A safety map is only as trustworthy as the data underneath it. So here is exactly where ours comes from, how precise it is, and — just as importantly — what it cannot tell you.
The source
The official layer of our map comes from data.police.uk, the open data service published by the UK police. It carries individual crime and anti-social behaviour records with street-level location information, along with the police and court outcomes attached to them.
Forty-three police forces across England, Wales and Northern Ireland feed into it. The data is refreshed monthly and published under the Open Government Licence v3.0, which is what makes it legal for anyone — us included — to build on.
What gets recorded
Every record falls into one of fourteen categories:
- Anti-social behaviour
- Bicycle theft
- Burglary
- Criminal damage and arson
- Drugs
- Other theft
- Possession of weapons
- Public order
- Robbery
- Shoplifting
- Theft from the person
- Vehicle crime
- Violence and sexual offences
- Other crime
For someone walking home, a handful of these carry most of the weight — robbery, theft from the person, public order, and violence and sexual offences. A cluster of shoplifting reports outside a supermarket tells you something about the shop, not about your walk. This is why we show you the category behind every marker instead of collapsing everything into a single risk score.
The part most maps do not explain: a pin is not an address
Street-level crime locations are deliberately imprecise, and this is the single most misunderstood thing about UK crime data.
Rather than publishing the exact spot, the police service snaps each crime to the nearest point on a master list of more than 751,000 anchor points — street centres, parks, shopping areas, transport hubs. Every one of those points is chosen so that it covers at least eight postal addresses.
This means a marker on any UK crime map — ours or anyone else’s — can never point at a particular house, and was never meant to. It tells you roughly where, not precisely where. That is a privacy protection, and it is a good one.
Accuracy varies by force
The police service publishes its own accuracy figures, and they are worth knowing. The share of records that are geocoded correctly ranges from roughly 60% to 97% depending on the force. Matching crimes to their eventual court results is more variable still.
So the honest summary is: this data is very good at showing you patterns across an area over time, and much weaker as a statement about any single incident at any single point. We build the map to suit what the data is actually good at.
What it cannot tell you
- It cannot tell you about unreported crime. A great deal of crime never reaches the police, and what never gets reported never appears on any map.
- It is not real-time. The feed updates monthly, so the freshest official data is typically several weeks old. Community reports fill some of that gap, which is why we carry both.
- It cannot tell you whether a street is safe right now. It describes what has been reported before, which is a guide to risk, not a forecast.
How we use it
We mark incidents, and only incidents. We do not mark areas as dangerous on the basis of who lives in them, we do not blend demographic data into a risk score, and we do not intend to. Every marker traces back to something that was reported.
If the data cannot support a claim, we would rather not make the claim.