Ranking on Google but Not in ChatGPT

A hypothetical owner sees her service page in Google's top three for her service and city, asks ChatGPT the same question, and gets three competitors. If you are ranking on Google but not in ChatGPT, the ranking proves Googlebot can read and rank your page; ChatGPT search uses its own crawler, OAI-SearchBot, and partner search providers. OpenAI says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, although they can still appear as navigational links.

Google AI Overviews follow a different eligibility rule: the page must be indexed and eligible to appear in Google Search with a snippet, and the site must be included in Search generative AI features. Before treating an absence as a selection issue, check nosnippet, max-snippet and data-nosnippet controls and the Search generative AI setting in Search Console. Each assistant retrieves evidence through its own crawler and index; the guide to SEO for AI search and answer engines explains the wider citation mechanism and crawler settings.

The order of the six checks is my judgment, not a measured frequency: rule out the fast access, rendering and indexing failures first, then inspect the evidence a buyer's question asks the assistant to use.

Is your site allowing AI search crawlers?

AI search crawlers need access to your site before they can retrieve its pages. A Google ranking does not establish that OAI-SearchBot, Claude-SearchBot or PerplexityBot has the same access as Googlebot.

Symptom: ChatGPT does not cite or mention the company, even when the relevant page ranks in Google. The same page may also be absent from answers in Claude or Perplexity.

Cause: A robots.txt rule can allow familiar search crawlers while denying everyone else. This hypothetical file gives Googlebot and Bingbot their own permissions, then disallows all crawlers without a more specific matching group:

User-agent: Googlebot
Allow: /

User-agent: Bingbot
Allow: /

User-agent: *
Disallow: /

OAI-SearchBot, Claude-SearchBot and PerplexityBot fall under the wildcard group in that example. OAI-SearchBot surfaces sites in ChatGPT search, while GPTBot is OpenAI's separately controlled training crawler. Anthropic says blocking Claude-SearchBot may reduce a site's visibility and accuracy in search results; ClaudeBot is its training crawler. PerplexityBot surfaces sites in Perplexity search.

OpenAI tells site owners to confirm that their host or CDN allows its published searchbot IP addresses. Cloudflare's bot controls let an owner block the Search class of AI crawlers; the page, last updated 1 July 2026, said Cloudflare would set new defaults for new domains from 15 September 2026, blocking Training and Agent bots on pages that display ads while Search remains allowed. The page does not confirm that the change took effect, so check your own zone's AI bot policy.

Check: Open your live robots.txt, such as https://example.com/robots.txt, and inspect both named groups and the wildcard. A CDN can prepend directives that do not appear in your repository. Then inspect your CDN's AI bot settings and firewall events, and grep your server access log for each of the three search bot names.

For a labeled illustration, our own site's nginx access logs from 18 October to 1 November 2026 contained 630 lines matching OAI-SearchBot, 283 matching PerplexityBot and 160 matching Claude-SearchBot. Those are user-agent string matches, not IP-verified visits; user-agent strings can be faked. A zero count is a reason to investigate access, not proof on its own that a firewall blocked a crawler.

Fix: Allow the search crawlers in the applicable robots.txt groups and resolve any CDN or firewall block. Training crawlers can remain blocked independently. OpenAI says its systems can take about 24 hours to adjust after a robots.txt change.

The page needs JavaScript to show its content

JavaScript-dependent content can rank in Google while arriving as an empty shell to another crawler. The deciding check is whether the sentence a buyer needs exists in the HTML returned before browser scripts run.

Symptom: You can read the service details in a browser and the page ranks in Google, yet ChatGPT or Claude cannot retrieve those details from the page. A polished browser view does not tell you what the initial response contains.

Cause: Google says it can process content within JavaScript when the content is not blocked. Vercel's crawler data published on 17 December 2024 found that none of the major AI crawlers rendered JavaScript: ChatGPT and Claude crawlers fetched JavaScript files without executing them, while Gemini used Googlebot's infrastructure and rendered JavaScript.

Check: Fetch the page's plain HTML with curl or open view-source and search for a sentence visible in the browser; the full JavaScript rendering test shows how to examine the result.

Fix: Serve the service content in the initial HTML, so a crawler does not have to execute JavaScript to read it. Check the actual sentence after the change, since a page can return an HTML document while still leaving its useful content to a script.

The ranking page is missing from Bing's index

Bing index coverage gives a ranking page a route into Microsoft Copilot and may give it another route into ChatGPT search. Google index coverage does not establish that Bing has indexed the same URL.

Symptom: The service page appears in Google, but Bing Webmaster Tools reports that the URL is not indexed. You may also find the company missing from answers that depend on Microsoft search.

