One Service. Seven Names. No Owner.

Illustrative scenario. Records, events and arithmetic are constructed examples, not findings about a named organisation.

The booking form calls it an assessment. The service page calls it a consultation. The staff directory names a clinic that closed last year. Seven systems describe the offer. None carries a shared service identifier.

Search returns all three. The caller asks which one to use.

Departments serve different audiences. Their terminology reflects professional practice, reporting obligations and local history. The same public label can conceal different eligibility rules; different labels can describe one service.

After maintaining provision through reorganisations, funding changes and the departure of staff who negotiated the original arrangements, the managers responsible for these records have reasons to resist a central naming exercise that treats every distinction as an editorial inconsistency.

Names can carry real duties.

They are right to distrust a committee that mistakes uniform wording for a data model. Replacing every label with the same phrase would erase some distinctions and leave the underlying relationships untouched.

An interface redesign cannot decide whether two records identify one service. A search replacement cannot supply an ownership decision nobody has made.

You can spare them the workshop.

Where every system maintains a separate bilateral mapping to every other system, seven systems create twenty-one system pairs. Ten create forty-five. This is a model of that architecture, not an inevitable count for every integration.

If each pair needs one twenty-minute review per monthly cycle, the workload rises from seven hours to fifteen. Corrections and disputed matches sit outside that allowance.

The extra systems inherit a bill.

A name change now requires several translations. Without a maintained identity, each team must determine whether it is seeing a new service, a replacement or another label for the existing one.

The search engine becomes the defendant. It returns duplicates, separates related records and ranks a retired page above the current booking route. The vendor proposes tuning. A replacement platform enters the budget discussion.

Search configuration can fail. Ranking and indexing deserve inspection. But a search system receiving contradictory records cannot establish the organisation’s intended identity merely by assigning those records better scores.

The procurement brief asks for relevance.

No.

Search did not lose the link. You never stored it.

Give the service a stable identity, then represent its labels, eligibility, locations and booking relationships explicitly. Record when a service is renamed, split, merged or retired. Someone must own those decisions.

The experienced coordinator who explains the differences to callers is performing entity resolution from memory. Each helpful answer conceals another request the published records could not settle.

You called that personal service.

It was an absent key.