SearchScore has just benchmarked 1,038 UK accountancy firms on whether AI search engines can find, understand and recommend them. One firm is AI-ready. One, out of 1,038.
But there is a different number that actually explains the first.
Zero of 1,038 firms lay their answers out in the format AI quotes from. Not a low number. Zero. For a profession that sells expertise in writing, almost none of that expertise is written in a way a machine can lift.
Where the profession sits
| Band | Score | Share of 1,038 firms |
|---|---|---|
| AI-Ready | 80 to 100 | 0.1% |
| Strong | 60 to 79 | 33.8% |
| Emerging | 40 to 59 | 50.4% |
| Low Visibility | 20 to 39 | 15.6% |
| Invisible | 0 to 19 | 0.1% |
Half the profession is in the crowded middle and a third is Strong. This is not a story about terrible websites. Most of these firms have a perfectly good site that loads fast and reads well. The average score is 53, and the ceiling for practically everyone is the Strong band.
To reach AI-Ready you need 80. Exactly one firm gets there.
Key Takeaways
- Where the profession sits
- Where the scores actually break
- The recipe already exists and almost nobody follows it
- Why nobody has noticed
Where the scores actually break
| Area | Average score out of 100 |
|---|---|
| Labelled information (structured data) | 34 |
| Brand authority | 41 |
| AI-platform readiness | 46 |
| Platform setup | 46 |
| Expertise and trust | 48 |
| Topic coverage | 70 |
| Easy for AI to quote | 73 |
| Technical basics | 75 |
Technical basics 75. Easy for AI to quote 73. Topic coverage 70. Those are fine. These firms can be crawled, they load properly, and they write about the right subjects.
Then labelled information sits at 34 and brand authority at 41. Everything a human judges is competent. Everything a machine reads is missing.
Of the firms that went to the trouble of writing a questions-and-answers section, 82% never added the tag that lets AI read it. They did the hard part. They sat down, worked out what clients actually ask, and wrote the answers. Then they skipped the thirty seconds of markup that makes the work legible to the thing doing the recommending. That is not a knowledge gap or a budget problem. It is a missing last step, repeated across four fifths of a profession.
The recipe already exists and almost nobody follows it
Eighteen firms, roughly one in sixty, do all five basics together. Those eighteen average 71, well into Strong.
So this is not mysterious. The gap between the top firms and everyone else is tidy, mechanical and entirely closeable. The leaders are not better accountants. They feed AI better.
The benchmark puts a figure on the biggest single lever: adding the standard set of data labels is worth nearly 20 points on its own, and four in ten firms still do not have them. Twenty points, from a block of markup, on a scale where the whole profession averages 53.
Why nobody has noticed
None of this gets fixed because nothing reports the loss.
When a firm slips down Google, it shows up. Someone sees the traffic dip, and there is an agency on a retainer whose job is to explain it. When an AI assistant quietly recommends the practice down the road instead, no dashboard anywhere records that it happened. There is no line item for the client you never heard from.
This is not really about accountants. A profession can be competent at the visible game and close to unquotable in the one that is quietly replacing it, and nothing in its own reporting will say so. Accountancy is just the profession that happened to get measured this time.
What I would do on Monday
- Tag the FAQ you already wrote. If you are in the 82%, this is the highest-return half hour available to you. The content exists. It just is not labelled.
- Add the standard data labels. Worth nearly 20 points on the benchmark’s own measure, and four in ten firms are still missing them.
- Put a named person, with credentials, on the pages that matter. Expertise and trust averages 48, which is the gap between having expertise and being able to prove it to a machine.
- Write one answer per page that stands on its own. Since zero firms currently do this, it is the cheapest way to be the exception.
The same shape shows up elsewhere
I ran the same reading over 9,487 UK law firms and found the identical pattern, plus a harder cause: the 964 firms with JavaScript-built websites are 76 times less likely to be visible.
Where these numbers come from
All figures above are from the SearchScore SAVI benchmark, The AI Visibility Index: UK Accountants 2026, covering 1,038 UK accountancy firms checked across more than 250 signals. I have not recalculated anything. The report is the benchmark; this is my reading of it.
The obvious disclosure: I built SearchScore, so I have an interest in you finding this convincing. The checks behind it are things you can verify by hand on any accountancy website in about ten minutes, which is the best answer I can give to that. If you want the reasoning behind the method rather than the numbers, start with my guide to generative engine optimisation.
Zero out of 1,038 accountancy firms write for AI. That is the whole problem.
