SEO·AEO·GEO

Guide · updated 2026-10-28

How to get your business mentioned by AI

AI assistants mention businesses they can find, read and trust. Three things decide it: whether each engine's crawler can reach your pages, whether those pages say plainly what you do, for whom and where, and whether other sites talk about you. You can check the first two in a minute. The third takes months. Here's what each engine reads, what to check, and what nobody can promise you.

Can ChatGPT, Google AI, Claude and Perplexity reach your site? Find out first.

Free score in about 20 seconds. No sign-in for your first site.

The short version

  1. Be reachable. Every engine reads the web through its own crawler, and a robots.txt rule, a firewall or a bot challenge can turn it away without you noticing. In our study of the top 5,000 sites, many sites whose robots.txt allowed AI search crawlers still refused a request sent as those crawlers, at the firewall.
  2. Be quotable. Put the facts an answer needs in the page's HTML, not behind JavaScript: what you are, who it's for, where you operate, prices and concrete numbers, and a visible date. Answer the question near the heading that asks it.
  3. Be talked about. The strongest correlates of AI citations sit off your site: press coverage, review sites, YouTube, Reddit and “best of” lists, with the same facts about you everywhere.
  4. Measure it honestly. Use the engines' own reports where they exist, ask your buyers' questions on a schedule, and treat any single answer as one sample.

Engine by engine: what each one reads

From each vendor's own crawler documentation. A crawler you block can't read you; a crawler you allow still has to get past your firewall.

Google AI Overviews and AI Mode

Reads from
Google's Search index, through its core ranking systems
Crawler to let in
Googlebot. Blocking Google-Extended doesn't remove you; Google says it “does not impact a site's inclusion in Google Search”.
Check
The page is indexed and allowed a snippet: no noindex, nosnippet or max-snippet:0. See the eligibility check.
Measure
Search Console's Generative AI report; our analyzer reads the export.

Gemini app

Reads from
Grounding with Google Search
Crawler to let in
Googlebot. Here Google-Extended does matter: it also controls grounding in Gemini Apps, so blocking it can keep you out.
Check
Everything for Google above, and Google-Extended not disallowed if you want Gemini to quote you.
Measure
No first-party report. Ask your buyers' questions in Gemini on a schedule.

ChatGPT search

Reads from
OpenAI's own search index
Crawler to let in
OAI-SearchBot. ChatGPT-User fetches a page when someone asks; GPTBot is for training only. See GPTBot vs OAI-SearchBot.
Check
robots.txt and your firewall both let OAI-SearchBot in. OpenAI says robots.txt changes take about 24 hours to apply.
Measure
Visits tagged utm_source=chatgpt.com in your analytics.

Microsoft Copilot

Reads from
Bing's index
Crawler to let in
Bingbot. As of October 2026 we found no Microsoft documentation of a separate Copilot crawler.
Check
The page is indexed in Bing; verify the site in Bing Webmaster Tools.
Measure
Bing Webmaster Tools' AI Performance report.

Perplexity

Reads from
Perplexity's own index
Crawler to let in
PerplexityBot (index). Perplexity-User fetches when someone asks and generally ignores robots.txt.
Check
PerplexityBot allowed in robots.txt and by your firewall; Perplexity publishes its IP ranges.
Measure
Referral visits from perplexity.ai.

Claude

Reads from
A search index Anthropic doesn't name publicly
Crawler to let in
Claude-SearchBot (index) and Claude-User (fetches when someone asks). Anthropic says blocking Claude-SearchBot may reduce your visibility in its answers.
Check
Claude-SearchBot allowed in robots.txt and by your firewall.
Measure
Referral visits from claude.ai.

Meta AI

Reads from
Meta's own web index
Crawler to let in
Meta-WebIndexer. Meta says allowing it “helps us cite and link to your content in Meta AI's responses”. See the new AI crawlers.
Check
Meta-WebIndexer not disallowed. Blocking Meta-ExternalAgent (training) is a separate choice.
Measure
Referral visits from meta.ai.

Siri and Apple Intelligence

Reads from
Apple's own index
Crawler to let in
Applebot. If your robots.txt doesn't name Applebot, Apple follows your Googlebot rules.
Check
Applebot (or Googlebot) allowed. Applebot-Extended only controls training.
Measure
No first-party report.

The full list of 31 AI crawlers, with their exact user agents and what each is for, is on the AI crawler list. To write robots.txt rules that allow search and refuse training, use the AI robots.txt generator.

