ChatGPT Search Optimisation

ChatGPT Search is OpenAI’s web-connected search interface, launched in late 2024 and progressively expanded throughout 2025. Across 10 industries analysed by Conductor, ChatGPT accounted for 87.4% of all AI referral traffic, making it the dominant AI search channel by a considerable margin.¹ It allows ChatGPT to retrieve current information from the web and cite sources within its responses. From an SEO perspective, it is a separate retrieval surface from Google AI Overviews, with overlapping but distinct mechanics. This page is about making your site visible when ChatGPT answers a query; it is a separate question from whether ChatGPT can do SEO as a workflow tool.

Following GPT-5.3 Instant becoming the default model in early 2026, average unique domains per answer dropped from roughly 19 to 15.² Citation behaviour has since shifted significantly across minor versions, though the evidence for the middle two rests on a small panel. Writesonic, which sells AI-visibility tooling, ran 50 prompts across three models from a single US ChatGPT Plus account on 27 April 2026, giving 150 conversations, 1,821 fan-out queries and 1,257 classified citations.³ On that sample GPT-5.4 cited brand-owned sites in 56.8% of responses, averaged 10.5 fan-out queries per prompt and used a site: operator in 40.5% of them, behaviour no prior ChatGPT model had shown. GPT-5.5 Instant, the default model for all users from May 2026,⁴ reversed part of that shift: brand-site citation fell to 47.2%, fan-out queries to 7.3 per prompt and site: usage to 12.6%. Fifty prompts on a single day is a narrow base, so read those as the direction of a change rather than as rates. GPT-5.6, which rolled out in early August 2026, moved further again and in the opposite direction on breadth. A re-run of roughly 4,000 prompts against a 2025 baseline found fan-out queries per prompt rising from 2.17 to 7.61, a site: operator appearing in 64% of them, and official and gov entering the five most common terms ChatGPT adds to a search the user never typed, displacing the reviews, free and comparison terms of a year earlier.⁵ Those figures are not continuous with the two before them: Nectiv ran its own panel against its own 2025 baseline, so 7.61 cannot be set against Writesonic’s 10.5 and 7.3 as though the four model versions formed one series. Each panel supports the direction it measured and not a rate that carries across the others. Sources cited per chat roughly doubled across the same rollout, which is why listicles and comparison pages lost around half their citation share without losing citations in absolute terms; the fan-out shift sets out that arithmetic. Citation breadth is less stable across minor model versions than initial data suggested, and the “15 domains” figure reflects the GPT-5.3 era specifically.

The same change moved citations away from third-party user-generated surfaces. Promptwatch measured Reddit’s share of ChatGPT Search citations at 3.8% from 18 July to 7 August 2026 and 0.5% across 14 to 17 August, an 86.4% relative fall inside a single day, while cautioning that a data-collection issue cannot be ruled out and the size of the drop is provisional.⁶ OpenAI has documented no retrieval change, and Reddit’s citation collapse sets out what the other tracking panels recorded. A surface you do not own can lose most of its citation share in a week with nothing about your content changing, which is the argument for treating a third-party profile as worth having for the audience it earns you directly, with any ChatGPT citation counted as upside.

How does ChatGPT Search retrieve content?

ChatGPT Search uses a combination of search providers (Microsoft Bing has historically been the underlying index, alongside OpenAI’s own crawling via OAI-SearchBot)⁷ and OpenAI’s own ranking and citation logic. When a user asks a question that benefits from current information, ChatGPT issues one or more search queries, retrieves results, reads relevant pages, and synthesises a response.

Bing supplies much of that retrieval pool, which makes Bing indexing the main route into ChatGPT Search. It is not the only one. Independent inspection of ChatGPT’s network traffic in 2026 found web results arriving through several distinct retrieval pipelines, two of which correspond to the third-party data providers Bright Data and Oxylabs rather than to Bing.⁸ One pipeline supplied the large majority of primary sources in the largest sample tested, but not all of them. Treat Bing indexing as the dominant path to citation rather than an absolute precondition for it.

Those pipeline findings are dated. OpenAI removed the response-stream field that named the retrieving pipeline around 21 July 2026, so the split between pipelines can no longer be read directly from the traffic.⁹ The distributions reported in the mid-2026 teardowns describe how retrieval was routed that month, not a stable architecture.

