There are some interesting nuggets of data here for anyone insuring London area retail premises, pubs, old churches, petrol stations, nail salons/barbers or HMO insurance;
A new survey by Seenity, the risk intelligence company in the insurance sector, reveals – using its platform and agentic AI technology – the London high streets that have decreased their risk. Insurers and lenders that still rely on pre-pandemic location scores are pricing London’s high streets on out-of-date assumptions. Based on data from 2016 to 2019 and 2023 to 2026, 29% of the capital’s 488 ranked high streets moved into a different risk band.
Seenity analysed 13 million police records, 287,000 road collisions, 222,000 fires and official air-quality data for every high street defined by the Greater London Authority. It then compared 2016 to 2019 with 2023 to 2026 data, weighting each crime by its seriousness.
Seenity uses its AI-powered risk intelligence platform to help insurers understand, predict and reduce risk. Its platform allows an insurer to combine its own data with real-world external data, including location, environmental conditions and asset. Asset characteristics include the physical, structural and operational attributes of insured assets, their age, condition and usage, as well as their geographic, environmental and surrounding risk factors. From this Seenity builds risk models that support an insurance company’s underwriting, pricing and claims decisions.
A mixed picture for London’s high streets
The data shows a mixed picture. Crime harm on London’s high streets rose 7%, driven by a 74% rise in shoplifting and a 53% rise in theft from the person. Over the same period, burglary fell 24%, non-residential fires fell 17%, and injury road collisions fell 16%. Nitrogen dioxide in the air across London’s high streets fell by about a third (34%).
New King’s Road in Fulham is a good example of how London’s high streets have changed since before the pandemic. On New King’s Road, crime harm fell by 40%, burglary fell by 51% and vehicle crime by 66%.
For commercial and property underwriters, the mix matters as much as the total. A count-based score would say London’s high streets got worse. A severity-weighted view shows that burglary, the heaviest-weighted high-volume crime, fell on all 22 of the most-improved streets (see complete list below in notes to editors), and by 38% to 77% across the top ten.
London’s ten most-improved high streets
|
# |
High street |
Borough |
Fall in overall risk |
Fall in crime harm |
|
1 |
New King’s Road, Fulham |
Hammersmith and Fulham |
29% |
40% |
|
2 |
Kingsbury Road (Roe Green), Kingsbury |
Brent |
29% |
39% |
|
3 |
West Green Road, West Green |
Haringey |
27% |
39% |
|
4 |
Brixton Road, Kennington |
Lambeth |
27% |
37% |
|
5 |
Burnt Ash Lane, Sundridge Park |
Bromley |
27% |
38% |
|
6 |
Packington Street |
Islington |
26% |
36% |
|
7 |
Lordship Lane, Wood Green |
Haringey |
26% |
35% |
|
8 |
Highgate High Street, Highgate |
Haringey |
25% |
36% |
|
9 |
St Margaret’s Road, St Margarets |
Richmond upon Thames |
25% |
44% |
|
10 |
Clapham Road, Kennington |
Lambeth |
24% |
33% |
Overall risk is calculated by combining multiple risk factors, including crime, road safety, fire exposure and air pollution, into a single composite risk score for each street.
When the data, not the street, changes
The study also shows how changes in source data can create false trends. In July 2022 police.uk replaced every anonymised crime location in London. Other changes in data included the Met’s move to the CONNECT records system in February 2024, which shifted about 1,500 offences a month between theft categories, and the April 2025 reclassification of shop theft involving threats as robbery. The 22 streets on the final list held up under three catchment definitions, 5,000 random weightings of the five hazards, a median-year test and a displacement check on the surrounding 300 metres.
Oren Atia, CEO of Seenity, said: “Most risk models still treat a place as if it had a fixed character: a postcode is scored and the label follows it for years. This study shows how costly that assumption can be. But change over time is also where data is most unreliable. A new police records system, a reclassified offence or a refreshed map of anonymised crime locations can manufacture a trend that never happened on the street. For an insurer or a lender, the right question is never only ‘how risky is this place?’, but ‘which way is it moving, and how confident are we?'”
What it means for risk assessment and insurers
Seenity draws four lessons for insurers and lenders:
- Re-score, don’t label. Portfolios priced on a pre-pandemic view of place carry stale assumptions in both directions.
- Measure severity, not counts. Crime volume rose while burglary fell.
- Look across hazards. Crime, road danger, fire and air quality can move in opposite directions on the same street.
- Treat source changes as part of the model. A data refresh or a reclassification can look like a real change in risk.

Be the first to comment