The data source: HM Land Registry Price Paid

When a property transaction completes in England or Wales, the solicitors are legally required to register the sale with HM Land Registry. That registration includes the full address, sale price, property type, whether it's a new build, and the date of completion.

The Land Registry publishes this as the Price Paid Dataset — updated monthly, free to access, and covering all residential sales since 1995. It's not a survey, an estimate, or a valuation: it's the actual price, at the actual completion date, on the actual property.

Important: The dataset covers England and Wales only. Scotland has a separate register (Registers of Scotland), and Northern Ireland uses the Land Registers of Northern Ireland.

How to search

The results group data by street within the postcode, not by individual property. You see the aggregated picture for each street: median price, average, year-on-year change, transaction count, and optionally broken down by property type.

What each metric means

Median price
Anchor figure
The midpoint of all sale prices on the street in a given year — half sold for more, half for less. More robust than the average because it isn't pulled up or down by a single unusual transaction. This is the figure to compare against an asking price.
Average price
Mean
The arithmetic mean of all sales. Useful as a secondary check, but can be skewed by outliers — one very expensive sale on an otherwise modest street can push the average up significantly. Compare median and average together: a large gap between them suggests a wide spread of property quality.
YoY change
Year on year
The percentage change in median price from the previous year to the current year. A positive figure means the street's median price rose; negative means it fell. This is your market direction signal — and three consecutive negative years is a meaningful trend, not noise.
3yr rolling avg
Smoothed trend
The average of the median price over the current and previous two years. Smooths out single-year volatility — a year with only two or three sales can produce a median that looks like a crash or a spike. The rolling average tells you where the underlying trend is heading, not just what happened last year.
Sales count
Confidence level
The number of registered transactions on the street in that year. Treat medians based on 1–2 sales as directional indicators, not benchmarks. 5+ sales in a single year for a single street gives you real statistical confidence in the figure.
New build premium
New vs established
Where available, the percentage by which new build properties sold above the median for existing (established) stock on the same street. Useful for understanding whether new build asking prices on the street represent genuine market value or developer premium.

Why full-postcode data gives you street-level precision

Street level on SoldByStreet is defined by the full postcode — something like SW11 4PG typically covers just one or two streets. That's meaningful precision. A postcode district average (e.g. SW11 as a whole) spans dozens of streets and can hide £100,000+ price differences between neighbouring roads depending on proximity to a station, school catchment, or road noise.

Searching a partial postcode — SW11 rather than SW11 4PG — won't give you the same granularity. You'll see more streets, but the data is spread across a broader area. For genuine street-level comparison, always use the full postcode.

What the data doesn't show

The Land Registry records the registered sale price, which is the price agreed at exchange (or sometimes renegotiated before completion). It doesn't record:

Fixtures, fittings, and chattels — if a buyer paid separately for the kitchen or white goods, that amount doesn't appear in the Land Registry figure.

Condition at sale — a price paid for a property in need of full renovation and one in pristine condition will look identical in the data.

Developer incentives — cash-back schemes or part-exchange arrangements can mean the effective price paid was lower than the registered figure, especially for new builds.

These caveats don't undermine the data — they're just context for interpretation. The median across multiple transactions smooths most of this noise out.