In plain English: every two years the government's National Fraud Initiative cross-checks councils' records against each other — pension rolls against death records, council-tax discounts against the electoral roll — and flags every payment that looks like it shouldn't have happened. This number is the total count of those flags a council received in one exercise: its raw fraud-risk exposure before anyone investigates anything. A match is a lead, not proven fraud, but the size of the pile tells you how much checking a council faces. No per-council figures are published centrally; they would have to be scraped from committee papers or requested from the audit body.
| name | nfi_matches_total |
|---|---|
| dataset | 08-national-fraud-initiative |
| kind | prospective |
| type | numeric (count) |
| unit | matches per biennial exercise |
| grain | council, per biennial NFI exercise |
| role | input / exposure measure |
| source | not published centrally — would require (a) scraping ~300+ councils' audit/governance committee papers (ModernGov etc.) for biennial NFI update reports, or (b) FOI/data request to the PSFA (or Audit Wales / Audit Scotland / NIAO) for per-participant extracts from the NFI web application |
| period | biennial exercises |
| missingness | expected high and uneven — only councils that publish committee-paper NFI updates would be covered via route (a) |
| score_note | Exposure/workload denominator, not an efficiency signal in itself. |
| peer_group | council class (district / county / unitary / metropolitan / London) — not yet scored on the leaderboard |
The number of data matches released to a specific council per NFI exercise (e.g. Bridgend, 2024–25: 17,255). It would indicate the council's data-matching exposure and workload, and serve as the denominator for investigation and hit rates. The NFI publishes no per-council data, so this exists only via bespoke collection; match volume reflects data-set size and participation, not fraud prevalence.