Data Score
10 questions. Find out how ready your data is for AI - and where to focus first.
AI is only as good as the data underneath it. Copilot, Fabric IQ and data agents read your semantic models and source data, so when definitions disagree or data is patchy, the answers come back confident and wrong. These 10 questions assess the four things that determine whether AI will work: your data infrastructure, your governance, how your team accesses and uses data today, and whether you have a clear AI use case to pursue.
Is your data stored in a centralised system rather than across spreadsheets and local files?
What your score means
Each of the 10 questions scores 0, 1 or 3 points, giving a total out of 30. Your result falls into one of three bands.
AI Ready (21-30).
The foundations are in place. Focus on choosing the right first use case and making sure your team can act on what AI surfaces.
Getting There (12-20).
Real progress, with gaps that will limit what AI can deliver before they are addressed. A focused readiness programme closes them without a full rebuild.
Foundation First (0-11).
AI will not add much yet. Centralise your data, clean it, agree definitions and open up access. Most mid-market businesses start here, and it is the right place to start.
What has changed for Power BI users
Copilot can now answer questions straight from your Power BI semantic models. Fabric IQ, generally available since FabCon Europe 2026, lets Microsoft Copilot choose the relevant model or report and answer from it, and its MCP server gives other AI agents the same access. Queries run as the signed-in user, so row-level security still applies.
One thing worth checking: by default, each semantic model lets read-only users reach it through Copilot and agents. Decide which models should be reachable and switch the setting off for the rest.