On 16 August we published a piece about the Housing Delivery Test results not being published. The following day, they were.
The government released the 2024 and 2025 measurements together on 17 August 2026, alongside the new National Planning Policy Framework, and said the two had been issued at once to re-establish a regular annual cycle. The gap this publication wrote about, twenty months in which no council in England could see its own score, closed with a day to spare.
So this is the piece we could not write last week. It is based on the published data files rather than the trade coverage, because the numbers are the story and several of them are not what anyone expected.
What the test does, and what the new framework changed about it.
The test measures net homes delivered against homes required over three years. The 2025 measurement covers 2022/23, 2023/24 and 2024/25, and it is the one that now applies to decisions.
Three thresholds sit underneath it, and they survived the framework rewrite intact. Below 95%, a council must produce an action plan. Below 85%, a 20% buffer is added to the housing land supply it has to demonstrate. Below 75%, the consequence is the serious one.
What changed on 17 August is the machinery, not the number. Under the old framework a score below 75% engaged the presumption in favour of sustainable development directly, by a second route independent of land supply. Under the framework published the same day, it means an evidenced unmet need for housing is deemed to exist for the purposes of policy S5(1)(j), which is the route by which housing may be approved outside a settlement (NPPF, August 2026, Annex D, paragraph 12).
The effect is much the same and the mechanism is different, which matters to anyone drafting an objection.
England delivered 92%. That is not the finding.
Across the 301 authorities measured, England was required to deliver 772,163 homes over the three years and delivered 707,744. That is 92% of the requirement, and a national shortfall of 64,419 homes.
By the standards of the last decade, 92% is not a collapse.
And yet 93 authorities came in below 75%, up from 87 on the 2024 measurement. A further 112 are below 85%, and 147, close to half of all authorities measured, are below 95%.
So roughly a third of English planning authorities are now in the position where an unmet need for housing is treated as proven, while the country as a whole sits within touching distance of its target.
England is close to its housing target. The failure to reach it is not spread evenly, and the places where it lands lose the ability to refuse. A national average conceals exactly the thing the test was built to expose.— The Editor · Ground Level, August 2026
That is the finding. The shortfall is not spread evenly. It is concentrated, and the places it lands in lose the ability to refuse the applications that follow.
The concentration shows up at the bottom of the table. Portsmouth scored 11%. So did Castle Point. Southend-on-Sea reached 20%, Basildon and Eastbourne 23%, Gosport and Sevenoaks 25%. These are not rural districts fending off speculative fields. Several are dense urban authorities with almost nowhere left to build.
The direction of travel, authority by authority.
Because two measurements were published together, we can see movement rather than a snapshot. Three hundred authorities appear in both.
179 of them scored worse in 2025 than in 2024. 121 scored better.
Nineteen crossed the 75% threshold downwards, into the deemed unmet need. Ten climbed back out of it.
Coventry fell from 111% to 47%, a drop of 64 percentage points. Brentwood went from 162% to 110%, Hart from 181% to 137%, Cannock Chase from 157% to 115%. Most of those remain comfortably above the thresholds, which is worth saying: a steep fall from a high base is not the same as failure.
Get the dispatch.
Long-form, data-first, independent. Housing, planning and infrastructure reporting delivered free to your inbox.
Cotswold District, and a council that marked itself down.
We have reported on this district all year, so its result is worth setting out in full.
| Measurement | Published | Score | Consequence |
|---|---|---|---|
| 2023 | 12 December 2024 | 95% | 5% buffer |
| 2024 | 17 August 2026 | 85% | Action plan |
| 2025 | 17 August 2026 | 68% | Deemed unmet need |
Over the three years the district was required to deliver 1,392 homes and delivered 945.
The year-by-year figures are where it becomes stark. Against requirements of 420, then 468, then 504, delivery ran 350, then 411, then 184. Housebuilding in the district more than halved in a single year, in the year the requirement was highest.
Here is the part that reflects well on the council, and we should say so, because we were critical of the position when we described it.
In July, working from the 95% score that was then the only published figure, Cotswold District applied a 20% buffer to its housing land supply. On 95% it was required to apply 5%. Its own Topic Paper explained that it expected to fall below the 75% threshold and expected that to persist, so it had taken a precautionary approach. We wrote that it had estimated its own likely failure and imposed the consequence on itself in advance.
It was right, and it was more right than it knew. The published figure is 68%, seven points below the threshold it was bracing for.
A council that marks its own homework down against its own interest, in a document that will be tested at examination, is doing something unusual. It is also, on this evidence, more reliable than the last published score anyone could see.
The estimate that was almost exactly right.
One footnote worth recording, because we reported the estimate at the time and it deserves checking against the outcome.
In the absence of official results, the consultancy Pegasus Group published projections. It estimated that the number of authorities subject to the presumption would reach 92 on a 2025 measurement.
The published figure is 93.
That is a remarkable piece of work on incomplete data, and it says something about the publication gap. The information required to produce these numbers was available to anyone prepared to assemble it. What was missing for twenty months was not the data. It was the official act of publishing it, and with it the legal consequence.
What to watch.
The action plans. Every authority below 95% has to produce one, and 147 authorities now do. Action plans are published locally, rarely read, and are the clearest statement a council makes about why it is not building.
Whether the cycle holds. MHCLG says the two measurements were published together to re-establish an annual publication. The rule book promised results every November. That promise has been broken more often than it has been kept, and the next test of it is November 2026.
The interaction with the new framework. A council below 75% now has an evidenced unmet need deemed against it under policy S5(1)(j), while a council that adopts a plan and holds five years of supply gains a five year protection under Annex A. For the 93 authorities in the first group, the second is a long way off.