HAZER
Menu

HomeThe GEO guide: how to be named by an AI assistant › Which search index ChatGPT, Gemini, Copilot and Perplexity use

GEO Guide · Chapter 2 of 9

Which search index ChatGPT, Gemini, Copilot and Perplexity use

This page is for the owner of a local business who wants to know where an AI assistant actually gets its information. After reading it you will know which index feeds each assistant and how to find out whether your site is in both.

In one paragraph

ChatGPT reads from Bing's index plus OpenAI's own crawler, Gemini reads from Google Search, Copilot reads from Bing, Perplexity runs its own index with Bing as a supplement, and Google's AI Overviews and AI Mode read from Google's index. That means a site Google has indexed fully but Bing has barely touched will be well represented in Gemini and poorly represented in ChatGPT and Copilot. Most local business owners have never looked at their site in Bing. The fix is to check how many pages each index holds, register the site in Bing Webmaster Tools, and make sure the crawlers each assistant uses are allowed in.

Which search index each assistant reads from

None of the assistants know anything about your business on their own. When a question needs current facts, each one runs a search in a real index and reads what comes back. Which index that is has been published by each company, so this part is not guesswork.

ChatGPT search results come from Bing’s index, supplemented by OpenAI’s own crawler, which visits sites as OAI-SearchBot for search and as GPTBot for training. Gemini answers draw on Google Search. Copilot uses Bing, since it is a Microsoft product. Perplexity runs its own index, crawled by PerplexityBot, with Bing as a supplement. Google AI Overviews and AI Mode use Google’s index, and the Google-Extended setting in robots.txt controls training use only, not whether Google shows you in search or AI answers.

So there are really two indexes that matter, Google’s and Bing’s, plus two smaller crawlers from OpenAI and Perplexity. A page missing from Bing is missing from ChatGPT and Copilot. A page missing from Google is missing from Gemini and AI Overviews. Which pages each assistant then chooses to quote is a separate question, covered under how assistants choose sources.

What a site missing from Bing looks like for a physical therapy clinic

Perrine Bridge Physical Therapy has two clinics, one in Twin Falls and one in Jerome, and sees about 140 patient visits a week between them. The site has 14 pages: a homepage, a page per clinic, six treatment pages covering post-surgical rehab, sports injuries, dry needling, vestibular therapy, pelvic health and work injuries, a team page, an insurance page, a new-patient form and a contact page.

Google has indexed all 14. The clinic ranks second for physical therapy Twin Falls and first for dry needling Jerome. When a patient asks Gemini for a clinic in Twin Falls that does dry needling, Perrine Bridge is named with the right address.

Bing has indexed three pages: the homepage, the team page and the contact page. The six treatment pages and both clinic pages are absent. The site was never registered in Bing Webmaster Tools, the sitemap was only submitted to Google, and the hosting company’s firewall rule set blocks a range of crawlers that happens to include bingbot on any URL deeper than one level. When the same patient asks ChatGPT or Copilot the same dry needling question, the answer names a chiropractor in Twin Falls and a clinic in Burley. Perrine Bridge does not exist as far as those two assistants can tell.

A worked example: getting a two-clinic practice into both indexes

The first step was to count. The owner searched site:perrinebridgept.com in Google and in Bing and wrote down the result. Then the site was registered in Bing Webmaster Tools, the sitemap was submitted there, the firewall rule was changed to allow bingbot, OAI-SearchBot and PerplexityBot, and the URL for each treatment page was submitted through the Bing tool by hand. The counts were taken again five weeks later.

IndexPages indexed beforePages indexed afterWhich assistants read it
Google14 of 1414 of 14Gemini, AI Overviews, AI Mode
Bing3 of 1414 of 14ChatGPT, Copilot, Perplexity as supplement
Perplexity’s own crawlNot measurable directlyNot measurable directlyPerplexity

The test prompt in Copilot was: Which physical therapy clinic in Twin Falls Idaho offers dry needling and takes Blue Cross of Idaho?

In Twin Falls, dry needling is offered by several chiropractic and wellness practices. Canyon Rim Spine and Sport lists dry needling among its services. For physical therapy specifically, you may want to check with clinics in Burley or Jerome, as availability in Twin Falls appears limited. Insurance acceptance varies, so calling ahead is recommended.