Cause: OpenAI's search help page names no search provider in its text but links the Microsoft and Shopify privacy policies for information about how search providers may process queries. Bing Webmaster Tools also reports AI citations for Microsoft Copilot. My inference is that a page absent from Bing's index loses its indexed route into Copilot and one possible route into ChatGPT search; OpenAI does not say how much ChatGPT search runs through Microsoft.

Check: Open URL Inspection in Bing Webmaster Tools and inspect the exact page that ranks in Google. Use the canonical service URL, including the correct path, rather than testing only the home page. If you already have a verified Google Search Console property, you can import it into Bing Webmaster Tools.

Fix: Verify the site in Bing Webmaster Tools, submit its sitemap and address whatever URL Inspection reports. Then inspect the same URL again to see whether Bing reports it as indexed.

The company is absent from sources assistants cite

Third-party pages give assistants evidence about businesses beyond the businesses' own service pages. For an unbranded question such as “best commercial cleaning company in Austin,” a top-ranking service page is only one possible source among lists, directories, reviews and forum discussions.

Symptom: ChatGPT names competitors for a category question, and the cited pages mention those competitors but leave your business out. Your own page may be accessible and indexed, yet the sources selected for that answer provide no independent mention of your company.

Cause: Google's generative AI guidance says its AI features can show what people say about products and services across the web, including in blogs, videos and forum discussions. Google also cautions that seeking inauthentic mentions is less helpful than it might seem.

In a study published on 26 May 2025, Ahrefs examined 75,000 brands and found a Spearman correlation of 0.664 between branded web mentions and brand mentions in Google AI Overviews, compared with 0.218 for backlinks. Those figures describe correlation with AI Overviews, not proof that mentions cause inclusion or that ChatGPT uses the same signals.

Check: Run the category prompt with search on, open every cited source and count how many name your company versus each named competitor. Use the AI visibility prompt test to run the prompt and record the cited sources consistently.

Fix: Pursue appropriate listings and genuine reviews on the pages the answer actually cited. Keep your Google Business Profile and Merchant Center information current as well; Google names both as products that can contribute to visibility in AI responses and other Search results. A cited directory that does not cover your service or location is a poor target, even if it appears in one answer.

The requested fact is missing from the page text

A service page cannot supply a plain-text answer to a question if the relevant fact exists only in a picture or an interactive control. The missing fact may be the price range, service area, intended customer or turnaround time.

Symptom: An assistant finds the business but omits a detail that a buyer can see on the page. For example, a service-area map may show the towns you cover while the page's HTML contains none of their names.

Cause: The price may live in an image or PDF, the service area in a map, and the turnaround behind a quote form or calculator widget. A crawler extracting the page's HTML may have no sentence containing the requested fact. Google says there is no requirement to break content into tiny pieces or write in a special way for AI search; the failure here is the missing plain-text statement, not the length of the page.

Check: Search the view-source HTML for the price range, the town names and the category word your buyer uses. Compare what you find with the visible page and the question the assistant failed to answer.

Fix: State each applicable fact once in a plain sentence on the ranking page. For a hypothetical company, “We clean offices in Austin and Round Rock” gives a reader and a crawler town names to work with; use your real service area and only prices or turnaround times you can substantiate.

Business facts disagree across sources

Conflicting business facts can make it harder for an assistant to resolve which company a source describes. A mismatch may lead an answer to hedge, repeat an old detail or merge your company with another business that has the same name.

Symptom: An assistant names your business but gives an old phone number, an incorrect address or a service you no longer provide. The answer may cite a directory that still carries the outdated information.

Cause: The website, Google Business Profile, Bing Places, Yelp and another directory can each carry a different version of the name, address, phone number or services. A correct service page does not remove the conflicting statements from those other sources.

Check: Put those four facts from all five sources into one table, using one row per source. Any mismatch is the finding: record the exact wording or number and the page where it appears, so the discrepancy can be checked against the current business record.

Fix: Use the order for correcting facts AI gets wrong to decide which sources to address first. Recheck the same five sources after updates, since a change on your own site does not automatically establish that a directory has changed its entry.

If crawler access, server HTML and Bing indexing pass but the company is still absent, Iron Goo's AI search visibility service covers third-party presence and consistency of business facts. OpenAI says placement in ChatGPT search is not guaranteed.

Triage the six failure modes

The fastest triage starts with checks that produce a visible access, HTML or index finding. Use the observed result to choose the first fix.

Work from the top of the table down: the first three are yes-or-no checks that take minutes, and a failure there can make later fixes pointless for the affected retrieval route. After a fix, run the same prompt more than once, because answers can vary between runs.