Methodology
How the ranking is made
1 · Coarse search
Listings are filtered on the Step-1 basics only: radius from your postcode, price band, property type, detachment and bedroom range. The filter is deliberately forgiving — a listing missing a value is kept and flagged, never dropped. Prices up to 10% below your floor and 5% above your ceiling stay in, marked for you to verify.
2 · Deep analysis
Each shortlisted home gets a full attribute read. Where a value is judged rather than stated, it is badged amber as AI-inferred. Analyses are cached, so a re-run of an edited brief is near-instant.
3 · Exclusions and score
Anything you marked Essential but the home lacks, marked Avoid and the home has, a tenure you didn't tick, or a value outside your min–max is removed — but only when the listing states it or it is seen in photos or the floorplan. If it was only assumed, the home stays with a penalty and a "Check" flag. Unknown values never count for or against a home. Survivors are scored additively and normalised to 0–100, then the top 10 are returned.
Weights in points
- price
- 25
- bedrooms
- 20
- distance
- 15
- propertyType
- 5
- detachment
- 5
- bathrooms
- 6
- receptionRooms
- 4
- size_sqft
- 8
- garden
- 6
- offstreet_parking_cars
- 5
- property_condition
- 5
- ownership_status
- 5
- garden_facing
- 4
- study
- 4
- master_ensuite
- 4
- annex
- 4
- outbuilding
- 4
- swimming_pool
- 4
- chain_free
- 4
- cash_buyer_only
- 4
- garage_cars
- 3
- property_period
- 3
- central_heating
- 3
- water_supply
- 3
- waste_water
- 3
- onstreet_parking
- 3
- log_burner
- 2
- range_cooker
- 2
- loft_space
- 2
- conservatory_garden_room
- 2
- solar_panels
- 2
- heat_pump
- 2
- ev_charging_point
- 2
Closeness credit
- atBoundary
- 0.5
- inferredPenalty
- 0.5
- boostCap
- 0.1