Where the numbers come from
- Sold prices — HM Land Registry Price Paid Data, Crown copyright, Open Government Licence v3.0. Every sale this report uses is a real, registered transaction — not an asking price, not an estimate.
- Price movement over time — HM Land Registry's UK House Price Index, by local authority and property type (flat/detached/semi-detached/terraced), Crown copyright, OGL v3.0. Each sale's price is adjusted forward to today's terms using this index, so a 2018 sale is compared on a like-for-like basis with a 2024 one.
- Floor area and energy rating — the Energy Performance Certificate register (England & Wales), Crown copyright, OGL v3.0. Floor area shown is the EPC's own figure — broadly Gross Internal Area, not a RICS-measured figure.
- Flood risk — the Environment Agency's open flood-monitoring data (England), OGL v3.0. We check how far the address is from a designated Flood Warning or Flood Alert Area, and whether a flood warning is currently active nearby. This does not cover surface-water or groundwater flooding — only the EA's own Warning/Alert Area boundaries.
- Plot and building shape — Ordnance Survey INSPIRE polygon data and OS Open UPRN (where ingested for the address's local authority), Crown copyright, OGL v3.0. Shown only where we can verify the geocoded point sits precisely on this address, not a postcode-level guess.
- Map imagery — Google Maps Platform static map tiles, shown with Google's own attribution.
How comparable sales are chosen
We look for real sales of the same kind of property, as close to this address and as recent as possible, in this order:
- This street — sales on the exact same street.
- This postcode — if the street alone doesn't have enough.
- The wider area nearby — a roughly 1-2.4km radius, nearest sales first, only if neither of the above has enough.
At each of those three, we look first within the last 10 years. Only if none of the three has at least 3 sales within 10 years do we widen the time window to the last 15 years — never further back than that, at any geography. A tight street match with stale evidence is treated as worse than a wider area with fresh evidence, not automatically better.
Within whichever geography and time window we land on, we also try to narrow to sales of the same property type (flat / detached / semi / terraced) specifically — only falling back to blending other types in when there genuinely aren't enough same-type sales nearby, and even then each blended-in sale is scaled to this property's own type level using real local price differences, not a flat guess.
Flats in a shared block are only compared against other flats in the same building at the street-tier step — never a different block on the same street, even if that block shares the street name.
This home's own last sale, when we can find one, is shown as its own separate fact — “this home's last sale” — never folded in as if it were one more comparable. We only count it as the subject's own sale when the SAON (flat/unit), PAON (building/house), street and postcode all match exactly — not just the house number, and never a loose text match.
How the headline figure is reached
By default, the figure is the median (the middle value) of the comparable sales found, each one indexed forward to today's terms as described above.
Exception — the floor-area method. When at least 5 of the comparable sales carry both a sale price and an EPC floor area, we also calculate each one's price per square metre, take the median of those, and apply it to this specific property's own EPC floor area. We only use that figure instead of the plain median when it comes out within 15% of what the plain median already said — otherwise we keep the plain median and don't use the floor-area figure at all. This guards against a single wrong floor-area match (the wrong flat's EPC certificate, for example) swinging the headline a long way from what every comparable sale actually sold for.
If this home has its own very recent sale (roughly within the last 5 years), the figure leans more on that real transaction and less on the general comparable-sales median — the more recent the sale, the more weight it carries. A sale from longer ago is shown as information only and never changes the figure.
The range shown alongside the headline is a fixed band around that central figure, not a separate statistical calculation.
What “confidence” means
Every report shows either Normal confidence or Limited confidence. It drops to Limited whenever any of the following is true for this specific address:
- Fewer than 5 genuinely comparable sales were found (fewer than 3 is treated as especially thin).
- Even the newest comparable sale used is more than 2 years old.
- No close match on property type was found, so other types had to be blended in.
- The comparable sales found needed the wider 15-year window rather than the usual 10.
- The only usable evidence came from the wider area nearby, with no floor-area-based cross-check available to narrow it down further.
The label always reflects the real evidence behind this report — it never defaults to “Normal” just because a number came out the other end.
What this is not
- Not a RICS valuation. This is an automated estimate built from public sold-price and registry data — it is not a Royal Institution of Chartered Surveyors valuation and carries no professional indemnity.
- No inspection. Nobody has visited, measured, or assessed the condition of this specific property. Owner-provided details (condition, recent works, photos) are recorded in the report for reference but do not currently change the figure.
- Not legal or financial advice. It's one input among several a buyer, seller, or lender might use — not a substitute for an independent valuation or professional advice.