people who produce that data what stops them entering it correctly.
- Improve that data first, and leave the rest for later.