How the figures are produced
Rent Increase Checker covers 1,796 decisions, the latest made on 19 October 2026. Read with extraction version 2026.11-L1.2.
Where the data comes from
Every decision comes from the register of residential property tribunal decisions on GOV.UK, in the category for market rents on assured tenancies. These are decisions of the First-tier Tribunal (Property Chamber) under sections 13 and 14 of the Housing Act 1988, as amended by the Renters' Rights Act 2025. Fair rent cases that are sometimes filed in the same category are left out. The register is published under the Open Government Licence v3.0.
Decisions are dated by when they were made. GOV.UK usually publishes a decision several weeks later, so the most recent weeks always look quieter than they will once the backlog is published.
How each decision is read
Each decision is read twice. First, pattern matching against the tribunal's standard wording picks out the rents, dates, property details and what the panel relied on. Second, an AI model (Anthropic's Claude Haiku 5.5, with Claude Sonnet 5.5 re-reading the harder cases) reads the whole decision and returns the same figures, each with the passage it came from. Where a decision was published as several documents, all of them are read.
The two readings are then merged. Where both find a figure and agree to within 1%, the pattern-matched figure is used. Where only the model finds one, the model's figure is used. Where they disagree, or a figure fails a sanity check, neither is published until it has been checked. Weekly rents are converted to monthly at 52/12. A figure is never guessed: if neither reading finds it, it stays blank.
Of the 1,617 rents the tribunal set, 1,567 were matched in the decision text and 50 came from the model-assisted reading. Each case page says which.
How accurate it is
The figures were checked against 60 decisions read and labelled by hand. “Right” is the share of published figures that matched the hand label; “Found” is, of the figures the decision actually states, the share the site has.
| Figure | Right | Found |
|---|---|---|
| Rent the tribunal set | 100% | 100% |
| Rent the landlord asked for | 98% | 100% |
| Rent before the notice | 100% | 96% |
| Outcome | 100% | 100% |
| Struck out or withdrawn | 100% | 100% |
| Which rules applied | 98% | 97% |
| Decision date | 92% | 92% |
| Number of bedrooms | 91% | 87% |
Sixty decisions is a small sample, so expect a small error rate across the full set. Decision dates can be a few days out where a decision states more than one date. Every case page links to the decision on GOV.UK, which is the authority.
The short summaries
The short summary on each case page is written by the same AI model from the published decision. Before it is published it is checked automatically: it must not contain the name of any individual from the decision's cover sheet or a full postcode, and every £ figure in it must appear in the decision itself. A summary that fails is not published. Summaries are not reviewed one by one by a person, so the decision is the authority.
Short-form decisions
249 of the decisions are short-form register notices rather than full statements of reasons. They record the rent the tribunal set and the date, but often not the rent before or the rent the landlord asked for, and never the reasoning. For those there is no summary, and the outcome is left blank unless all three rents are known.
Properties like yours
The property checker on the home page finds the ten past decisions closest to the property you describe. It only considers properties with the same number of bedrooms and a current rent within 40% of yours. It then ranks them: the same postcode district first, then the same postcode area, then the same tribunal region; then the closest current rent; then the most recent; with new-rules decisions slightly ahead of pre-Act ones.
Decisions under both sets of rules count, because the tribunal values the market rent in the same way under each. The checker's second answer, how often an ask is cut, uses new-rules decisions only, grouped by the size of the increase asked for.
What the outcome labels mean
Full increase: the tribunal set the rent the landlord asked for. Part of it: above the rent before, but below the ask. No increase: the same as the rent before. Rent cut: below what the tenant was already paying. Above the ask: more than the landlord asked for, which was possible before 1 May 2026 but no longer is.
Struck out or withdrawn: the application ended without a rent being set, most often because the landlord's notice was invalid or the tribunal had no jurisdiction.
Which rules applied
Which rules apply depends on when the landlord's notice was served: on or after 1 May 2026, the Renters' Rights Act rules; earlier, the pre-Act rules. So many decisions made since May still apply the old rules. Where the notice date isn't stated, a decision counts as new-rules when it refers to the Act, to section 14(A3), or to the tribunal having no power to go above the landlord's figure, or when its case reference uses the new case type.
Under the pre-Act rules the new rent could take effect from a date before the decision; those cases are marked Backdated. The new rules don't allow it.
Privacy
Decisions are public documents and name the parties. Rent Increase Checker doesn't show tenants' names, and shows individual landlords only as “Individual”. Companies, housing associations and other organisations are named where the name read from the decision is clearly an organisation; otherwise the landlord is shown as “Not stated”. The full text of decisions is not republished; each case links to GOV.UK.