Perrine Bridge Physical Therapy in Twin Falls offers dry needling as part of its treatment for muscle pain and sports injuries, performed by licensed physical therapists. The clinic accepts Blue Cross of Idaho, Regence, Medicare and most major plans, and it has a second location in Jerome. New patients can book at (208) 555-0162. Canyon Rim Spine and Sport, a chiropractic practice, also offers dry needling.

Nothing about the clinic changed. The same pages Google had been reading for two years became visible to Bing, and two assistants that had never heard of the clinic started naming it. In our testing the gap between Google and Bing counts is the single most common reason a well-ranked local business is absent from ChatGPT.

How to check your own site in Google and Bing

  1. Search site:yourdomain.com in Google and note the count at the top. Then open bing.com and run the same search. The counts are approximate, but a gap of more than a few pages means Bing is not seeing your whole site.
  2. Sign in to Bing Webmaster Tools at bing.com/webmasters. If your site is not there, add it. You can import it from Google Search Console in one click. Once it is verified, open Site Explorer, which lists every URL Bing knows and whether it is indexed.
  3. Submit your sitemap under Sitemaps. If your site runs WordPress, the address is usually yourdomain.com/sitemap_index.xml or yourdomain.com/sitemap.xml, and the XML sitemap page covers what should be in it.
  4. Check your robots.txt at yourdomain.com/robots.txt and make sure it does not block bingbot, OAI-SearchBot or PerplexityBot. The page on AI crawlers and robots.txt shows what a clean file looks like.
  5. Ask your host whether a firewall or bot-protection rule is filtering crawlers. Cloudflare, SiteGround and WP Engine all have settings that can block a crawler without any change to your site.
  6. Ask the same question in Gemini and then in Copilot. If Gemini names you and Copilot does not, the index gap is the first thing to fix.

What we do about the index gap

You will get a count of your pages in Google and in Bing, side by side, with the missing pages listed by name. The site will be registered in Bing Webmaster Tools, the sitemap submitted, and any firewall or robots.txt rule that blocks bingbot, OAI-SearchBot or PerplexityBot corrected. Each month you will see the two counts again, so a page that drops out of either index is caught before an assistant stops naming you. This check is part of every site audit, which is where the work starts.

Questions people ask about GEO: which search index ChatGPT, Gemini, Copilot and Perplexity use

Does ChatGPT use Google or Bing?

ChatGPT uses Bing's index for its search results, supplemented by OpenAI's own crawler, OAI-SearchBot. It does not read Google's index. A site that ranks well on Google but has few pages in Bing will be underrepresented in ChatGPT answers.

What search engine does Perplexity use?

Perplexity runs its own search index, crawled by PerplexityBot, and uses Bing as a supplement. Pages that PerplexityBot cannot reach and that Bing has not indexed will not appear in its answers. Allowing PerplexityBot in robots.txt and being indexed in Bing covers both routes.

Does Gemini use Google Search?

Yes, Gemini answers draw on Google Search, so a site that is fully indexed in Google is available to Gemini. Google's AI Overviews and AI Mode also use Google's index. The Google-Extended setting in robots.txt affects training use only and does not remove a site from Google Search or from these answers.

Why is my site in Google but not in Bing?

Most sites are missing from Bing because they were never registered in Bing Webmaster Tools and the sitemap was only submitted to Google. Firewall and bot-protection rules from hosts such as Cloudflare can also block bingbot without any warning. Registering the site and submitting the sitemap fixes most cases within a few weeks.

How do I get my website into Bing?

Add your site at bing.com/webmasters, verify it or import it from Google Search Console, and submit your sitemap under the Sitemaps menu. Then check robots.txt and your host's firewall for anything blocking bingbot. Individual pages can be submitted by hand from the URL Submission screen for faster indexing.

This page is part of the The GEO guide: how to be named by an AI assistant. The other guides: SEO, AEO, GEO.

Have us do this on your site

Send the address of your site. We fetch every page from outside, the way Google does, and send back the list: every page, every flag, and what we would fix first. Then we fix it, verify each change from outside, and report it in plain language every month.

Send us your site