Bing Index vs Google Index: Which AI Answers Use Which

Most AI answers that cite the web draw on a stored index of pages. The question for a site owner is which index each surface reads from, because a page missing from that index cannot be cited there.

As of October 2026, the documented split looks like this:

• Google's index backs AI Overviews and AI Mode. Google says that to be eligible as a supporting link, "a page must be indexed and eligible to be shown in Google Search with a snippet." It adds that there are no additional technical requirements.
• Bing's index backs Microsoft Copilot, Bing's AI summaries, and Microsoft's Grounding with Bing Search for developers. The mechanics are covered in how Copilot grounds answers in Bing.
• Their own crawlers feed ChatGPT search (OAI-SearchBot), Perplexity (PerplexityBot) and Claude (Claude-SearchBot). None of these is documented as a copy of Google or Bing.

Eligibility is not inclusion. Google's same page says indexing and serving "isn't guaranteed." Being in an index puts you in the candidate pool. What the model then picks is retrieval and generation, which no publisher controls.

OtterlyAI ran a public test in 2026. It built two new domains about capybaras, loaded each with roughly 1,000 AI-generated posts, and tracked Google, Bing and seven AI engines for about four months. The full method is in our write-up of the Otterly 2,000 AI blogs experiment.

Google removed both sites. Otterly reports one lasted about 14 days in Google's index and the other about 90, with zero manual actions in Search Console on either. Bing never deindexed them. Otterly reports combined Bing impressions of 119,935 in July.

The AI citations followed the indexes:

• Google AI Overviews, AI Mode and Gemini almost never cited the sites.
• Of the 3,934 citations Otterly tracked for capybara-world.com, 2,718 (69%) came from Microsoft Copilot.
• That site went from 100 tracked citations in April to 1,666 in the first 27 days of July, while it was absent from Google.
• Perplexity and Claude cited the sites too.

The lesson is mechanical. An engine cannot cite what its index no longer holds. Google dropped the pages and its AI surfaces stopped using them. Bing kept them and Copilot kept citing them. If you see the same split on your own site, why Copilot cites you when Google's AI doesn't walks through the causes.

"ChatGPT uses Bing" gets repeated as a reason to chase Bing rankings. The Otterly test cuts against it. Both sites ranked in Bing, ChatGPT's crawlers kept visiting, and ChatGPT citations were effectively zero.

OpenAI's documentation explains why the shortcut is weak. ChatGPT search "sometimes partners with other search providers" and rewrites the user's query into targeted queries it sends to them. That is hedged language, and nobody has measured how much Bing feeds ChatGPT's candidate pool. More on that in whether ChatGPT uses Bing.

What OpenAI does state plainly is the access rule. Sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers." GPTBot is a separate setting that governs training only. OpenAI's help center also says the host or CDN must allow traffic from its published searchbot IP addresses. So for ChatGPT, the gate to check is OAI-SearchBot access, not your Bing position.

Two notes on the Google rows. There is no robots.txt token that blocks AI Overviews or AI Mode on its own, because Googlebot is the control. And inclusion in Google's generative features is opt-out: a Search Console property can be excluded, including by inheriting the setting from a parent property, and nothing on the public site shows it.

Otterly published its data, which is more than most GEO advice does. It is still one pair of sites, and it supports less than the headlines claim.

• One site per arm. The sites differed in more than one way, so no single variable explains the outcome.
• A low-stakes topic. Capybara questions have little competition. Copilot citing thin pages here says little about commercial categories.
• Prompt-based counts. Microsoft's own AI Performance report credited capybara-world.com with 42,817 citations, more than ten times Otterly's 3,934. The time windows may differ, but any prompt tracker is a sample.
• A vendor ran it. OtterlyAI sells AI visibility tracking, and so does 99Visibility. That is a reason to want replication, not a reason to dismiss the data.

The test also shows what not to do. Google's spam policies name generating many pages with AI tools without adding value as scaled content abuse, and the removal came with no warning. Copilot citations did not replace Google.

You don't pick one index. You check both, plus the independent crawlers, because each feeds different answers.

1. You care about Google AI Overviews or AI Mode. Confirm your key pages are indexed in Google and eligible for a snippet. Look for stray noindex, nosnippet or max-snippet:0, and check the Search Console generative AI setting on every property.
2. You care about Copilot. Verify your site in Bing Webmaster Tools and confirm the pages are in Bing's index. Its AI Performance report is the only place that shows grounding queries, the queries an AI actually ran. See whether Bing Webmaster Tools is worth it for AI search.
3. You care about ChatGPT, Perplexity or Claude. Index status in Google or Bing is not the gate. Allow OAI-SearchBot, PerplexityBot and Claude-SearchBot in robots.txt, then confirm your CDN or firewall does not return 403 to them. The AI Crawler Checker requests a page as each crawler and shows which ones get turned away.
4. You see uneven results across engines. Read each engine on its own. A blended score would have hidden the Otterly pattern entirely.

For a step-by-step routine, follow checking index status engine by engine, or start with the free index status check for Google and Bing. None of this guarantees a citation. It removes the blocks that make one impossible.