Benchmark
How cloud, data and AI consultancies sound
Forty firms, six pages each. Nobody established, and one voice between them.
Forty cloud, data and AI transformation consultancies, read by the same code on the same day, six pages each. Not one reaches the established band. All forty fail on the same measure. Thirty-four of them use the same two words, and the sector has, on the numbers, one voice between it.
What was measured
Six pages a firm, nobody named
Fifty-two firms were listed: the global integrators, the offshore majors, the mid-size specialists and the UK boutiques that sell cloud migration, data platforms and AI transformation. Seven of the largest block automated readers and are not here; one page had gone, one timed out, and three firms had so little continuous prose across six pages that the method refused to score them, which is a finding in its own right. Forty remain.
Each was read the way a Wordprint reads a customer: the homepage and up to five of its own about, story and services pages, as one body of prose, 3,400 words for the median firm. The published method scores own words, rhythm, landing and evidence, and reports eighteen more things. No firm is named and no page text was kept.
Same code, same day, same answer next Tuesday.
The headline
A median of 56, and a ceiling of 79
The sector averages 56.8 and the median firm scores 56.5. The bottom tenth is at 40; the top tenth clears 71; the best firm measured reaches 79. Nobody is established. Five firms are in the seventies, thirty-one are in the middle band where a voice exists and is buried, and four are below 40, where the method calls the problem structural.
For comparison, UK accountancy, measured the same day on the same instrument, has a median of 72 and two firms at 93 and 100. The best AI consultancy would place in the middle of the accountancy table. This is a sector that sells the future to boards, and it writes fifteen points worse than the people who do the boards' tax.
Rhythm is not the problem: variance has a median of 8.0 and only two firms are flat. Sentences average seventeen words. Evidence is, at first sight, a strength, with the median firm putting 423 figures into every ten thousand words and the top tenth 1,761. The count cannot tell whether a figure is about the client or about the firm, and a page that is 40% faster, 60% cheaper and 3x more scalable is showing its working in a way a buyer has learned to discount. But the figures are there, and that is more than most sectors manage.
The failure
Borrowed vocabulary: forty firms out of forty
Every firm crosses it. The method calls vocabulary borrowed when more than sixty phrases per ten thousand words come from the shared phrasebook, the few hundred terms that read like a proposal template and that nobody could attribute to anybody. The median consultancy runs at 146. The top tenth runs at 273, which is one borrowed phrase every thirty-seven words, or roughly one a sentence for as long as the page lasts. Accountancy's median, for scale, is 42.
The second failure is the accountants' failure too: nothing lands. Twenty-eight of forty put fewer than one sentence in twelve at five words or fewer. The median is 5%. A sector that talks about outcomes does not, on the page, stop to state one. Not once.
And the pages themselves are more furniture than prose. Only 57% of the words on a median consultancy site sit in continuous sentences; the rest is headings, cards, bullet fragments and tiles. Accountancy runs at 76%. Three firms had so little prose across six pages that there was nothing to score, and a buyer reading them meets the same thing: a wall of labelled boxes with nobody talking.
One voice, forty logos.
The shared phrasebook
Thirty-four of forty firms use the same two words
These are the phrases used by a fifth or more of the forty firms, counted as firms rather than occurrences. In accountancy the list had eleven entries and the commonest reached twenty firms of thirty-one. Here the list is longer than the table below, which stops at fifteen, and the commonest reach thirty-four of forty.
| Phrase | Firms using it |
|---|---|
| empower | 34 of 40 |
| ecosystem | 34 of 40 |
| journey | 33 of 40 |
| excellence | 31 of 40 |
| unlock | 31 of 40 |
| seamless | 29 of 40 |
| end-to-end | 28 of 40 |
| agility | 26 of 40 |
| business value | 25 of 40 |
| leverage | 25 of 40 |
| streamline | 22 of 40 |
| robust | 21 of 40 |
| landscape | 21 of 40 |
| real-world | 20 of 40 |
| cutting-edge | 19 of 40 |
Read the top of that table again as a buyer would. Two consultancies out of any three shortlisted will promise the same thing in the same words, and the third will say it slightly differently by accident. The sector has converged on a house style that belongs to no house, and it is the style that ChatGPT, Claude, Copilot and Gemini all produce when asked for "professional copy for a technology consultancy", because this is the sector those tools have read most of. Whether any given page was written by a person or a tool the method cannot say and does not try to. It can say the result is indistinguishable either way.
Who is speaking
The firm, on forty sites out of forty
On every site measured the firm outnumbers the client: "we" and "our" at a median of 474 per ten thousand words, "you" and "your" at 163. One firm in forty lets a person speak as "I" anywhere in six pages. In accountancy it was seven in thirty-one. A sector that sells transformation to named executives has decided, unanimously and probably without noticing, that no named person will speak back to them.
One more number, offered as a measurement rather than an accusation. Long dashes, the punctuation every AI writing tool reaches for by default, appear on twenty-seven of forty sites, at a median rate of ten per ten thousand words. Accountancy's rate is under one. The method is not an AI detector and nothing claiming to be one works; a person can type a long dash, and many do. But a twelvefold difference between two sectors measured on the same day is a difference in how the pages are being made.
What the top decile does
Three habits that separate 71 from 56
- It names the work. The firms in the seventies describe a migration, a data platform, a model in production, with the client's sector and the constraint attached. The middle describes capability, which is the phrasebook by another route.
- It lets a sentence end. The top tenth lands one sentence in six. The bottom tenth lands one in a hundred and twenty.
- It has an about page with a person on it. The few sites where anybody speaks as "I" are all in the upper half, and not by coincidence: a sentence with a person behind it is hard to borrow.
None of that needs a repositioning workshop. It needs the services page read aloud with the phrasebook highlighted, and the discipline to replace each highlighted phrase with what the firm would say to a client across a table.
Limits
What this benchmark is not
It is not a ranking, and no firm is named here or in the data behind it, which stores numbers only. It is not a judgement on the work: a consultancy can score 45 here and run the best data practice in the country, and some do. Not an AI detector either. And the largest firms are absent by their own choice, so what is measured is the sector that lets itself be read.
What the number says is how hard it is for a buyer to tell one firm from the next by how it writes. In this sector, on the page, a buyer cannot. That is a diagnosis, and a cheap one to act on, because everybody else is standing still.
Measured 3 September 2026. Forty firms, six pages each, numbers only.
Where does your firm sit?
Paste a page you have published into the free test and get the same four measures in about a minute, with the sector's phrasebook marked and the sentences that are genuinely yours quoted back. A Wordprint turns the result into rules the whole firm, and the tools it drafts in, can follow. A Soundalike puts you against three named rivals on the same instrument.
Written by Ian, with Claude assisting, and scored by the smarticulate method at 98.7 out of 100 before it went up. Ian spent twenty-odd years turning thinking, his own and other people's, into something a board would act on, and built smarticulate to put numbers on it. His practice is ConsultingPartner.io: enterprise experience, right-sized. Who is he? →
More guides: the UK accountancy benchmark, how to measure brand voice, ChatGPT custom instructions for your brand voice.