Page one, answer zero

My Site Ranks on Google but Not in ChatGPT or AI Answers: Finding the Missing Link

You did the SEO. You rank. Then a customer mentions that ChatGPT gave them three names for your trade in your town and none of them was yours. Google's results and an AI's shortlist are built from different indexes, different crawlers and different questions, and the gap between them usually has a findable cause.

October 9, 2026 · 9 min read

The short answer

A site can rank on Google and still be missing from ChatGPT because most AI engines do not use Google's index. ChatGPT's citations track Bing closely, Copilot runs on Bing, and Perplexity and Claude use their own crawlers. Check Bing's index, your robots.txt and firewall rules for AI search bots, and whether third-party sources mention you.

On this page
  1. Ranked by one machine, ignored by the next
  2. Google is one index; the chatbots read others
  3. Six causes, most common first, with a test for each
  4. The sequence that closes the gap
  5. Training crawler or search crawler: the robots.txt mix-up
  6. When the fixes show up in answers
  7. Keeping your place on both lists

Ranked by one machine, ignored by the next

The symptoms rarely arrive as data. They arrive as a remark from a customer or a late-night experiment. You type the question you would ask if you needed your own service, and the answer names competitors you outrank on Google every day. A few versions of the same story:

  • Your service page sits in Google's top three, but ChatGPT recommends two rivals and a directory listing.
  • Google's AI Overview cites you, while Copilot and Perplexity act as if you do not exist.
  • ChatGPT describes your company correctly when asked by name, yet never volunteers you when asked for the best option.
  • GA4 shows a trickle of visits from chatgpt.com to an old blog post, and none to the service pages that earn money.
  • Every AI engine names the same three competitors, and all three have more reviews and directory listings than you.

Each points at a different cause. The first four suggest a plumbing problem: an engine cannot read or does not index you. The last suggests a reputation problem: it can read you, but the rest of the web vouches for someone else. Our free AI visibility check shows which engines name you and who they name instead.

Google is one index; the chatbots read others

The comfortable assumption is that winning Google wins everything. It does not, because only Google's own AI features draw on Google's index. The rest lean on Bing, on their own crawlers or on partner search providers. Ranking well on Google remains a strong signal, but it is not a pass that every engine honors.

Where each AI engine finds pages, and what to check first
Where answers come fromCrawler token that mattersCheck first
[ChatGPT](/generative-engine-optimization/chatgpt)Third-party search providers and partners; a 2025 Seer Interactive test found 87% or more of its citations matched Bing's top results, against 56% for GoogleOAI-SearchBotBing index, robots.txt
[Microsoft Copilot](/generative-engine-optimization/microsoft-copilot)Bing's search index, per MicrosoftBingbotBing Webmaster Tools AI Performance report
[Perplexity](/generative-engine-optimization/perplexity)Its own index built by PerplexityBotPerplexityBotrobots.txt and firewall rules
[Claude](/generative-engine-optimization/claude)Web search that testers in 2025 found matching Brave's results, plus its own search crawlerClaude-SearchBotrobots.txt
[Gemini app](/generative-engine-optimization/gemini)Google's crawl, with the Google-Extended token controlling grounding in the Gemini appsGoogle-Extendedrobots.txt for a Google-Extended block
Google AI Overviews and AI ModeGoogle's Search index; pages must be indexed and eligible for a snippetGooglebotSearch Console's generative AI setting, nosnippet tags

Read across the rows and a pattern appears: for a business that has spent years on Google alone, Bing is often the weak link, and Bing feeds both ChatGPT and Copilot. That makes it the first place to look, not the last.

Six causes, most common first, with a test for each

1. Bing barely knows your site

Search Bing for your business name and for site:yourdomain.com. If key service pages are missing, verify the site in Bing Webmaster Tools, submit the sitemap and turn on IndexNow, which Microsoft recommends for reporting updates. Its AI Performance report then shows whether Copilot cites you at all.

2. A robots.txt line or a firewall turns AI search bots away

Open yourdomain.com/robots.txt. A Disallow under OAI-SearchBot removes you from ChatGPT search answers; one under PerplexityBot or Claude-SearchBot does similar damage there. Check your CDN too: since July 1, 2025 Cloudflare asks every new domain at sign-up whether to allow AI crawlers, and a block at the network edge overrides a permissive robots.txt. Our AI crawlers guide lists every token.

3. Your words only exist after JavaScript runs

Googlebot renders scripts; Vercel's 2024 analysis found OpenAI's, Anthropic's and Perplexity's crawlers did not. We covered the two-minute test in your site ranks on Google and is invisible to AI, so we will not repeat it here.

4. You rank for searches, not for questions

Google rewards you for "water heater repair Boerne." A person asks ChatGPT "who should I call for a leaking water heater in Boerne, and what will it cost?" If no page on your site names the town, the price range and the service in plain sentences, a page that does will be the one quoted.

5. The rest of the web does not mention you

A 2024 BrightLocal study of local searches in ChatGPT found 58% of sources were business websites, 27% third-party mentions and 15% directories. Ahrefs, studying 75,000 brands in 2025, found branded web mentions correlated far more strongly with AI Overview visibility (0.664) than backlinks did (0.218). Correlation, not proof, but a pattern. See brand mentions versus backlinks.

6. You switched Google's AI off without meaning to

If you are missing only from Google's AI features, check Settings > Search generative AI in Search Console, and look for nosnippet or a tight max-snippet in your page headers. Either one keeps a ranking page out of the summary.

