If you open a UK crime map and a whole city-region looks almost empty, the usual reaction is that the area got safer. Often that is the wrong reading.
Street-level figures on data.police.uk are not a complete count of crime. They are the subset each force chooses to publish that month. When a force pauses or severely reduces publishing, the map goes quiet even if reporting and recording have not.
CrimeMapsUK treats that as a publication gap. This article explains how we detect gaps, why they matter, and which forces currently fall below our threshold, using data loaded through June 2026.
What a publication gap is
A healthy force series produces a fairly stable monthly volume of published street-crime records for years. A gap is when that volume collapses relative to the force’s own history.
Our rule of thumb (in code under apps.data.coverage):
- look at the last twelve calendar months of published totals
- compare that total with the highest twelve consecutive months in earlier history for that same force (the peak year-long stretch, not an average and not a national figure)
- if recent volume is under 15% of that historical peak, and the peak was large enough to judge, we flag a gap
For West Midlands, that peak is roughly 348,000 records in a high earlier twelve-month window; recent twelve-month volume is about 25,000.
That rule is deliberately blunt. It is meant to catch “used to publish tens of thousands a month, now publishes almost nothing,” not gentle year-to-year change.
A gap is not the same as:
- a genuinely quieter year in a force that still publishes normally
- one missing month that gets backfilled later (see the data.police.uk changelog)
- our own site being down
Why the map can mislead
Hex maps and force totals only show what was published. In a gap:
- hexes look sparse or empty
- year-on-year charts can look like a miracle improvement
- postcode district pages for that force are withheld, because a district page built on paused data would invent a false calm
- we refuse to sell a neighbourhood report when the local series is in this state for the same reason
Greater Manchester is the clearest public example: street-level publishing fell to near zero for a long stretch, then returned only in a thin trickle. West Midlands shows a similar pattern from May 2023 - monthly published volume fell from around 30,000 records to around 2,500 and has stayed there. That is why Birmingham can look “low” next to forces that still publish fully.
Forces currently flagged (crime data)
Snapshot based on CrimeMapsUK rollups through June 2026. Recent totals are the last twelve published months. “Historical peak” is the highest earlier twelve-month stretch for that force. Ratios are rounded.
Little or no recent street-level crime data
These forces have roughly zero published street-crime rows in the last twelve months in our database (or their monthly series ends before the current tip):
| Force | Notes |
|---|---|
| British Transport Police | No recent street-crime months after early 2023 in our load |
| Northumbria Police | Series ends April 2023 |
| North Yorkshire Police | Series ends April 2023 |
| Police Service of Northern Ireland | Series ends April 2023 |
| South Wales Police | Series ends December 2023 |
Still publishing, but far below their own history
These forces still appear in recent months, but recent twelve-month volume is under 15% of their historical peak:
| Force | Recent 12 months | Historical peak year | Approx. share of peak | Sharp drop around |
|---|---|---|---|---|
| Cleveland Police | ~100 | ~95,000 | under 1% | May 2023 |
| Greater Manchester Police | ~4,500 | ~390,000 | ~1% | mid-2022 (after earlier disruption) |
| Thames Valley Police | ~10,000 | ~211,000 | ~5% | long-running reduction since 2018 |
| West Midlands Police | ~25,000 | ~348,000 | ~7% | May 2023 |
| Hampshire and Isle of Wight Constabulary | ~15,000 | ~193,000 | ~8% | May 2023 |
| Cumbria Constabulary | ~3,800 | ~44,000 | ~9% | May 2023 |
| Nottinghamshire Police | ~12,000 | ~130,000 | ~9% | May 2023 |
| Lancashire Constabulary | ~19,000 | ~200,000 | ~10% | May 2023 |
| Kent Police | ~27,000 | ~222,000 | ~12% | May 2023 |
| Norfolk Constabulary | ~9,500 | ~79,000 | ~12% | May 2023 |
| Gloucestershire Constabulary | ~8,800 | ~66,000 | ~13% | early 2026 (after an earlier step-down) |
| Dyfed-Powys Police | ~7,100 | ~50,000 | ~14% | May 2023 |
Several of those drops cluster on May 2023. That timing shows up across multiple forces in the published files. We do not claim a single official explanation here - only that the open data volume changed sharply.
Force pages such as Crime in Birmingham (West Midlands) and Crime in Manchester still exist where we have history, but recent months should be read with this gap in mind.
Are there public explanations?
Greater Manchester is the well-documented case. Local reporting and HMICFRS material describe long-running problems after GMP’s iOPS / records-management change: the force has often been unable to supply a clean street-level feed to data.police.uk even while other Home Office returns continued. The data.police.uk changelog has carried repeated “crime data not provided” notes for GMP, and GMP has said publicly that aligning its systems to the police.uk return is complex.
For most other gap forces, including the May 2023 cliff shared by West Midlands, Lancashire, Kent, and others, there is usually no clear public write-up of a lasting policy change. The changelog mostly logs single-month misses and later refreshes (“data not provided… refresh as soon as possible”). That does not explain a multi-year step-down to a thin trickle. Until a force or the Home Office publishes a reason, the honest reading is simply: street-level open data volume fell; we do not know why from the public record.
What this does not mean
- It does not mean those places have almost no crime.
- It does not mean every sparse hex elsewhere is a gap - some areas really do publish few records.
- It does not mean CrimeMapsUK is missing a secret complete dataset. If it is not on data.police.uk, we do not have it.
For the wider limits of the source (approximate locations, outcomes, reporting), see the methodology page and why we don’t predict crime.
How to check an area that might be affected
- Open the relevant force area page and look at the monthly chart, not just the latest total.
- Compare recent months with 2021-2022 (or earlier) for the same force.
- If the line falls off a cliff and never recovers, treat the map as incomplete publishing, not as a safety score.
- For a single postcode, try the free check on the reports page - if we detect a gap, we say so plainly instead of selling a misleading PDF.
Will this list change?
Yes. Forces resume, thin out, or pause again. The gap rule is re-evaluated from the monthly series whenever derived stats are rebuilt. This article is a readable snapshot for August 2026, not a permanent league table.
If you need the precise definition used in product decisions (district publishing, report sales), the implementation lives in the coverage helper documented from the methodology page.