Building a state service index that departments would accept
A composite index is only useful if the departments being measured believe it is fair. How confidence scoring made that possible.
38
Districts compared
7 factors
In the confidence model
Versioned
Index weights
0-100
Composite scale
An illustrative scenario composed from how the platform is designed to be used. Not an account of a named customer, and the figures are not audited customer results.
Situation
Where the team started
A state wanted a single service index to compare districts. Two previous attempts had failed — not technically, but politically. Districts with thin data collection scored badly and argued, correctly, that the index was measuring reporting rather than service.
The index had no versioning, so when weights changed, a district's score moved without anything in the world having changed. That destroyed whatever credibility remained.
Approach
What changed
- 01
Publish confidence alongside score
Every district score carried a confidence value computed from freshness, survey coverage, field reporting, community coverage, media coverage, source diversity and model confidence. A thin-data district was visibly a thin-data district.
- 02
Version the weights
The index model version was recorded on every score. A change in weighting produced a new version rather than silently rewriting history, and both could be compared.
- 03
Show the contributors
Each score exposed its per-signal-family contribution, so a district could see exactly which input was pulling its number down and whether that input was fair.
- 04
Fix coverage rather than argue about it
Because low confidence was visible on the main dashboard rather than buried in a data quality annex, coverage gaps became field deployment decisions instead of debating points.
Result
What it produced
- Districts could see and contest the specific input driving their score
- Coverage improved in the districts that had previously argued about the index
- Score changes could be attributed to the world or to the model, never ambiguously both
- The index survived its first contested review
A composite index earns acceptance through legibility, not accuracy. Departments will accept a number they can take apart and disagree with far more readily than a better number they cannot inspect.
The part teams tend to underestimate
Ready to transform civic outcomes?
See how Civiq.plus can help your teams make better decisions, faster.