The sequence that closes the gap

  1. 1

    Record a baseline

    Write ten questions a customer would actually ask, with your town in them. Run them through ChatGPT, Copilot, Perplexity, Gemini and Google, and save who gets named. Prompt tracking explains how to pick them.

  2. 2

    Open the doors

    Allow OAI-SearchBot, PerplexityBot, Claude-SearchBot, Bingbot and Googlebot in robots.txt, and confirm your CDN or security plugin is not refusing them. Training crawlers such as GPTBot are a separate decision.

  3. 3

    Get into Bing properly

    Verify in Bing Webmaster Tools, submit the sitemap, fix crawl errors and set up IndexNow so changes are reported, not waited on.

  4. 4

    Make the text readable without scripts

    If view-source shows an empty shell, the fix is server rendering. Read JavaScript and AI crawlers before you pay for anything.

  5. 5

    Write the answer pages

    One page per service and area that states what you do, where, for roughly how much, how fast and with what license or warranty, in sentences a model can lift whole.

  6. 6

    Earn the outside vouchers

    Claim Bing Places and Apple Business, complete industry directories, collect reviews steadily, and get mentioned by local press and associations. Entity optimization keeps the name, address and phone identical across them.

  7. 7

    Rerun the baseline monthly

    Compare the same questions on the same engines. Movement shows up as a trend across engines, not in any single answer.

Training crawler or search crawler: the robots.txt mix-up

The most expensive misunderstanding in this whole problem is a single line copied from a forum post. Several AI companies run two kinds of bot: one that gathers text to train models, and one that fetches pages to answer questions today. Blocking the first is a defensible policy choice. Blocking the second removes you from the answers.

Training controls (blocking does not hide you from answers)

  • GPTBot: OpenAI model training only; it does not control ChatGPT search
  • ClaudeBot: excludes future content from Anthropic's training data
  • Applebot-Extended: Apple model training; pages still appear in Apple search
  • CCBot: Common Crawl's open archive

Search controls (blocking removes you)

  • OAI-SearchBot: opt out and you are not shown in ChatGPT search answers
  • PerplexityBot: surfaces and links sites in Perplexity
  • Claude-SearchBot: Anthropic says blocking it may reduce visibility in Claude's search results
  • Google-Extended: also governs grounding in the Gemini apps, so a block there reaches beyond training

When the fixes show up in answers

Some changes travel fast. OpenAI says its search systems adjust to a robots.txt change in about 24 hours. IndexNow lets you tell Bing about a change when you publish it rather than waiting for a recrawl. Other changes depend on calendars nobody publishes: when an engine refreshes its index, when a directory pushes data, when a journalist writes the article that mentions you. Answers also vary from one run to the next, so a single good result proves little. Plumbing fixes tend to register sooner; reputation work builds over months. Measure both with the same question set.

Keeping your place on both lists

Most relapses come from routine changes nobody connected to AI: a new security plugin with a "block AI bots" toggle, a CDN migration, a redesign that moved service text into a script, a staging robots.txt that went live. Add a robots.txt and view-source check to every site launch, and rerun your question set after any infrastructure change. Our AI visibility monitoring does this on a schedule, using AICiteBeam to track which engines name you and which sources they cite.

Call for help when the crawlers are open and Bing indexes you but the answers still name others, or when the fix is a rebuild. A free GEO audit tests your questions across the engines, reads your site the way their crawlers do, and comes back with a fixed monthly quote. If the gap is mostly one rival, ChatGPT recommends my competitor goes deeper; the rest of the AI search fixes cover related problems.

Questions owners ask us about AI search

ChatGPT does not use Google's index. OpenAI says ChatGPT search uses third-party search providers and partner content, and a 2025 Seer Interactive test found most of its citations matched Bing's top results. If Bing indexes you poorly, or your robots.txt or firewall blocks OAI-SearchBot, ChatGPT may never consider your pages.

No. GPTBot gathers content for training OpenAI's models. Appearance in ChatGPT search is controlled by OAI-SearchBot, and OpenAI says sites that opt out of OAI-SearchBot are not shown in ChatGPT search answers. You can block GPTBot and allow OAI-SearchBot if you want to stay in answers without contributing to training.

It is one of the most useful free steps. Microsoft says Copilot runs on Bing's index, and independent testing links ChatGPT's citations to Bing results. Verify your site in Bing Webmaster Tools, submit your sitemap and set up IndexNow. Its AI Performance report then shows how often Copilot cites your pages.

It is most of the foundation. Google says optimizing for its AI features is still SEO, and Seer Interactive found page-one Google rankings correlated with brand mentions in GPT-4o answers for finance and SaaS queries. The gaps are usually crawler access, Bing's index, content written as answers to real questions, and how often other sites mention you. Our GEO vs SEO page covers the overlap.

OpenAI says its search systems take about 24 hours to adjust after a robots.txt change. Being crawled is not the same as being recommended, though. Whether you appear in a given answer still depends on what the engine finds about you and your competitors, so allow weeks before judging results.

They read different indexes. AI Overviews draw on Google's Search index, while Perplexity relies on its own crawler, PerplexityBot. Check that robots.txt and your firewall allow PerplexityBot, and that your key text is in the raw HTML, since Vercel's 2024 analysis found Perplexity's crawler did not run JavaScript.

Keep reading

Free · No obligation

See which engines skip you, and why

Send your site and the questions your customers ask. We will test them across ChatGPT, Copilot, Perplexity, Gemini and Google, read your site the way their crawlers do, and tell you what is in the way.

Which of these sounds like you? (pick any)

Rather talk now? Call (210) 346-0848.

CallGet a quote