How AI assistants find a business
When someone asks an AI assistant for “a good plumber in Laval”, it searches the web live or uses a search engine’s index: Bing for Copilot, Google for Gemini and AI Overviews, and their own crawlers for ChatGPT (OAI-SearchBot), Claude (Claude-SearchBot) and Perplexity (PerplexityBot). It then reads the pages it found and summarizes.
There is no way to pay to be cited, and nobody can guarantee it. You can only make the site easy to find, read and quote.
The 7 steps
- Let the crawlers in: robots.txt must not block OAI-SearchBot, Claude-SearchBot, PerplexityBot, Bingbot and Googlebot.
- Text readable without JavaScript: services, city, hours and contact written in the page, not only in an image or an app.
- A short answer at the top of each page: who you are, what you do, where.
- Structured data (Schema.org LocalBusiness, FAQ, services) so machines understand without guessing.
- Google and Bing Places listings up to date, with the same name, address and phone as the site.
- Register the site in Google Search Console and Bing Webmaster Tools, and announce new pages (IndexNow for Bing).
- Get mentioned elsewhere: directories, Google reviews, associations, local media. Assistants trust what several sources confirm.
The llms.txt file
A text file at the site root that summarizes the business for language models. It is a recent proposal: useful and cheap, but no assistant commits to reading it.
How to check
Once a month, ask ChatGPT, Claude, Perplexity and Gemini the same questions (“[service] in [city]”) and note whether your business appears, and from which source.
To get a site built for you
Email info@ateliersurmesure.ca or use the quote form. The published price is in the guide how much does a website cost.
Frequently asked questions
Can you pay to appear in ChatGPT?
No. Answers cite pages found on the web; your site has to be accessible, clear and confirmed by other sources.
How long before a business appears?
Often a few weeks after the pages are indexed, but no timeline is guaranteed.
Is a Facebook page enough?
A website crawlers can read is the most reliable source; assistants often read Facebook pages poorly.