WebScore Guru · Visual explainer · 31 August 2026

Why can't AI assistants see the services on my website?

Because assistants like ChatGPT, Perplexity and Copilot don't know your business the way a person does. They retrieve and quote individual pages — and anything they can't reach or read might as well not exist.

GOOGLEBOT · GPTBOT · CLAUDEBOT · PERPLEXITYBOT · BING / COPILOT

When someone asks an AI assistant to recommend a plumber in Leeds or an accountant who understands farms, the assistant does not answer from memory. Its crawler programs visit web pages, extract the readable text, and store what they find. When the question arrives, the assistant searches that store, picks the pages that answer it, and quotes them.

That is why ranking well on Google is no longer the whole job. Google can rank your homepage highly while an assistant's crawler never reaches the page that actually lists your services, your prices or your phone number. If that page can't be fetched, or its words can't be extracted, you are invisible to that question — no matter how good the page looks to a human.

Industry research suggests that most UK small businesses currently have little or no measurable visibility in AI answers. That is worth checking on your own site rather than assuming either way. The three failure points below are the ones we see most often, and all three are fixable in an afternoon.

The chain from crawler to citation

Five steps sit between your website and a quoted recommendation. A break at any one of them ends the journey.

  1. Crawl

    Googlebot, GPTBot, ClaudeBot, PerplexityBot and Bing's crawler follow links around the web and discover your pages. A page nothing links to is rarely found at all.

  2. Fetch

    The bot requests the page itself. A login screen, a block rule or a broken link ends the visit here — the content behind it is never seen.

  3. Parse

    The bot extracts readable text from what it fetched. Words baked into a photograph or a scanned image do not extract; the bot has the picture but not a single word on it.

  4. Retrieve

    When a real person asks a question, the assistant searches what it has gathered and selects the pages whose text answers that question.

  5. Cite

    Only a retrieved page can be quoted or linked in the answer. Everything else — however good it looks in a browser — is invisible to that question.

The same services page, built two ways

Both pages below show identical information to a human visitor. To a crawler they are completely different documents.

Our services & prices

[ one image file:
photograph of a printed price list ]

A crawler sees: an image, and no words it can quote.

Our services & prices

  • Boiler serviceprice in text
  • Annual safety checkprice in text
  • Emergency call-outhours in text
  • Call usphone number in text

A crawler sees: text it can fetch, store and quote.

Same business, same prices. To a machine, the first page is a blank sheet of paper; the second is a menu it can read aloud to a customer.

What a machine can use, by content type

Whether four common ways of publishing business details survive the crawl-to-citation chain
Where the detail lives Can a bot fetch it? Can an assistant quote it?
Plain text on a linked page Yes Yes — this is what gets cited
Words inside an image or photo The file, not the words No
Details in an unlinked PDF Nothing links to it No
Pages behind a login Blocked at sign-in No

The three places the chain snaps

These are the failure points we find most often on ordinary small business sites. Each card names the break and the fix.

BREAK POINT 1

Text trapped in images

Price lists, opening hours and service descriptions saved as photographs look tidy but extract as nothing.

Fix: retype the words as real page text; keep the photo alongside if you like.

BREAK POINT 2

The unlinked PDF

A brochure or menu uploaded to the site but linked from no page is a file no crawler ever discovers.

Fix: link to it from a relevant page — better still, put the key details on the page itself.

BREAK POINT 3

The login wall

Service details inside a customer portal or members' area stop at the sign-in screen, for every bot without an account.

Fix: keep accounts private, but publish a public summary of what you do and how to contact you.

What to do this week

  1. Open your services page and try to highlight the text with your cursor. If you can't select it, it's an image — retype it as text.
  2. List every PDF on your site and check each one is linked from a page a visitor would actually land on.
  3. Move anything that describes your services, prices or contact details out from behind any login.
  4. Give each main service its own plainly written page, so a crawler can match it to a specific question.
  5. Run the free WebScore Guru check to see your site the way a machine sees it, before and after your changes.

Common questions

Does ranking on Google mean ChatGPT can recommend me?

No. Google ranking measures how your pages perform in search results. An AI assistant can only recommend you if its own crawler fetched a page containing the relevant detail as readable text. A top-ranked homepage does nothing for a services page the crawler can't read.

Which bots actually visit a small business website?

The main ones are Googlebot for Google, GPTBot for ChatGPT, ClaudeBot for Claude, PerplexityBot for Perplexity, and Bing's crawler, which feeds Copilot. Each one must independently be able to reach, fetch and parse your pages.

Do I need to rebuild my whole website?

Usually not. The common fixes are retyping image-based text, linking PDFs properly, and moving key details onto public pages. Most small business sites can be made machine-readable without changing how they look to customers.

Is this the same thing as SEO?

It overlaps, but it is not identical. SEO aims at ranking in a list of links; AI visibility aims at being quoted inside a generated answer. The foundations — reachable pages and extractable text — serve both, so fixing one almost always helps the other.

Sources

This page makes no statistical claims and cites no external figures, so no sources are listed. It explains how AI crawlers reach, read and quote web pages, based on WebScore Guru's own monitoring of how these bots treat small business websites.

If you'd like to know what the machines can actually see on your site, get in touch — the first check is free.