Identical-on-paper homes sold £63,000 apart: the limit of every online valuation

Research by ValuQ: we paired 13.7 million repeat sales in HM Land Registry data and grouped homes that are identical on paper, the same property type, on the same postcode sector, bought in the same year and held for the same length of time. Inside those groups the whole local market is held constant, so whatever spread is left is the part of a home's value no area-level model can predict. This is part of the ValuQ Property Watch series.

Identical-on-paper homes still sold tens of thousands of pounds apart

Take two homes of the same type, on the same streets, bought in the same year and held the same number of years. You would expect them to perform almost identically. In fact the middle half of such homes ended up 21 percentage points apart in total return, around £63,000 on a £300,000 home. The full tenth-to-ninetieth spread was 43 points, closer to £129,000.

This is not a few oddities. More than half of all homes (51%) finished more than ten points away from what their look-alike neighbours did, and almost three in ten finished more than twenty points away.

Why a single online estimate cannot capture this

The spread is not a data error. It is the real, irreducible part of a home's value that area-level information cannot see: condition, the exact spot on the street, internal layout and light, the work the owner did, and the particular buyer who turns up on the day. A model fed postcode averages knows none of it. A person standing in the room can price all of it.

Instant online valuations are now everywhere, and lenders increasingly lean on automated models too. They work well enough across a whole market. For one specific home they carry this built-in error bar, and it does not shrink with a cleverer algorithm, because the missing information was never in the data to begin with.

Widest for terraces and in the North

The size of the blind spot is not the same everywhere. It is widest where homes vary most behind similar front doors. Terraced homes carry the widest spread of any type; detached homes the narrowest. By region the spread is widest in the North and narrowest in the South.

What it looks like on real streets

• Canary Wharf (E14): 118 flats bought the same year and held three years ranged from a 12% gain to a 31% gain.
• Bishop's Stortford (CM23): detached homes bought the same year and held four years ran from a 53% gain to a 72% gain across the middle half.
• A block of Manchester terraces (M9) bought the same year ran from a 51% gain to a 127% gain.

What it means if you are selling

1. Treat any single online estimate as the middle of a wide range, not a price.
2. The error bar is widest for terraced homes, homes in the North, and homes you have improved, exactly where a generic estimate is most likely to leave money on the table.
3. The things that close the gap are the things only a person can see: condition, layout, the exact position, the quality of a refit.
4. Several real local valuations, side by side, are how you find the range your home actually sits in.
5. Checking costs nothing, and it is your strongest defence against selling on a number that was never more than a guess.

ValuQ gives UK homeowners free, side-by-side property valuations from competing local estate agents, and the seller stays anonymous until they choose to connect. It is free for homeowners, always. One number from a website is a guess. Several local valuations, compared together, show you the real range.