Six signs a Health Check is worth its fee.
If two or more of these describe your programme, the underlying cause is usually configuration decisions taken early — and they are still fixable.
The business keeps its own spreadsheet
The clearest signal of all. When teams maintain a parallel list, they have already judged the golden record and found it wanting.
Records are merging that should not be
Over-matching destroys confidence faster than under-matching and is far harder to unwind. One wrongly merged customer can discredit an entire dataset.
The stewardship queue only grows
The platform is generating more tasks than the team can clear. Downstream, an unworked backlog is indistinguishable from bad data.
Go-live has slipped more than once
Usually a symptom of unresolved ownership rather than technical difficulty — a disagreement about who is authoritative, surfacing as a delivery problem.
Nobody can explain why two records merged
If the match logic is not explainable to a business owner, it will not be trusted by one, regardless of how well it performs statistically.
An AI or analytics initiative is blocked
The programme downstream has stopped, and the reason given is data quality — but nobody has isolated which definitions are actually in conflict.