Car advice should start with the driver.

Every required review section is complete

Generation, model years and recommended version are stated

Specifications are checked against authoritative or manufacturer material

Safety claims link to the relevant official record and are reproduced only where permitted

Independent editorial conclusions are compared across multiple named UK publications

Research sources are classified by reuse rights before drafting begins

No first-hand driving claim is made unless RoadMatch actually tested the vehicle

Source scope differences and uncertainty are disclosed

Image reuse rights and UK context are recorded

An editorial QA pass is required before publication

Collect a practical profile

Budget, condition, body style, seats, doors, powertrain, mileage, driving environment, charging access, main-driver age, economy, costs, desired car character, older-premium appetite, features, vehicle age, mileage cap, makes and search area.

Apply firm eligibility rules

Avoided makes, body style, fuel, vehicle age, seating, doors, gearbox, must-have equipment, review-based reliability minimum, indicative budget fit, sporty driver appeal and age-related insurance safeguards are checked before ranking. An ineligible model cannot be rotated into the shortlist, and we return fewer cars rather than pad a result.

Plug-in hybrids require confirmed regular home/work charging in our initial policy. Choosing no regular charging excludes electric cars too; public-only or unconfirmed EV charging is clearly flagged. Adaptive cruise can meet an ordinary cruise-control requirement, but ordinary cruise cannot meet an adaptive requirement. Optional equipment is always flagged for confirmation on the individual car.

Score every eligible model

Each active dimension returns a 0–100 fit. A cost dimension carries no weight when the user selects “No preference”. The overall review verdict and the user’s stated preference for value, refinement, driver appeal or technology are separate inputs. These are base weight points, not percentages; active weights are normalised for each profile. Character rises to 36 points for Sporty, 32 for Premium or Technology, 28 for Value and 22 for Balanced. Driving fit uses 10 points; review quality uses 8–10. The base set is:

budget

14 points

body Style

7 points

practicality

7 points

powertrain

9 points

driving

6 points

economy

5 points

ownership

6 points

features

6 points

age

3 points

make

2 points

reliability

5 points

driver Age

0 points

character

14 points

quality

6 points

ownership Tradeoff

10 points

A larger budget with sporty character and no cost preference increases the emphasis on refinement and premium feel within genuinely driver-engaging choices. It does not reward expense alone. No economy preference removes MPG scoring; an explicit MPG target, reliability floor or newer-ownership choice remains independent. Match scores are editorial fit estimates, not a probability that a particular car will suit you.

Keep the reasoning visible—and the choice varied

The result stores the explanation for every dimension. Strong contributions become “why it fits”; low contributions become trade-offs. Reliability uses our desk-researched model assessment and a confidence label. A device-local seed can reorder only 1.5-point bands of comparable eligible cars. Clearly stronger matches stay ahead; near-equal leaders can rotate. Each model family gets one place in that rotation through its best-fitting variant, and different models from the same make are allowed. This means a Toyota Aygo can get a fair turn beside a Skoda Citigo without promoting a materially worse match. The optional AI beta receives the same interpreted priorities; server checks restore stronger match tiers if its ordering overlooks them. It cannot invent cars or override firm filters.

Hand off only supported filters

A separate marketplace adapter converts the model and profile into an allow-listed search object. Unsupported or uncertain filters remain visible for the user to apply manually.