Get in touch

Tell us about your project.

Just the essentials: your name, your email, and a line about your business or project.

All articles

How does AI know about your business (and is it getting you right)?

Ask an AI for a good business in your area and it answers, often with a name. Here is how it learns who you are, and how to make sure it gets you right.

A website feeding structured data into an AI that understands the business

Ask ChatGPT, Perplexity, or Google’s AI for a good plumber (or accountant, or web designer) in your town, and you’ll get an answer, often with a name and a recommendation. So here is a question worth sitting with: when an AI talks about your business, where is it getting its information, and is it getting you right?

Your website is often one of the clearest sources it can use, but not the only one. Answer engines also read your Google Business Profile, directories, review sites, social profiles, and anywhere else your name appears, weighing them against each other. The businesses AI describes well tend to have done a few quiet things most never bother with, so every source tells the same story. Here is what those are, and how to tell whether your presence has them.

Why this matters more than it used to

Old-style search gave people ten blue links to choose from. AI search increasingly gives one answer, often a single recommendation, and most people take it. Being somewhere on page one is no longer enough; you want to be the business the AI is confident enough to name. For a machine, that confidence comes from four plain things: being able to find you, read who you are and what you do without ambiguity, check it against other sources, and describe you without hedging. How clearly your business presents its facts, everywhere it appears, has become a commercial question, not a technical footnote.

How an AI actually builds an answer about you

An AI does not keep a tidy file on your business. When someone asks about you, it gathers pages and sources, reads and cross-checks them, and stitches together an answer on the spot. It works only with what it can find and understand. If the facts that matter — what you do, where, and for whom — are buried in prose, inconsistent across the web, or missing, it fills the gaps with guesses. Guesses are where wrong hours, wrong service areas, and “I could not find much about them” come from.

An AI reading a structured-data (JSON-LD) code block

It also helps to know that not every AI reads a page the same way. Some extract the structured, machine-readable data a good site carries; others read only the visible words a person sees. Since you cannot control which is asking, the safe approach is simple: make your visible content clear, and back it with structured data that says exactly the same thing. Then it does not matter how the AI reads you — it arrives at the same answer.

Start with what everyone can read: your visible words

Before any hidden code, the most reliable thing you can do is state the obvious, in plain sight. Near the top of your key pages, say what you do, who for, and where. “We are a family-run heating and plumbing firm covering Wolverhampton and the towns around it” does more for an AI than any amount of clever markup, because every system can understand it. A surprising number of sites never quite say this — they open with a slogan or a mood and leave the machine, and the hurried human, to infer the basics. Do not make either guess.

JSON-LD: your facts, in a language machines read

Once your visible content is clear, a well-built site restates the same facts in a hidden block of code called JSON-LD, or structured data. Visitors never see it; it is written for machines. It states your facts unambiguously: name, what you do, where, hours, services, areas covered, and links to your other profiles. It will not, on its own, make an AI recommend you — anyone who promises that is overselling. What it does is quieter and genuinely useful: it removes ambiguity. Where a paragraph must be interpreted, structured data reads the same way every time.

Its most valuable job is helping an AI recognise you as one specific business, a distinct entity, rather than confusing you with a similarly named firm two towns over or a national chain. That recognition, reinforced by the links between your profiles, helps an AI settle on you as the business in your area that does the thing. Structured data does not buy confidence; it removes the reasons to withhold it.

The structured data that actually earns its place

You do not need to know how to write any of this — only that it should be there, be accurate, and match what a visitor can see. For a typical business, a handful of types do most of the work:

  • Organization or Local Business: your name, address, phone, hours, and area served — the backbone of who you are.
  • Service or Product: the specific things you offer, so an AI can match you to what someone is asking for.
  • FAQPage: your real questions and answers, mirroring a visible FAQ on the page (more below).
  • Review or AggregateRating, only where it reflects real, visible reviews on the page — easy to abuse and closely policed, so treat it with care. Genuine reputation lives mostly on third-party sites, not your own code.
  • BreadcrumbList and Article: structure for your pages and posts, so an AI understands how your site fits together.

Each is a plain statement of fact that matches the page. Together they turn your site from a wall of text a machine must interpret into a profile it can read and verify.

FAQs: answer the questions buyers actually ask

People do not query an AI in keywords. They ask questions in their own words: Who does this near me? Does this firm do that? Are they any good, and how much, roughly? A genuine FAQ answers those in plain sight, and that visible answer does the work — it reassures the unsure human and hands every AI a clear, specific response in your customer’s language. Marking the FAQs up can reinforce the structure where supported, but the value is that the answers are there, visible, and true.

