How to get your brand into ChatGPT and AI answers

How AI answers pick sources

When someone asks ChatGPT, Perplexity, Claude, Copilot or Google's AI features a question with current or specific needs, the assistant often runs a web search, reads a handful of pages, and writes an answer with citations. When it doesn't search, it answers from what the model learned in training.

That gives you two routes in: be retrievable (crawlable, indexed, clearly answering the question so it gets picked and quoted) and be known (mentioned consistently across the web so the model and the search step both associate your brand with the topic). Generative engine optimization, or GEO, is the practice of working both routes. Most of it is good SEO with a sharper focus on clarity and corroboration.

Step 1: let the right AI crawlers in

AI companies run separate crawlers for separate purposes, and you can allow or block each one in robots.txt. A common mistake is blocking everything with “AI” in its name and disappearing from AI search along with training.

If you want to appear in AI answers but not be used for training, allow the search and user crawlers and disallow the training ones. Blocking Google-Extended does not remove you from Google Search or its AI Overviews; those follow Googlebot.

User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: GPTBot
Disallow: /

Also check your CDN or firewall: bot protection often blocks these crawlers even when robots.txt allows them. The free AI visibility checker tests a page against the main AI crawlers' robots.txt rules. See AI crawler.

Step 2: add an llms.txt (cheap, optional)

llms.txt is a proposed standard (llmstxt.org) for a Markdown file at your site's root that tells language models what the site is and links its most useful pages with one-line descriptions. Think of it as a curated table of contents for machines.

Be realistic: support from the big AI search crawlers is not confirmed, and llms.txt won't rescue a site that is hard to crawl. It is still worth an hour, because coding assistants and agents that do read it get a clean map, and writing one forces you to decide which pages matter. Our llms.txt generator drafts one from your sitemap; you can see ours at /llms.txt.

Step 3: make pages machine-readable

Structured data in JSON-LD tells machines what a page is about without guessing: Organization with your official name, logo and profiles, Product with pricing, Article with author and dates, FAQPage for genuine question-and-answer content.

Google narrowed FAQ rich results in 2023 to a small set of authoritative sites and dropped HowTo rich results, so don't expect search snippets from them. The markup still helps any system that parses pages to understand them. Beyond schema:

• Server-render the important text. Content that only appears after JavaScript runs may never be seen by some crawlers.
• One clear H1, descriptive headings, and a canonical URL per page.
• Name things consistently: the same product name, the same company name, everywhere.

The SEO audit tool checks titles, descriptions, canonicals, noindex and structured data on any URL.

Step 4: write pages AI can quote

Assistants quote passages that answer a question completely in a few sentences. Structure pages so those passages exist:

• Answer first. Put a two- or three-sentence answer directly under the heading that asks the question, then the detail.
• Be specific. Numbers, formulas, named steps and dates beat adjectives. “Pacing % = spend ÷ expected spend” is quotable; “we optimize budgets intelligently” is not.
• Cover comparisons honestly. “X vs Y” and “best X for Y” questions are common in AI search. Pages that state where a competitor is better are more credible, to humans and to models.
• Publish original data. Benchmarks, surveys and teardown analyses get cited because nobody else has them.
• Keep pages current. Show an updated date and actually update the content.

This hub is built that way: every guide opens with a short answer, and every glossary term opens with a one- or two-sentence definition.

Step 5: earn the mentions AI trusts

Assistants corroborate. If the only site saying you're the best tool for something is yours, that claim carries little weight. Work on the places answers draw from:

• Review sites and directories in your category, with complete, current profiles.
• Comparison and “best of” articles by independent publishers.
• Community discussions (forums, Reddit, Q&A sites) where real users describe what they use. Participate honestly; astroturfing gets noticed.
• Partner and integration directories, conference talks, podcasts and guest research.
• Your own profiles on knowledge sources, kept factually consistent.

Several assistants also lean on Bing's index for web results, so verify your site in Bing Webmaster Tools and use IndexNow to tell search engines about new and updated pages quickly.

Step 6: measure AI visibility

1. List the prompts buyers ask.

   Twenty to fifty real questions: “best X for Y”, “X vs Y”, “how do I…”. Use sales calls and support tickets, not guesses.

2. Test them monthly.

   Ask each in ChatGPT, Perplexity, Claude and Google's AI features. Record whether you're mentioned, linked, and how you're described.

3. Track AI referrals.

   In GA4, look at sessions from referrers such as chatgpt.com, perplexity.ai and copilot.microsoft.com. They are small but growing, and they convert differently.

4. Watch crawler hits.

   Server or CDN logs show whether OAI-SearchBot, Claude-SearchBot and PerplexityBot actually fetch your pages.

5. Fix what's missing.

   No mention? Check crawl access, then whether a page clearly answers that prompt, then whether anyone else on the web says it.

Example prompts

Copy these into Claude, ChatGPT or Cursor with MegaMCP connected. Each lists the tools it typically calls; more are in the prompt library.

• Reads data

  Run an AI visibility audit on https://example.com/pricing: crawler access, llms.txt and structured data.

  Tools it calls: web_ai_visibility_audit

• Reads data

  Draft an llms.txt for https://example.com from its sitemap.

  Tools it calls: web_draft_llms_txt

• Reads data

  Audit https://example.com/blog/guide for title, meta description, canonical and structured data problems.

  Tools it calls: web_seo_audit

• Proposes a change

  Add Organization JSON-LD to our WordPress homepage and show me the plan.

  Tools it calls: wordpress_set_schema

  Any change becomes a plan you review and approve. Nothing moves until you do.

• Reads data

  Show sessions from chatgpt.com, perplexity.ai and other AI assistants in GA4 for the last 90 days, by landing page.

  Tools it calls: ga4_run_report

• Proposes a change

  Tell search engines about these five updated URLs through IndexNow.

  Tools it calls: gsc_indexnow_submit

  Any change becomes a plan you review and approve. Nothing moves until you do.