How Dwellio measures rent

We store every rental listing we find with its link and the last time we saw it on its property site. A listing counts as live if we have seen it in the last 14 days. Once a day, everything older is marked as removed. If a removed listing turns up again, it goes straight back to live.

That daily check stopped running in March 2026 and only came back on 14 October 2026. In between, nothing was ever marked as removed, so every listing we had seen since March still counted as live. When the check returned, the counts fell to the homes we had actually seen in the previous fortnight.

So listing counts from before and after 14 October 2026 can’t be compared, and neither can monthly changes across that date. Wherever a month-on-month change would compare the two, we hide it and label it “not comparable before 14 Oct 2026”. Our September 2026 report and the cheapest cities post quote the older counts and say so.

A live count is our sample, not the size of the market. In the biggest cities we read a fixed number of result pages from most sites, so we see a large share of what is advertised, not every listing. One property site on its own may show more London homes than our whole count.

Asking rent is what a landlord or agent is advertising for a home that is available now. That is what we measure. It moves first, because it reflects new lets only.

The ONS figure comes from the Office for National Statistics’ Price Index of Private Rents. It is the average rent being paid across private tenancies, including long-running ones at older rents. It changes more slowly and is published about two months behind.

Open-market rent is a figure the First-tier Tribunal sets for one specific home in England when a tenant challenges a rent increase (Housing Act 1988, section 14; see GOV.UK’s guide to applying for an open market rent determination). It looks at that property, not a town. Our medians cover every size and condition of home in a city, so they are not a valuation of any one home and can’t stand in for the tribunal’s figure. This is general information, not legal advice.

We publish the median: line the prices up and take the one in the middle. A mean (the usual “average”) is pulled about by a few luxury or mispriced listings. A median barely moves.

Before taking a median we clean the listings:

• Weekly rents are converted to monthly: the weekly price × 52 ÷ 12.
• Whole homes only. Rooms in shared houses are kept out of the overall and bedroom figures, and where a city page shows a room figure it is per room.
• Anything under £400 a month is dropped as a scam, a per-room price or a typing error. For homes with three or more bedrooms there is a second floor of £200 per bedroom, which catches student houses advertised at one tenant’s share with the whole house’s bedroom count.
• A 4+ bedroom listing priced below the city’s 1-bed median is treated as a per-room price and left out of the 4+ bed figure.
• We need at least 3 clean listings to work out a median at all; publishing needs more (below).

Every figure we publish names, next to the amount, the size of home it covers, how many live listings sit behind it, its month, and whether it is an asking rent or the ONS figure, so it still says what it is when it is quoted on its own. For one size the count is the live listings of that size; for “all sizes” it is the live whole homes, because rooms are left out of that median. Counts are taken before junk prices are dropped, so a median’s own sample is smaller: usually a little, but for 4+ bed homes in student towns much smaller, because houses advertised at one tenant’s share are left out of that median. The sites named are the ones the figure draws on: whole-home figures never include SpareRoom, which lists rooms.

Every figure passes the same checks before it appears anywhere: a city page, the UK rental index, our emails or a tool. A figure that fails is left out. We don’t estimate a replacement.

• Enough listings. A city page shows figures once it has 10 live listings. Each bedroom figure needs at least 10 live listings of that size on a city page, and 30 in any table that ranks cities against each other.
• An overall figure that fits. A city’s overall median must sit between the lowest and highest of its published 1-, 2- and 3-bed medians. If it doesn’t, we show the bedroom figures and leave the overall one out.
• Like with like. A month-on-month change is shown only when both months were counted the same way (so not across 14 October 2026), both overall figures passed these checks and, in the index, both months had at least 50 live listings.
• Postcode areas need at least 3 listings, and postcode areas that are rare in a city’s listings (usually strays from a neighbouring town) are left out.

Why there are no studio figures

When we can’t read a listing’s bedroom count, it is currently stored as zero, the same as a studio. That lets whole flats into the studio figure: on 18 October 2026, 44 of the 152 cities with both figures had a studio median above their 1-bed median. Until that is fixed we show no studio medians anywhere.

These checks apply everywhere we publish a figure, city pages included. Each time the figures are rebuilt we log everything that was held back, so a problem like Exeter’s is seen and fixed at the source rather than quietly hidden.

Listings come from six UK property sites: Rightmove, OnTheMarket, OpenRent, PrimeLocation, SpareRoom and Leaders. SpareRoom lists rooms, so it only feeds room figures, never the whole-home medians.

We aim to search every city on every site at least once a week, and the busiest 30 cities more often: 10 of them three times a week and 20 twice. Searches cover the town as each site defines it; we don’t add a radius around it.

Why there’s no direct Zoopla scraper. Zoopla’s site blocked our scraper, so we stopped reading it in March 2026. PrimeLocation belongs to the same group as Zoopla and carries listings from many of the same agents, so it is our route to that stock. We only read PrimeLocation’s newest listings (those added in the last 24 hours), though, so our figures don’t contain everything Zoopla shows.

The same home is often advertised on several sites. Our alerts match listings across sites by address, postcode area, price and bedrooms, so you hear about a home once.

The rent statistics don’t do that matching yet. Each site’s advert is counted, so a flat listed on two sites counts twice in the listing count and in the medians. Within one site a listing is counted once, however often we see it. Our September 2026 report said these figures were de-duplicated across sites; that was wrong, and the report now says so.

The ONS publishes its rent figures for local authorities and wider areas, not for every town. We match each town to the ONS area that covers it. When that area is bigger than the town, the page names it, so a county figure is never passed off as a town’s own: the Edinburgh page, for example, shows the figure for Lothian.

• In Scotland the ONS uses Broad Rental Market Areas, which are usually wider than the town.
• Northern Ireland has no ONS rent series for the towns we cover, so those pages have no ONS figure.
• The share-of-pay figure divides the ONS rent by median gross pay for people living in the area (ONS Annual Survey of Hours and Earnings, 2025). It isn’t available for Scottish or Northern Irish towns.

• Student lets. Some student houses are advertised per person per week. Our floors catch the cheapest of them, not all, which is why a town like Exeter can produce an overall figure that fails our checks.
• A sample, not a census. We read a fixed number of result pages per site in big cities, and only new listings on one site.
• Duplicates across sites are counted (see above).
• Bedroom counts we can’t read are stored as zero, which is why studio figures are hidden.
• Scotland and Northern Ireland. Police crime data (data.police.uk) doesn’t cover Scotland. Council tax bands and energy ratings cover England and Wales only, and the schools list covers England only. Northern Ireland has no ONS rent series, and broadband figures aren’t mapped for its towns.

• Listings are collected around the clock, on the schedule above.
• Every city’s figures are rebuilt after each 15 search runs (roughly every one to two hours) and again every night. They describe the current calendar month, and a month’s final figures are the ones from its last day.
• The 14-day removal check runs once a day.
• ONS figures are refreshed monthly, when the ONS publishes.
• City pages refresh at least daily; the UK rental index is rebuilt on every visit.

You can reuse our published city tables under CC BY 4.0: credit the Dwellio UK Asking-Rent Index and link to dwellio.co.uk. The September 2026 table is at uk-asking-rent-2026-09.csv. Its listing counts predate the 14 October 2026 change, and it doesn’t apply the checks above.

Each table on a city page and on the UK rental index has a “Cite this” line with its title, link and date, and a button that copies it.

If you’re writing about rents and want to check a figure or ask for a different cut of the data, email [email protected] or use our contact form, and say it’s a press enquiry.