Blog / Web data by industry

UK Sold House Price Data: What Listing Sites Add

29 Jul 2026Perceptlayer7 min read

For UK sold house prices, start with HM Land Registry's Price Paid Data: it is free, covers residential sales in England and Wales, and is the authoritative record of what was paid. Listing sites add what the register leaves out: the asking price and how it changed, the agent, rooms and size, and valuation history. The useful dataset joins the two, and since August 2026 Land Registry publishes a UPRN for new sales, which makes that join exact.

Where should UK sold prices come from?

From HM Land Registry, for England and Wales. Its Price Paid Data covers residential property sold for value and lodged for registration, is free to use for commercial and non-commercial purposes under the Open Government Licence, and is updated monthly (HM Land Registry, 2026). It is the register, so it is the figure to use for "what did this sell for".

Each row has a transaction ID, the price from the transfer deed, the completion date, the postcode and address parts, a property type code (D detached, S semi-detached, T terraced, F flat, O other), an Old/New flag (Y for a new build), a duration (F freehold or L leasehold) and a PPD category (A for standard sales, B for additional types such as repossessions and buy-to-lets). The files have no header row; the columns come in that documented order.

It has limits worth knowing before you build on it. The two most recent months are incomplete, because the gap between a sale and its registration typically runs from two weeks to two months. Prices are transaction prices, not market values: a transfer within a family can be fixed far from market value. And it covers England and Wales only, so Scotland and Northern Ireland need other sources.

What do listing sites add to sold prices?

The context around the sale. Land Registry records the end of the process, the price paid. A listing site records the process itself: what the seller asked, when the property was listed, whether the price was cut, who the agent was, and the rooms and size buyers saw. Zoopla also exposes valuation history and nearby sales for a property by its UPRN. The table maps each question to the source that answers it.

Rightmove's sold-properties endpoint filters by postcode, property_type (DETACHED, FLAT, TERRACED, SEMI_DETACHED), tenure and sold_in (3, 6, 12, 24 or 120 months). Zoopla's sold endpoint takes a location and sale-date bounds. Both are listing-site views of completed sales, useful for joining to listings; for the price itself, the register wins.

Which listing fields are worth keeping?

Keep the fields that explain a price, and the ones you need to match records. Our recorded Rightmove search for location=London&channel=BUY shows the asking side. This excerpt is its first listing, trimmed with …:

{
  "id": 90486465,
  "bedrooms": 3,
  "bathrooms": 2,
  "displayAddress": "Cavaye Place, Chelsea, London, SW10",
  "location": { "latitude": 51.486678, "longitude": -0.180225 },
  "propertySubType": "Apartment",
  "tenure": { "tenureType": "LEASEHOLD" },
  "listingUpdate": { "listingUpdateReason": "new", "listingUpdateDate": "2026-07-03T22:30:31Z" },
  "price": { "amount": 1495000, "currencyCode": "GBP", … },
  "customer": { "branchId": 298991, "branchDisplayName": "OB Private, Mayfair", … },
  "displaySize": "1,285 sq. ft.",
  "channel": "BUY",
  "addedOrReduced": "Added on 03/07/2026",
  …
}

Two details from the same recording matter for sold-price work. The second listing in it is Land at 65000000 with channel COMMERCIAL_BUY, inside a residential BUY search, so filter on channel and propertySubType before computing anything. And two of the three displayAddress values stop at the street and outcode (SW10, W1W), which is too coarse to match one listing to one Land Registry sale.

How do you join a sale to a listing?

By UPRN where you can, by address where you must. HM Land Registry now publishes a lookup from Price Paid transaction IDs to UPRNs. It was first published on 28 August 2026, isn't back-dated, includes a UPRN only where Land Registry can reliably link the title address to an addressable location, and carries its own licence conditions and an Ordnance Survey attribution (HM Land Registry, 2026).

With a UPRN, every Zoopla property endpoint is one call: /v1/zoopla/property, property-valuation-history, property-agents and property-nearby-sold each take uprn and cost 1 credit. That turns a bare sale row into a sale with its valuation history and agents. We don't have a recorded Zoopla response, so check which fields each returns in the playground on the Zoopla API page before you design your tables.

For older sales, or for Rightmove, match on address. Normalise the postcode (upper case, single space), the house number or name (PAON), the flat (SAON) and the street, and treat a match as certain only when all of them agree. The article on deduplicating listings across sources covers the matching method. Where the listing address stops at the outcode, don't force a property-level match; use the aggregate comparison below instead.

How do you measure the asking-to-sold gap for an area?

Compare like with like at area level: median asking price from current listings against median sold price from the register, for the same outcode and property type. This works without any property-level match. It measures the market, not individual negotiations, so it's a first look rather than a precise discount. The script reads a Price Paid CSV by column position, which follows Land Registry's documented order, and pages a Rightmove search for one outcode.

import csv
import os
import statistics
import requests

OUTCODE, LR_TYPE, RM_TYPES = "SW10", "F", {"Flat", "Apartment"}

# Land Registry Price Paid Data: no header row; columns in documented order.
sold = []
with open("pp-2026.csv", newline="") as f:
    for row in csv.reader(f):
        price, postcode, ptype, category = int(row[1]), row[3], row[4], row[14]
        if postcode.split(" ")[0] == OUTCODE and ptype == LR_TYPE and category == "A":
            sold.append(price)

asking, credits = [], 0
for page in range(1, 6):
    r = requests.get("https://api.perceptlayer.com/v1/rightmove/search",
                     params={"location": OUTCODE, "channel": "BUY", "page": page},
                     headers={"X-API-Key": os.environ["API_KEY"]}, timeout=60)
    if r.status_code != 200:
        break
    credits += int(r.headers["X-Credits-Used"])
    data = r.json()["data"]
    for p in data["properties"]:
        if p["channel"] == "BUY" and p.get("propertySubType") in RM_TYPES:
            asking.append(p["price"]["amount"])
    if not data.get("has_next_page"):
        break

if sold and asking:
    s, a = statistics.median(sold), statistics.median(asking)
    print(f"{OUTCODE} flats: median sold {s:,.0f} (n={len(sold)}), "
          f"median asking {a:,.0f} (n={len(asking)}), gap {(s - a) / a:+.1%}")
print("credits used:", credits)

We tested the script on a sample row and the recorded Rightmove fields, not against live calls. It keeps only category A sales, which drops repossessions and other non-standard transfers, and only BUY listings of flats. Read the result with its sample sizes. The two sets differ in timing (sales completed months ago against today's asking prices) and in mix (which flats happen to be listed now), so a gap of a few per cent can be noise. Run it monthly per outcode and watch the trend.

What does the listing side cost?

Land Registry's data is free. Each listing-site call is 1 credit on success and 0 on failure, so the cost is pages and lookups. The script above reads up to 5 Rightmove pages per outcode: 20 outcodes monthly is at most 20 × 5 = 100 credits. Adding Zoopla valuation history and agents for 200 sales with a UPRN is 200 × 2 = 400 credits.

That's 500 credits, inside the 1,000 free credits every verified account gets each month. The credits docs explain the balance and pricing has what packs cost. The Rightmove API guide covers search paging and its record cap; to start, run a sold-properties call for your own postcode in the playground on the Rightmove API page.

uk propertysold pricesland registryhouse pricesreal estate data