Evidence
We did not assume the problem existed. We asked 90 English councils, in writing, under the Freedom of Information Act 2000, and published what came back — including the parts that do not help us.
Freedom of Information requests sent, July–September 2026
substantive replies — a 71.1% response rate
able to produce a figure whose scope was consistent enough to compare
figures modelled, estimated or completed by us
A council needs two numbers to know what it costs to write to a resident: the spend and the volume. Most can produce one of them.
District, borough and city councils in England.
71.1% — courteous, inside the deadline, and mostly complete in form.
76.6% of those who replied.
51.6%. Seventeen councils knew the money but not the volume. The reverse happened once.
22.2% of all requests sent. Everything else is a bound, not a measurement.
“No such job title exists for this work. Various teams across the council are responsible for resident communications and customer contact strategy within their respective service areas.”
An English borough council, answering who owns resident communications
Expenditure sits in the finance ledger. The activity that generated it is recorded nowhere. What nobody owns, nobody counts.
What the money looks like
Median cost of one letter, print and post, across councils whose figures cover the same activity.
Core band £0.75–£1.61. Remove the scope filter and the same variable spans £0.24 to £15.01 — which is a definition problem, not a price range. We do not report that number.
What councils pay for one text message. A 69-fold range, same country, same year.
Eight of fourteen pay more than the 2.4p list price of the government’s own platform. Fifteen more report £0 — they are inside a free allowance and therefore have no unit price at all.
Median spend per resident per year on print and post.
Range £0.60 to £4.43 — a seven-fold difference that is not explained by council size (Spearman ρ = +0.099, p = 0.75).
What residents do
Median share of annual canvass responses arriving over the internet, against a median 3.7% on paper — a ratio of 24 to 1, and in one council 88 to 1.
Nine councils gave a channel breakdown. Only nine — 14.1% of those who replied could say how their residents answered.
Among households legally required to respond, the median proportion that did not answer the first communication is 50.1%.
Residents have moved to digital channels. The institution still cannot establish whether its message was received. The deficiency is not the absence of a digital channel. It is the absence of proof inside one.
On the public record
Submitted 21 September 2026 to the House of Lords Communications and Digital Committee inquiry, The Online Safety Act: implementation and impact.
Twenty-three numbered paragraphs on why the three channels institutional impersonation actually travels through — email, SMS and one-to-one calls — sit outside the Act under Schedule 1, Part 1.
What nobody owns, nobody counts — thirteen tables on what 90 councils did and did not report, including an eight-part taxonomy of the reasons a figure does not exist.
Circulated to local government and electoral administration researchers at Birmingham, the Open University and Newcastle. No council is named.
“Counting the cost of talking to residents — a question for reorganisation”, published by the Institute of Local Government Studies (INLOGOV), University of Birmingham, 8 September 2026.
Written by our founder from the same dataset.
Take it and check it
One workbook: a read-me, a full data dictionary and codebook, the observation-level data with every derived field’s formula stated, the item non-response analysis, and summary statistics as live formulas over the data. The analysis scripts and the random seed are included, so every figure reproduces.
Nothing in it is imputed, corrected or completed by us. Every blank is a blank in the source.
90 observations · scope flags on every figure · declared non-measurement codes · per-question response rates with Wilson confidence intervals · canvass, SMS and print-and-post unit costs · channel breakdowns.
Nothing. It is free, carries no conditions, and you may publish from it. Council names and exact populations are withheld so that the whole set can be used rather than a subset. We hold the unredacted version and will answer questions about coding, scope flags and exclusions.
Honesty
Councils were not selected at random. We make no national estimate and advise against one.
Rates are measured among councils that replied. Twenty-six did not, and they are unobserved.
Most councils reported net of VAT; not all said so. This is a known and unquantifiable part of the spread.
We asked institutions what they spend. We did not ask residents what happened to them.
