Google AI Overviews vs ChatGPT search: index vs crawler

What is the one difference that explains the others?

Where each gets its sources. AI Overviews is a feature of Google Search, built on the same index as the ordinary results. Google states there are no additional requirements to appear in AI Overviews or AI Mode: a page must be indexed and eligible to be shown in Search with a snippet to be a supporting link. If you already do SEO, your pages are already eligible.

ChatGPT search depends on OpenAI's own crawler. OpenAI documents OAI-SearchBot as the user agent used to surface websites in ChatGPT's search features, and states that sites opted out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. A page Google indexes perfectly can be invisible to ChatGPT search because of one line in robots.txt, or a firewall rule nobody connected to AI. The stored ChatGPT answer above gives the same first step: allow OAI-SearchBot, and check that a CDN is not blocking it.

What does that mean for the work?

The page work is shared and the access work is not. On both surfaces the pages that win answer the question directly, near the top, with numbers and sources inside the passage. Google adds query fan-out: it may issue multiple related searches across subtopics to build an Overview, so pages that answer the sub-questions behind a search have more chances. OpenAI does not describe ChatGPT's retrieval in that detail, but the same principle holds in practice: a page that owns one clear question is easier to cite than one that covers everything at the same depth.

The access work is all on the ChatGPT side. For AI Overviews you already have Googlebot crawling you. For ChatGPT search you have to allow OAI-SearchBot in robots.txt, make sure your CDN does not challenge it, and decide GPTBot, OpenAI's training crawler, as a separate question. OpenAI says robots.txt changes take about 24 hours to apply. Answers move more slowly than that, because a page has to be fetched again and then chosen.

How do the controls compare?

Each surface now has a clean opt-out, and they live in different places. Google has added a setting to Search Console that decides whether a site appears in, and helps ground, its generative AI features: AI Overviews, AI Mode and AI Overviews in Discover together. Google says the setting is not used as a ranking signal for ordinary results. To leave ChatGPT search, you disallow OAI-SearchBot in robots.txt. Neither control touches the other company's surface.

The finer controls exist only on Google's side. nosnippet, data-nosnippet and max-snippet limit what Google may show from a page, so you can keep one passage out of an Overview and leave the rest of the page eligible. OpenAI documents no equivalent: robots.txt works on URL paths, not on passages. And Google-Extended, the token people often reach for, governs Gemini training and grounding. Google states it does not affect inclusion in Search, so it does nothing to AI Overviews.

What did ChatGPT cite when asked about itself?

Mostly OpenAI. We track "how to get cited by ChatGPT" for our own brand. In the 30 days to 4 October 2026, ChatGPT answered it 57 times, and 1,205 of the 1,831 citations in those answers pointed at help.openai.com. Perplexity answered the same prompt 60 times and leaned on agency and marketing blogs, with no OpenAI domain among its six most influential sources.

For questions about its own behaviour, ChatGPT trusts its maker's pages. That is reasonable, and it also means no third-party guide, ours included, is likely to be cited for that question on ChatGPT. Neither engine named Attensira in any of the 117 answers, and no attensira.com page was cited in them either, which is the honest baseline for a question a vendor's own documentation already answers. The lesson generalises: for each question you care about, find out whose pages each surface already trusts before you decide what to publish.

How do large sites treat the two crawlers?

Very differently. On 3 September 2026 we read the robots.txt files of the top 10,000 domains in a dated Tranco list and parsed 8,036 (the full study). Only 156 (1.9%) disallow Googlebot at the root. 467 (5.8%) disallow OAI-SearchBot, three times as many. Few large sites would cut themselves out of Google Search on purpose; a meaningful number have cut themselves out of ChatGPT search, often through blanket rules written against AI training crawlers.

On our own site both crawlers are allowed. In the 30 days to 4 October 2026, Googlebot made 189 requests across 80 pages, and about 86% of the 7,536 requests carrying OpenAI user agents were classed as search indexing, which is OAI-SearchBot's job. OpenAI's search crawler visited far more often than Googlebot did. User agents are self-reported, so treat both counts as directional.

How can you measure each one?

Google reports appearances; OpenAI reports nothing. Search Console's generative AI performance report, rolled out to all sites by 31 August 2026, shows how often your pages appeared in AI features such as AI Overviews and AI Mode, by page, country, device and date. Clicks from those features still sit inside overall Search traffic, under the Web search type. OpenAI's crawler documentation describes no reporting surface for publishers.

There is one practical aid on the ChatGPT side. Some cited links in ChatGPT's answers carry utm_source=chatgpt.com: of the 61 ChatGPT answers we stored on 1 to 4 October 2026, 6 did. Clicks from those links can be counted in ordinary analytics. What neither surface tells you is what the answer said about you, or who it recommended instead.

That leaves reading the answers. For AI Overviews, a tool can load the results page for each tracked search every day and record whether an Overview appeared and which pages it cited, keeping "no Overview" separate from "could not see". For ChatGPT, a tool has to ask your buyers' questions on a schedule and record who is named and what is cited, with the number of answers behind every rate.

Who owns each surface inside a company?

Usually the same person, with different blind spots. The SEO owner already manages Google: indexing, snippets, Search Console. AI Overviews falls naturally into that remit, and most of the work is checking that nothing in the existing setup holds back the pages that should be quoted, now including the generative AI setting.

ChatGPT search often falls between teams. robots.txt is owned by engineering, the CDN by infrastructure, content by marketing, and nobody is reading the answers. The first job on the ChatGPT side is to name one owner who can see all four, and to give that owner a record of what ChatGPT says for the questions that matter.

How should you report the two to leadership?

As two lines, never one. For AI Overviews: the share of tracked searches that showed an Overview and the share of those that cited your pages, each with the number of days behind it, with Search Console's AI impressions beside them. For ChatGPT: the share of tracked questions where you were named and where your pages were cited, with the number of answers behind each, and utm-tagged referrals beside them. A blended "AI visibility score" hides the one thing leadership needs to know: which surface to work on next.