Publishing service data the public can audit
Twelve datasets published on a schedule, with documented methodology and a lineage trail from source system to published figure.
- Datasets published
- 12
- Publication cadence
- Monthly
- Manual prep hours
- -80%
- Methodology pages
- 12
The situation
The council published service statistics annually, prepared by hand over several weeks. Freedom of information requests about the same figures took three weeks to answer because the working files were not retained.
Constraints
Everything published had to be explainable to a committee without a data specialist in the room, and the methodology had to be public alongside the numbers.
What we did
Twelve datasets now generate on a monthly schedule from the source systems, with lineage recorded from record to published figure. Each has a methodology page written in plain language, published at the same time.
Where a definition was ambiguous we wrote the decision down rather than resolving it silently in SQL.
What changed
Preparation dropped from forty hours a release to eight. The more useful change is that information requests about published figures are now answered in days, because the lineage already exists.
Results
| Measure | Before | After |
|---|---|---|
| Datasets published | 0 | 12 |
| Hours to prepare a release | 40 | 8 |
| Published methodology | None | Every dataset |
| Freedom of information response time | 21 days | 4 days |
What we left behind
- A project in the client's own repository, with tests and CI
- Documented definitions with named owners
- Runbooks for every scheduled job, written for a non-engineer
- Two weeks of training for the team taking it over
Other engagements you might want to read
Tell us which number you cannot trust
Send one paragraph about the report that always starts an argument. A data lead, not a salesperson, replies within one business day.