Bringing the details together.
One moment…
One moment…
Methodology
Where the data comes from, how the Buy Score is worked out, and what a report can’t tell you. Plain answers, including the limits.
Put in a UK registration and we look up the car’s MOT history with the DVSA. From that record we work out the MOT status, how the mileage has built up and which defects keep coming back. We add statistics for similar cars, such as how often the model fails its MOT and on what, and blend it all into one Buy Score.
Each report is saved as a snapshot when it’s made, with the records it used, the source versions and the date. Opening it again shows the same snapshot, not a fresh lookup.
AB19 CBB, EF20 CBB, GH18 CBB are made up to show what a full report looks like. Their MOT records and model statistics are invented, and they’re always labelled Sample.
Our data
Live reports use the DVSA MOT History API for the car itself. Model statistics come from DVSA’s anonymised MOT test data and DfT collision statistics. They describe similar cars, never the individual car.
The Buy Score
The Buy Score is one number from 0 to 100 for ranking the cars on your shortlist. It’s built from nine parts. Each part is scored 0–100 on its own, then blended using these weights:
A part with no data is left out and counts against coverage. It’s never filled in with a guess. If less than 50% of the weight is backed by data, the overall score is withheld and the report says why. Three parts aren’t connected yet, so no car can currently reach more than 82% coverage.
Confidence follows the weakest part that counts, and drops when coverage is lower.
They reflect how much each part tells a buyer. They aren’t a calibrated model: that needs outcome data we don’t have yet. Use the score to rank a shortlist, not as a verdict on whether to buy.
Every report has a panel of lights, one for each thing a buyer should know at a glance.
Amber: ask about it. A defect flagged at more than one MOT, a part that fails on similar cars over 25% more often than on peer models, a mileage reading that goes backwards, an expired or failed MOT, a dangerous defect, or a recall not recorded as fixed.
Green: the record checks out. For example, every mileage reading goes up, or the car has never failed on anything dangerous.
Off: no data to judge. Never a guess. The price light stays off until a listings source is connected.
The latest recorded test and its expiry give the status on the report’s date. A later failure outranks an earlier pass. This isn’t a certificate that the car is roadworthy today.
Readings are put in date order and converted to miles where needed. A drop of more than one mile is flagged as a potential discrepancy, which allows for rounding when converting units. Recording mistakes happen: a discrepancy isn’t proof of clocking.
Miles a year is the difference between the first and last readings, divided by the days between them and multiplied by 365.25. It’s withheld if readings go down or there’s less than 90 days of history.
A part at the same location is flagged when it appears in two inspection cycles in a row, or in at least three of the latest five. A failure and a retest within 14 days count as one cycle. A repeat doesn’t prove the defect is still there.
Checklist items come from repeated defects, the latest advisories, mileage discrepancies and parts that fail more often on similar cars. Each one links back to the evidence in the report.
Model figures describe a group of similar cars, not the condition of the car you’re looking at. For real registrations the groups come from DVSA’s anonymised MOT data (2025, cars and light vehicles, initial tests only), matched on make, model, fuel and age band. Mileage and test year aren’t adjusted for yet.
Weighted defects per test are compared with peer models. A reliability of 35 means the rate is lower than 35% of that peer set. It isn’t a 35% chance of anything.
The model’s weighted defects per test divided by the peer average. 1.31× means a 31% higher weighted defect rate than average. It doesn’t estimate repair bills.
A group is only used if it has at least 1,000 tests, mapping confidence of 90% or more, a positive peer rate, at least 10 peer models and an update in the last two years. Otherwise reliability and maintenance burden are withheld, not estimated. These thresholds are a starting point and still need calibrating.
Annual fuel cost = miles a year ÷ UK mpg × 4.54609 litres per gallon × price per litre. The price starts at the latest DESNZ weekly UK average, with its date shown. The MOT record doesn’t include fuel economy, so for real registrations you enter the car’s mpg. You can change every input.
If you enter vehicle tax, we add it. Tax isn’t looked up. Insurance, servicing, depreciation, finance and repairs aren’t included, and there’s no EV charging calculator yet.
We can’t see the car in front of you, so look it over, or have it inspected, before you buy. These aren’t checked at all: outstanding finance, write-off history, stolen status, VIN identity, ownership history and model theft rates. DVLA and commercial history providers aren’t connected. A missing check is never a clear result.
Collision involvement is a statistic for the model, never this car’s accident history. Vehicle crime near a postcode is an optional, separate lookup: it isn’t a theft risk for the model or a stolen-vehicle check.
Safety recalls come from the recall state DVSA returns with live MOT records. “No known recall” isn’t the same as all clear, because it depends on manufacturer reporting. Confirm any campaign with the manufacturer.
CarBuyerCheck is in early access. Full reports are shown with their prices, but payments aren’t switched on yet, so nothing can be bought and nothing is charged. Until they are, real registrations show the free sections with the paid ones locked, and the sample reports open in full.
MOT history, mileage and the warning lights for any car.
Model weak spots, running costs, a viewing checklist and questions to ask the seller, for that car.
Three full reports and a side-by-side comparison. 30 days to use them.
Report snapshots are stored so their links keep working. My cars and your checklist ticks are kept in this browser, on this device only. There are no accounts yet, so nothing syncs between devices. Clearing your browsing data removes them, but not the report snapshots.
For the local vehicle crime lookup, your postcode goes only to Postcodes.io to find its location; Police.uk receives the coordinates. The postcode isn’t saved.