UK Land Registry Sold Prices: What the Data Does and Does Not Tell You

Your gross development value rests on comparables, and the only free comparable evidence a valuer and a lender will both accept is HM Land Registry Price Paid Data. It is genuinely good data. It is also incomplete in ways that will embarrass you if you quote it without knowing them. This post covers what the dataset contains, what it leaves out, how far behind reality it runs, and how to use it to price a scheme.

What Price Paid Data actually is

Price Paid Data records residential property sales in England and Wales lodged with HM Land Registry for registration. It runs from January 1995, it is published monthly, and it is released under the Open Government Licence, so you can use it commercially.

Each record gives you the price paid, the date of transfer, the address and postcode, and:

• Property type – detached, semi-detached, terraced, flat or maisonette, or other.
• Whether the property was newly built or an existing building.
• Tenure – freehold or leasehold.
• A transaction category. Category A is a standard residential sale for value. Category B covers everything else HM Land Registry captures, including repossessions, buy-to-lets, and transfers to non-private individuals.

Scotland and Northern Ireland are outside this dataset. Registers of Scotland and Land and Property Services hold the equivalent records there.

What it leaves out, and why that matters to you

This is the part people skip. HM Land Registry lists the exclusions plainly, and several of them go straight to the heart of a development appraisal:

• Commercial property. Transactions where VAT is stated to be included are excluded. If you are pricing a commercial building for conversion, Price Paid Data will not give you the purchase comparable.
• Transfers not for value. Gifts, divorce settlements, transfers between parties, compulsory purchase and transfers under a court order.
• Right to buy sales at a discount. Which means some estates look thinner in the data than they were.
• Leases of seven years or less.
• Anything not lodged for registration. Land that has never changed hands since registration became compulsory in the area will simply not appear.

One more absence worth naming: there is no floor area in Price Paid Data. A price per square foot from this dataset is always the product of joining it to something else, usually EPC records. That join is not perfect, so treat a £/ft² figure as an indicator, not as evidence.

The lag is longer than you think

Price Paid Data records the date a sale was lodged for registration, and HM Land Registry notes that the gap between sale and registration typically runs from two weeks to two months – which means the two most recent months in the dataset are always incomplete.

Practically: if you are reading the last two months and concluding the market has gone quiet, you are probably reading the lag. Use a trailing twelve-month window for a GDV and treat the most recent quarter as provisional.

How to build a defensible comparable set

A lender’s valuer will discount any comparable they can pick a hole in. Give them fewer holes.

1. Match the product, not the postcode. Same property type, same tenure, same broad size. Three matched comparables beat thirty postcode averages.
2. Separate new-build from second-hand. The new-build flag exists for a reason. A new-build premium is real and it is not transferable to a refurbishment.
3. Screen out Category B. A repossession is a forced sale. It is evidence of something, but not of open market value.
4. Twelve months, then widen. Go back further before you go further out. Distance damages a comparable faster than age does.
5. Write down what you excluded and why. That note is what turns a spreadsheet into an argument.

Doing this across a search area rather than a street

The free government search answers one address at a time, which is fine for one site and useless for a search area. BOOM! puts the same sold price records on the map as a Sales History layer, with each transaction showing the address, date of transfer, price, property type, tenure, whether it was new build, and floor area and price per unit area where those are available. Transactions that can only be placed to a postcode rather than an address are marked as such, so you know which ones to treat with care.

You can then filter the set the way a valuer would, on:

• Date of transfer, price, and price per square foot or square metre.
• Property type, tenure, floor area and number of habitable rooms.

Above that sit average freehold and leasehold sale prices by postcode, on both gross internal and gross external area, and House Price Index prices and volumes by authority with charts – which is the quickest way to see whether a local market is thinning out before you commit to it.

In SiteSeeker, last sold price and last sold date are constraints in their own right. Filtering for titles that last changed hands a long time ago, at a low price, is one of the more productive ways to build an outreach list – see finding off-market sites without an agent for how that fits into a sourcing process.

Comparables tell you what the finished units are worth. They say nothing about whether you can build them – for that, start with the constraints that stop a scheme.