Available for projects & agency overflow · Quick reply, from the person who does the work

Send a message contact@allaux.fr
FR EN TR
allaux.
Emergency Discuss my project

Making your shop visible in AI search: Google AI Mode, ChatGPT, Perplexity

Since 22 July 2026, Google’s AI Mode and AI Overviews have been live in France, and ChatGPT and Perplexity already answer shopping questions. There is no magic tag to appear there: assistants’ crawlers must be able to read your pages, those pages must state clearly what you sell, and the shop’s data must be consistent. Here is what works, what doesn’t, and what I check on a shop.

Describe my issue Send a message

A magnifying glass and product cards link a search panel to a cart and a store.

Check what AI sees of your shop

How AI search engines find a shop

Each vendor uses separate crawlers depending on the purpose. At OpenAI, OAI-SearchBot indexes pages for ChatGPT search, GPTBot collects for model training, and ChatGPT-User opens a page when a user asks. Anthropic likewise separates ClaudeBot, Claude-SearchBot and Claude-User; Perplexity uses PerplexityBot and Perplexity-User.

At Google, AI Overviews and AI Mode rely on the regular search index, hence on Googlebot. The Google-Extended token only governs the use of your content for Gemini training and grounding: blocking it doesn’t remove your pages from search. The upshot: good Google SEO remains the foundation, and the same fundamentals serve the other assistants.

robots.txt: example, refuse training but stay citable
User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

What I check on a shop

  1. Real crawler access

    Reading robots.txt, firewall rules and access logs: which assistant crawlers come, which are refused, which pages they get.

  2. Rendering without JavaScript

    Price, availability, variants and descriptions must be in the HTML sent by the server. Content loaded only through JavaScript is often ignored by assistants’ crawlers.

  3. Structured data

    Complete Product and Offer on product pages (price, currency, availability, identifiers, shipping, returns), Organization site-wide, a consistent breadcrumb. I fix the theme or module that generates them, not just one page.

  4. Content specific to the shop

    Descriptions copied from the supplier add nothing an assistant can’t already find elsewhere. Buying guides, compatibility, usage advice and answers to frequent questions are what get a shop cited.

  5. Consistent information

    Same price on the product page, in the cart, in the Merchant Center feed and in structured data; the same shipping terms everywhere.

  6. Measurement

    A segment for visits and orders from ChatGPT, Perplexity, Gemini or Copilot in your analytics, and query tracking in Search Console, to judge on real figures.

Want to know whether your firewall blocks assistants?

llms.txt: what it does, and what it doesn’t

The llms.txt file offers language models a summary of the site and a list of links to text versions of pages. On the site you are reading, I have actually implemented it: one llms.txt per language, a catalogue of published pages, and a Markdown version of every page, generated from the CMS without ever exposing a draft.

Yet I’ll say it plainly: no major engine has committed to using it. Google has stated that none of its systems read it, and several server-log studies published in 2026 show that the big assistants’ crawlers almost never request it. Its value is limited to agents and developer tools already on a site. On a shop it replaces neither structured data nor useful content, and anyone selling it to you as a visibility lever is overselling.

Clues that assistants read a shop poorly

  • No visits from ChatGPT or Perplexity in analytics, even on best-selling products
  • Assistant crawlers getting 403 errors or challenge pages in the logs
  • Price or availability missing from the product page’s source code
  • Structured data flagged as errors in Search Console
  • Product pages word-for-word identical to other resellers’

Related pages

Make your shop readable by AI search

Frequently asked questions

How do I get my shop into ChatGPT?
Allow OAI-SearchBot, OpenAI’s search crawler, in robots.txt and at firewall level, serve product pages readable without JavaScript with complete structured data, and publish content specific to your shop. No action guarantees being cited.
Does blocking GPTBot remove me from ChatGPT?
No: GPTBot is for model training. ChatGPT search relies on OAI-SearchBot, which is set separately. You can refuse one and allow the other.
Does Google’s AI Mode need special optimisation?
It relies on Google’s search index. Best practice remains SEO best practice: fast pages, useful content, accurate structured data. Queries there are longer, which favours pages that answer a need precisely.
Should I add an llms.txt file to my shop?
It is cheap, but no effect on visibility has been shown, and Google says it doesn’t use it. I treat it as a bonus for agents, never as a priority ahead of structured data and content.
How do I know whether assistants send me customers?
By isolating in analytics the visits whose source is an assistant (ChatGPT, Perplexity, Gemini, Copilot), then following their orders. Part of that traffic arrives without an identifiable source, so these figures should be read as a minimum.

Describe your need in one minute

A few targeted questions so I can reply with an estimate rather than another questionnaire.

symptome
plateforme
catalogue (facultatif)
mesure (facultatif)
Please provide an email or a phone number so I can get back to you.

Please provide an email or a phone number so I can get back to you.

Related pages