Services described in accountant language
Pages titled "statutory compliance" and "management information" when the question being asked is "who files my accounts". A model matches the words the customer used, not the ones on your certificate.
Accountancy is a referral business meeting a technology that makes referrals at scale. Practices that state what they do, who for, and on which software are the ones being named.
The trigger is a specific event — incorporating, hiring a first employee, a Making Tax Digital deadline, an HMRC letter. Each one produces a question with a clear answer shape, and each is an opportunity to be the practice cited.
Specificity beats size here more than in almost any sector. "Accountants for e-commerce businesses on Xero" is quotable in a way that "full-service accountancy for businesses of all sizes" can never be, and the small practice usually has the sharper specialism.
Your qualifications are verifiable facts that most practices leave unreadable. ICAEW, ACCA, CIMA or AAT membership, and whether you are a registered auditor, are precisely what a cautious model wants to corroborate before naming you.
Get a free report on your site
The questions that matter here
These are the shapes of question an assistant answers with a shortlist of named companies. We test yours against the real models and record who gets named instead of you.
Common failures
Pages titled "statutory compliance" and "management information" when the question being asked is "who files my accounts". A model matches the words the customer used, not the ones on your certificate.
Monthly packages are the norm in this sector and are ideal for structured pricing data, yet most practices publish no number at all and lose the most common question outright.
ICAEW and Xero Partner logos in a strip at the bottom of the homepage, as images with no text equivalent, so the strongest trust signals you have are invisible.
Price
The full technical diagnostic on your site, generated by the engine and delivered in minutes. Every check scored, every finding evidenced with the response code or the markup behind it, and a prioritised fix list your developer can act on without ringing us.
If you want live testing against the models, competitor citation analysis and somebody to interpret it, that is the expert audit at £950. If you want the fixes built rather than described, we do that too. Most companies in your sector need the first one.
Most of them are a consultancy pitch with an audit attached, priced as a monthly retainer. Ours is a product: the technical diagnostic is automated, it costs £150 as a one-off, and you can request the free report without speaking to anybody. We also build the fixes, which most of them do not.
The diagnostic is the same in every industry — crawler access, structured data, rendering and content structure do not vary by sector. What varies is the questions your customers actually ask an assistant, and those we build with you. We publish what we find sector by sector, and this page lists the failures we see most often in yours.
Because this is a product rather than a consultancy pitch. The technical diagnostic is automated, so the full report is £150 once rather than a monthly retainer, and you can request the free report without speaking to anybody. We also build the fixes rather than sending a list of them. If the other three offered you a retainer and a document, that is the difference.
There is overlap, and good SEO helps. But an SEO audit tells you how you rank in a list of links. It will not tell you that your CDN is returning a 403 to Perplexity’s crawler, that your itinerary pages are empty without JavaScript, or that a model currently describes your business incorrectly. Different failure modes, different tests, different fixes.
No, and treat anyone who does with suspicion. AI answers are non-deterministic: the same question asked twice can return different sources, and model updates change results without warning. What we can commit to is the foundation — crawlers reaching your site, structured data they can parse, content they can read — and honest measurement of what changes.
It takes a domain and nothing else. If it comes back clean, you have a straight answer and you can stop thinking about this.