How CompareForecourt works

You save listings yourself: by pasting a link and the advert text, by uploading screenshots of the advert, or with our bookmarklet. We do not crawl or scrape listing sites. An AI extraction step turns the advert into structured fields (price, mileage, year, VAT status, service history claim and so on), together with a confidence level. Extraction reads what the seller wrote; it cannot add information the advert does not contain, and it can occasionally misread an unusual advert. Every listing keeps a link to its source so you can check the original.

MOT history is sourced from official DVSA data via the MOT history service, looked up by registration plate. It includes past test results, recorded mileages and advisories. We flag a possible mileage discrepancy when a recorded odometer reading is lower than an earlier one. Results are cached for up to seven days, so a test taken yesterday may not appear immediately.

DVSA data verifies the MOT record only. It does not verify the price, the seller, the service history, the mechanical condition, or anything else in the advert.

Each listing gets a completeness score from 0 to 100. It measures how much decision-relevant information is available, not how good the vehicle is. The score is a fixed calculation, not an AI judgement: we check a list of core fields for the vehicle's category (a van's list includes VAT status; a caravan's skips mileage and MOT because caravans have neither) and count what is missing. Fields that a particular listing site never publishes are not held against the advert. A low score means "you do not yet know enough to compare this fairly", and the listing page spells out exactly which fields are missing.

The value score compares each listing against the others in your own shortlist, never against the whole market. It is a weighted blend, and the weights are fixed:

• Price: 30% (compared against similar vehicles in your shortlist)
• Mileage: 25% (within the same vehicle category)
• Age: 10% (within the same vehicle category)
• Service history: 10%
• Damage/write-off markers: 10%
• Previous owners: 5%
• MOT time remaining: 5%
• ULEZ compliance: 5%

Price, mileage and age are percentile ranks against your other saved vehicles of the same kind: vans are compared with vans, never with cars. The other components use fixed scales (for example, a Cat S marker scores far lower than a clean history). When a component is unknown, it drops out and the remaining weights are rescaled, so an advert is not silently punished for a field the seller never stated; the completeness score is where missing information shows up. Wording in the advert that suggests mechanical trouble (non-runner, head gasket, sold as seen) applies a separate penalty of up to 40 points. New vehicles are not penalised for having no MOT or service history before either is legally due.

The score is guidance for prioritising your own research. A high score does not mean a vehicle is mechanically sound, and we never declare a winner for you.

Total-cost-of-ownership figures are fixed calculations from published rates: current UK vehicle excise duty (road tax) tables, national average pump prices, and the Ofgem-capped domestic electricity rate for EV charging (assuming mostly home charging). Fuel costs use the advert's stated economy figures and your annual mileage. Depreciation uses typical age-based curves, not a valuation of the specific vehicle. Insurance is never estimated: premiums are too personal to guess responsibly, so we show the insurance group where known and leave the rest to your own quotes. Where an input is unknown we show a dash, never a made-up zero.

The shortlist assistant answers questions using the listings you have saved and their MOT history. The model reputation profile additionally uses web search and shows its citations. Deep analysis compares your saved listings in detail. None of these features can:

• verify that a seller's claims are true;
• check outstanding finance, theft records or insurance write-off registers (see data sources for what we do not have access to);
• inspect the physical vehicle or predict a specific vehicle's faults.

AI output is generated text. We ground it in your saved data, but you should treat it as a well-organised starting point for your own checks, not as a verdict.

Everything you see in the app belongs to one of five buckets:

• Official records: MOT history sourced from DVSA data.
• Seller claims: everything extracted from the advert. Unverified.
• Your input: annual mileage, postcode area, notes.
• Calculated: completeness score, value score, running-cost estimates. Deterministic, from the rules above.
• AI-generated: assistant answers, deep analysis, model profiles. Grounded but generated.

Always, for any vehicle you are serious about, and especially when: the price is high relative to your budget, the advert scores low on completeness, the MOT history shows recurring advisories, the vehicle is a motorhome or caravan (damp cannot be assessed remotely), or you are buying at a distance. A professional pre-purchase inspection costs a small fraction of the repair bill it can save you from.

If anything on this page does not match what you see in the app, or a guide contains an error, tell us via the feedback page. We correct mistakes and date our updates.