Every November the trade press reports the headline from the Regulator of Social Housing's Tenant Satisfaction Measures release, and then the file goes quiet. Which is a shame, because the interesting part isn't in the headline. The full 2024/25 data set records, for each landlord, how many completed responses came in by each collection method and the TP01 result for each of those methods separately.
That means the file can answer a question the summary can't: does the way you collect a tenant perception survey change the number you report?
It does. Here is what the 2024/25 data says, with the method set out at the end so you can run it yourself.
What is in the 2024/25 file
The low cost rental accommodation (LCRA) perception data set covers 353 large landlords — 195 housing associations and 158 council landlords — reporting 464,280 completed perception interviews for the survey year to 31 March 2025.
Telephone dominates. 288,040 completes (62.0%) were collected by telephone, and 242 of the 353 landlords used telephone as their predominant collection method. Of the 353, 265 drew a sample and 88 surveyed their whole tenant base as a census.
On the compliance picture the file is worth reading plainly, because it is often assumed otherwise. 349 of the 353 landed at least the sample their accuracy band required, and only 2 reported an achieved margin of error outside their band. The regulator sets no minimum response rate — what binds is the achieved margin of error at 95% confidence — and on that test the sector came in broadly compliant.
The real pressure point is narrower than compliance. 111 landlords finished within 10% of their required minimum — a third of the sector landing close enough to the line that one weaker fieldwork month would have put them under. Of the four that did finish below their required minimum, two still came in inside their band and two did not.
TP01 by collection method, across the sector
Taking the per-method TP01 figures and weighting each by the number of completes collected that way:
| Collection method | TP01 — tenants satisfied with the overall service |
|---|---|
| Postal | 80.5% |
| Face to face | 79.9% |
| Telephone | 71.3% |
| SMS | 63.2% |
| Internet / online | 61.5% |
A nineteen-point spread between the highest and lowest method. But read this table carefully, because on its own it proves nothing. Different landlords use different methods, and they have different tenants, different stock and different service histories. A landlord running a postal wave may simply serve a more satisfied tenant base than one running an online wave. This table is a mix of a mode effect and a population difference, and you cannot separate them from these rows alone.
So separate them.
The within-landlord comparison
The file lists completes and TP01 per method for the same landlord. That allows a paired comparison: for each landlord that ran two methods in the same survey year, compare their own two results against each other. Same landlord, same tenant base, same questionnaire, same year — so a population difference cannot explain the gap.
132 landlords ran both telephone and online fieldwork with at least 50 completes in each. Comparing each landlord's two figures:
- Telephone TP01 came in a median 14.9 percentage points higher than that landlord's own online TP01 (mean 14.4).
- Telephone was the higher of the two in 125 of the 132.
That is not a subtle tendency. In roughly 95% of landlords that ran both, the telephone number was higher — and the typical gap is about fifteen points on the measure the regulator publishes and tenants' representatives quote.
The practical consequence: a landlord that ran telephone fieldwork last year and shifts to online this year should expect its published TP01 to fall, by something in the order of ten to fifteen points, with no change whatsoever in the service tenants received. The reverse is equally true.
The part that complicates the story
If the telephone gap tempted you towards "telephone produces flattering numbers", the same file contradicts you.
Postal is the highest-scoring method in the data, not telephone. And the paired test holds it up: across the 31 landlords that ran both telephone and postal with at least 50 completes in each, telephone came in a median 3.3 points lower than that landlord's own postal result, and telephone was the higher of the two in only 10 of the 31.
So the ordering isn't "interviewer-led modes inflate". Postal, which has no interviewer at all, sits at the top. Face to face, which is as interviewer-led as it gets, sits just below it. Online and SMS — both self-completion, like postal — sit at the bottom.
Whatever is going on is not a simple social-desirability story, and anyone selling you one is going beyond the evidence. Plausible contributors are visible in the data's shape rather than provable from it: who each method actually reaches within a tenant base, which tenants are willing to answer an unsolicited call versus click a link, and what proportion of a small online return comes from tenants with a live complaint. The file records the results, not the mechanism.
What to do with this if you run a TSM survey
One. Treat a change of collection method as a break in your own time series. If your TP01 moves by ten points and you changed method that year, the method is the most likely explanation on this evidence, and your board deserves to hear that before it hears a story about service improvement or decline. Record your method mix alongside the number, every year.
Two. Be careful comparing yourself to a peer on a different method. Benchmarking a telephone result against another landlord's online result is not a like-for-like comparison, and the size of the distortion — around fifteen points at the median — is larger than most of the differences landlords actually argue about.
Three. Watch the mix inside a mixed-mode wave. If you collect by two methods, your headline figure is a blend, and the blend shifts whenever the split shifts. Two years with the same fieldwork design but a different telephone-to-online ratio will not produce comparable headline numbers.
Four. Judge a method on whether it lands your accuracy band, not on the score it produces. The regulator's test is the achieved margin of error against your band — ±5% under 2,500 LCRA tenants, ±4% at 2,500–9,999, ±3% at 10,000–24,999, ±2% at 25,000 and above. Choosing a method because it scores well is, apart from anything else, indefensible; choosing one because it reliably delivers the completes your band requires is the actual job. Our TSM sample size calculator gives the required achieved sample for your tenant count.
Method, so you can reproduce this
Everything above comes from one file: File 01 - LCRA_perception.csv in the Regulator of Social Housing's Tenant Satisfaction Measures 2024/25 full data set.
- Per-method sector figures. For each method, take the
TP01_Satisfied_<mode>_LCRAcolumn and weight it by the matchingResponses_<mode>_LCRAcolumn — so a landlord contributing 4,000 telephone completes counts proportionally more than one contributing 80. Weighting by completes rather than by landlord is why these figures differ from a simple average across landlords. - The paired comparisons. For each landlord, compute its own difference between two methods'
TP01_Satisfied_<mode>_LCRAvalues, and include the landlord only whereResponses_<mode>_LCRAis at least 50 in both methods. The 50-complete floor keeps out tiny cells where a handful of responses swing a percentage by twenty points. Report the median of those within-landlord differences, and count how often each method came out higher. - Totals and counts. Landlord counts, provider types, completes and sampling approach are read directly from the same file's identification and response columns. The sample and accuracy figures compare
Total_sample_size_LCRA(the achieved sample, which equals the sum of the per-method completes for every one of the 353 rows) againstMinimumSample_LCRA(the required minimum), andEstimated_margin_LCRAagainst the band set byRelevant_ten_pop_size_LCRA.
Two things worth flagging about the figures on this page. They are computed from the published per-method columns as they stand, with no imputation and no exclusions beyond the 50-complete floor. And the paired subsets are subsets: 132 landlords for telephone-versus-online, 31 for telephone-versus-postal. The telephone-versus-online result rests on a substantial share of the sector; the postal comparison rests on 31 landlords and should be read as an indication, not a settled coefficient.
When this page next changes
The 2024/25 file was published on 4 November 2025. When the regulator publishes the 2025/26 file, we will rerun every figure on this page against it, and add the obvious question this analysis raises: among landlords that switched method between years, how far did their own published TP01 move?
Source: Regulator of Social Housing, Tenant Satisfaction Measures 2024/25, full data set — gov.uk/government/statistics/tenant-satisfaction-measures-202425. All figures on this page are our own calculations from that file; any error in them is ours, not the regulator's.
Related: Tenant Satisfaction Measures, explained · TSM survey methodology: postal vs phone vs online · How to run a TSM perception survey programme · TSM reporting deadlines and submission