The Surrey Sector Index
We scanned 450 professional websites in Surrey. The profession made no difference.
Five professions, five separately built lists, all checked the way an AI assistant sees them. Five points separate the best sector from the worst.
The headline
What AI can currently see
Everyone in professional services has been told to worry about AI search. Almost nobody has checked what an AI assistant can actually read on their own website. So we checked, for five whole professions in one county, in one week, with the same tests.
One hundred of the 450 sites returned nothing readable at all. Dead domains, expired certificates, servers that never answered, and a handful of live sites that simply refuse anything which is not a browser. Those are excluded from every average below, which flatters the figures.
The finding
Five professions, five lists, the same answer
We expected differences. Law firms are larger and better resourced than accountancy practices. Architects sell on design and spend accordingly. Surveyors and structural engineers work to published technical standards. The five lists were built separately, from different sources, and scanned as separate exercises.
| Profession | Scored | Average | Median | Above 80 | Agent ready |
|---|---|---|---|---|---|
| Accountants | 81 | 51 | 56 | 2 of 81 | 12/100 |
| Solicitors | 121 | 54 | 56 | 0 of 121 | 11/100 |
| Chartered surveyors | 70 | 56 | 58 | 2 of 70 | 14/100 |
| Architects | 46 | 54 | 59 | 0 of 46 | 17/100 |
| Structural engineers | 32 | 56 | 60 | 0 of 32 | 13/100 |
Five points separate the best profession from the worst. Four websites out of 350 scored above 80. Whatever is holding these firms back, it is not their profession.
That matters more than any single sector number. If one profession had scored badly you could blame the profession. Five independent lists landing in the same place says this is simply what a professional services website looks like in 2026, and that being average is not a safe place to be.
Where the points go
The one fix that would move almost everybody
Most of what separates a readable site from an unreadable one comes down to a handful of checks. One of them appears near the top in every sector and is worth more than the rest.
Known as primary page schema. Without it an assistant has to guess what your page is about from the prose, the way a stranger skim reads a leaflet. With it, the page states plainly that it is a law firm in Guildford offering conveyancing, in a format built to be read by machines.
The other recurring gaps are security headers, which most sites are missing, and a content map for AI, which almost nobody publishes. Both are worth fixing. Neither changes as much as telling an assistant what your page actually is.
What comes next
Not one of the 350 is ready for AI to act
Everything above is about whether an assistant can read a website. There is a second question arriving quickly behind it: whether an assistant acting for a customer can do anything on that website. Fill in an enquiry form. Book an appointment. Request a quote.
Every site in all five indexes failed every check in this group. Not most of them. All of them.
- A machine readable version of the page, so an assistant reads all of you rather than a fraction
- Any way for an assistant to take an action on your behalf
- A published description of what those actions would be
- Discovery tags that tell an assistant any of the above exists
This is not a criticism of any firm. These are new standards and hardly anyone has adopted them yet, in any industry. It is worth knowing because the assistants people already use to choose a supplier are moving from reading websites to using them, and at the moment there is nothing in any of these five professions for them to use.
The five indexes
Each index is re-run after three months, so the same firms can be measured again on the same tests.
Method
How the scanning was done
Each website was checked from the outside, exactly as an AI assistant or answer engine sees it. One anonymous request per public page, no JavaScript, no logins, and nothing that is not already published. The same tests were applied to every site, and each sector was scanned in a single day.
Of the 450 websites we tried, 350 returned a readable page and are included in the figures. The remaining 100 did not, mostly domains that no longer resolve, certificates that have expired, and servers that did not answer. Those firms are excluded from every average on this page.
No business is named here and no individual score is published. Any firm in any of these indexes can ask us for its own full result, and we will send it whether or not they want anything else from us.
The honest part
What this does not tell you
A score is not a measure of whether a firm is any good, and we would not want it read that way. Several of the lowest scoring practices here are excellent businesses with long client lists who have simply never had a reason to think about this.
The five samples are not the same size. Structural engineering is a small profession locally and that index covers 32 websites, which is enough to see a pattern and not enough to publish a percentage with confidence. The architects index describes only the 46 sites of 77 that could be read at all, and those are likely to be the better maintained ones.
These tests measure what a machine can read on a single day. They do not measure whether an AI assistant actually recommends you, which depends on what is written about you elsewhere as much as on your own website. That is a separate piece of work and we will not pretend a scan answers it.
We also built the scanner, so we have an obvious interest in what it measures. The tests it applies are public standards rather than our own invention, and you are welcome to check any of them yourself. Where a test is for something new and not yet widely adopted, we have said so on this page rather than counting it against anybody quietly.
Find out where your site sits
The same scan used for these indexes is free to run on your own website. It takes about a minute and gives you the same tests, the same score, and a list of what to fix first.
In one of these indexes and want your own result? Email dev@shopstartup.co.uk with your domain and we will send it.