Write them the way customers actually ask, not the way you describe your services. Listen to what comes in on the phone and by email: the worries, the “do you also do”, the “how soon can you”, the “what does it cost, roughly”. Marketing lines dressed up as questions help no one and impress no machine. Real answers to real questions help both.

It is not only your website: agreement across the web

An AI does not trust a single page in isolation; it cross-checks, and this is where many businesses quietly lose. It compares your Google Business Profile, social profiles, directories, review sites, and every other mention, looking for agreement. If your site says one thing and your listings say another — a different address, an old phone number, a name spelled slightly differently, hours that no longer match — that disagreement reads as doubt. And doubt is the enemy of being recommended.

So the facts on your site work best when the rest of your presence backs them up: the same name, address, and phone everywhere, links between your profiles, and listings that match. No single source is decisive — an AI triangulates, and what it rewards is agreement. You are not just telling it who you are; you are making sure every source tells the same story.

Can the page even be read?

One more quiet signal sits underneath all of this: can your pages actually be reached, loaded, and understood? If your key content only appears after heavy scripts run, the site is slow, the navigation is broken, or important pages are hard for automated visitors to reach, then everything trying to read you — search engines and answer engines alike — has less to work with and more reason to hedge or skip you. A clean, fast, well-structured site is not just pleasant for people; it gives machines fewer excuses to misunderstand you. It is the least glamorous item here and one of the most load-bearing.

What you actually get from this

  • You get the facts right — less chance of an AI quoting the wrong hours, services, or area, or saying you are closed when you are open.
  • You get found and mentioned. Clear, consistent facts are easier to surface and cite than facts buried in prose or contradicted elsewhere.
  • You get described the way you would describe yourself. When an AI can find, read, and verify you without ambiguity, it can talk about you confidently — and that makes you a safer name to put forward.

If you only do a few things, do these

If this feels like a lot, here is where the effort pays off. None of these is the single magic marker — an AI triangulates across them all — but together they cover most of what decides whether it gets you right:

  • Say plainly, in visible text on your site, what you do, who for, and where.
  • Keep your name, address, phone, hours, and services consistent everywhere they appear.
  • Keep your Google Business Profile and other listings complete, accurate, and genuinely reviewed.
  • Write real FAQs in your customers’ words.
  • Make sure your site carries accurate structured data that matches all of the above — cheap, low-risk insurance that makes the rest easier for a machine to read, not a substitute for any of it.

Where most sites fall down

Plenty of sites have no structured data at all. Many that do have the wrong kind: a generic theme that drops in a thin, boilerplate block, or one that stopped matching the page after a few edits and now contradicts it. Most have no real FAQs, or marketing lines dressed up as questions. And plenty are slow, hide content behind scripts, or say one thing on the site and another on their listings. None of this is hard to fix. It rarely gets done, because it is invisible — and invisible things get skipped, until the day a customer asks an AI about you and it shrugs.

A caveat, because honesty matters

This is not a trick or a ranking hack, and anyone selling it that way is overselling. None of it works unless it is accurate and matches what is on your page and across your listings. Faking FAQs or reviews backfires, because search engines and AI cross-check against your real content and reputation. It is necessary, not sufficient: you still need clear writing, a genuine reputation, and consistent information across the web. Making your business legible to machines helps a good business get recognised. It cannot rescue a bad one, and was never meant to.

How to tell if AI can read your business

You can check most of this yourself in a few minutes:

  • Ask the AI directly. Search your own business in ChatGPT, Perplexity, or Google’s AI. Is it right? Is it confident? Does it even find you?
  • Look for the structured data. Run your homepage through Google’s Rich Results Test or a schema validator to see what is there and what is broken.
  • Read your own pages as a stranger. Do they say, in plain visible words, what you do, who for, and where?
  • Check your FAQ. Do you have a real one, in the words customers actually use?
  • Check your listings. Does your Google Business Profile match your site — same name, address, phone, hours, and services?

If you would rather not pick through all that yourself, that is what our free site score is for. We look at what AI can actually read about your business — your visible content, structured data, FAQs, and whether your presence hangs together across the web — and tell you plainly what an answer engine can find, understand, and confidently say about you, and what is quietly holding it back. Get your free site score and find out whether AI can find, understand, and confidently recommend your business.