What to put on the page

  • Match the question. In controlled studies, how closely a page matches the question asked is the most reliable lever. Write the page for the question your buyer types, and use their words in the heading.
  • State the facts plainly. One paragraph that says what you do, for whom, where, and from what price. Engines quote specifics; “industry-leading solutions” gives them nothing to quote.
  • Keep it current, really. A visible, recent date helps when the content behind it changed. Changing only the date hasn't been shown to help.
  • Say who's behind it. A named author or organisation, with the same name and links on every profile you control.
  • Skip the tricks. Formatting-only rewrites, FAQ blocks for their own sake and llms.txt files haven't been shown to earn mentions. See the GEO myths, graded.

Local business? Your Google Business Profile does much of this work: see getting a local business recommended by AI. Selling products? See getting products into ChatGPT shopping and AI Mode.

Off your site: where most mentions come from

When an assistant names one business over another, it's usually repeating what many independent sources say. Earned media, review sites, YouTube videos, Reddit threads and “best of” lists are the best-evidenced correlates of being cited. The evidence is correlational, and there's no shortcut: Google's spam systems discount inauthentic mentions, and fake reviews damage you once found. Earn coverage the slow way, and keep your name, address, prices and descriptions consistent wherever you appear. If a competitor keeps getting named instead of you, read why ChatGPT recommends your competitor and how to get ChatGPT to recommend your business.

What nobody can promise you

No tool, agency or consultant controls what an AI engine says. Answers change with the question, the person asking, the model and whether web search runs; the same prompt asked twice often cites different pages. Anyone who guarantees a ChatGPT recommendation or a fixed rise in “AI visibility” is guessing. What you can control is whether the engines can reach and read you, and what your pages and the wider web say about you. That's also all we measure: our audit scores AI readiness (page fit), not whether you're cited today.

A 30-minute plan for this week

  1. Audit your most important page and fix anything marked as a blocker first.
  2. Check every AI crawler against your robots.txt and firewall with the AI crawler checker. On Cloudflare, also run the Cloudflare AI check.
  3. Write the one paragraph that says what you do, for whom, where and from what price, and put it on your home and about pages.
  4. Make sure your Google Business Profile, main directories and social profiles say the same thing.
  5. List 10 questions your buyers ask. Put them to ChatGPT, Gemini, Claude and Perplexity, and note who gets named.
  6. Set up an AI assistant channel in GA4 so you can see the visits these answers send.

Choosing software for the rest? We compare GEO tools by what they do, including the trackers we don't make.

The detailed guides

Frequently asked questions

How long does it take to get mentioned by AI?
Access fixes apply within days: OpenAI says robots.txt changes take about 24 hours, and Google reflects them as it recrawls. Being chosen is slower, because it depends on what other sites say about you and on your rankings. Expect months, not weeks, and be wary of anyone who names a date.
Can I pay to be mentioned in ChatGPT's answers?
Not inside the answer. OpenAI says ads in ChatGPT are shown separately and don't influence the answers it gives. A service that sells a guaranteed mention is usually selling PR and content work, which you can judge on its own merits.
Do schema markup or an llms.txt file get my business mentioned?
No one has shown that they do. Google says you don't need new machine-readable files, AI text files or special markup to appear in its AI features, and a correlational study of about 300,000 domains by SE Ranking found no measurable link between llms.txt and citations. Structured data that matches the page is still worth having for ordinary search features.
Do I need a Wikipedia page?
No, and writing a promotional one usually backfires under Wikipedia's notability rules. What helps is consistent facts about your business everywhere it appears: your site, Google Business Profile, review sites, directories and a Wikidata entry if you qualify. The audit checks your Wikidata entity.
How do I know whether AI already mentions my business?
Write down 10 to 20 questions your buyers really ask, put them to each engine once a month, and note who is named and which pages are cited. Add the engines' own data where it exists: Search Console's Generative AI report and Bing's AI Performance report. Trackers automate this if you need it at scale.

Free to run. Pay only for the fixes you want.

Free

Free account

Your score without signing in; with a free account, every check's result and the full fix for the top 3 issues.

★ Starter

$9 / ₹499 once

Every issue with its exact fix for one site, PDF/Markdown/JSON exports, re-audit for 30 days. Nothing renews.

Pro

$29 / ₹1,499 / month

Full reports on every site, weekly alerts when AI access breaks, competitor compare.

Agencies: white-label widget, branded reports and lead capture on Agency.