In June 2026 Microsoft launched Web IQ, an AI-native grounding API built on Bing’s index.¹⁰ Jordi Ribas, Microsoft’s President of Search and AI, told Search Engine Land that Web IQ “is used by ChatGPT for some of its web answers and it is used directly in Copilot for answers”.¹¹ Web IQ returns passage-level evidence objects rather than full pages, with retrieval quality evaluated against GDSAT (grounding satisfaction), a metric covering completeness, freshness, and authority. Passage-level selection is therefore the mechanism behind a substantial share of ChatGPT’s web answers: each section of a page is assessed on its own, not the page as a whole.

Citations appear inline as numbered references and as a list at the end of the response. Users can click through to source pages.

ChatGPT’s research mode goes further, decomposing a query into many sub-searches and synthesising findings across a multi-step task. This is a form of agentic search, and optimising for it means being retrievable for the underlying sub-queries rather than the single original question.

What does ChatGPT store about your page?

In most answers, ChatGPT does not open your page. It queries its retrieval pipelines, receives a title, a URL and a short snippet for each result, and writes from those three fields. Resoneo, a French SEO consultancy, captured 1,249 real ChatGPT answers in July 2026, then fetched 534 of the pages ChatGPT had cited and compared each against the snippet the index had stored for it. The stored snippet is cut just past 200 characters and taken from the start of the rendered page body, not from your meta description.⁹ The meta description still does work for results arriving via the scraped-Google pipeline, which picks it up roughly one time in three.

That snippet is anchored on the H1. Of the 463 cited pages that had one, 387 snippets contained it (83.6%), and in 77% of those the H1 was the first thing in the snippet. With a median H1 of 51 characters, that leaves around 150 characters of actual content. Whatever the template prints between the top of the body and the first paragraph spends part of the same budget: a section kicker on 29% of pages (18 characters), a publication date on 11% (25 characters), the alt text of the first image on 9% (50 characters), a site or brand name on 7% (44 characters). One page in seven had no H1 markup at all, and the snippet then starts on whichever subheading the template supplies.

The snippet is also frozen at crawl time rather than assembled per query: the same page seen under different queries returns an identical snippet, and 13% ran more than a month behind the crawl that produced them.⁹ Only OAI-SearchBot refreshes it. ChatGPT-User visits, which fetch the full page when the model decides to read one, never update what search displays, so a page can be re-read repeatedly while the stored snippet stays stale.

Two qualifications belong with these figures. Resoneo sells SEO consulting and distributes the browser extension that captured the data, and it did not test whether changing what sits above the H1 makes a page more likely to be cited.¹² The study reads observed behaviour rather than any documented design, on a system that has already moved once during the measurement window.

What does ChatGPT Search reward?

Patterns observed across cited sources suggest the model favours:

Content that directly answers the query. ChatGPT issues queries that are often more specific than the original user prompt. Pages whose headings and opening sentences map closely to those queries are retrieved more often.

Recency for time-sensitive topics. For news, product updates, and rapidly evolving subjects, ChatGPT shows a marked preference for recent content. Articles older than 12 to 18 months are less commonly cited unless they are clearly canonical references.

Recognised publishers. Domain authority and editorial reputation carry weight. Established news outlets, trade publications, and known expert bylines are cited disproportionately compared to anonymous content from low-authority domains.

Concise, well-structured passages. ChatGPT extracts shorter snippets than Google AI Overviews tends to. A page with a clear, paragraph-length answer is easier to cite than a page where the answer is distributed across multiple sections.

Original analysis or data. When multiple sources cover the same factual ground, ChatGPT often prefers the source that adds something distinct: original research, expert commentary, or first-hand experience.

What does ChatGPT Search deprioritise?

• Aggressive SEO content. Pages that are obviously optimised for keyword density or that read as machine-targeted are cited less often than naturally written content of comparable factual depth.
• Heavy ad and pop-up loads. Pages where content extraction is impeded by interstitials, pop-ups, or aggressive paywalls are harder to retrieve from cleanly.
• Pure listicles without analytical content. “Top 10 X” articles that summarise without contributing original judgment are cited less reliably than articles that take a position and defend it.

How do you optimise for ChatGPT Search?

Most of the work overlaps with general GEO and SEO best practice. Specific to ChatGPT Search:

1. Allow OAI-SearchBot in robots.txt. OpenAI’s search crawler is a separate user agent from GPTBot (which is used for training data). Blocking GPTBot does not block OAI-SearchBot, but blocking OAI-SearchBot will exclude you from ChatGPT Search citations.
2. Maintain content recency. For evergreen topics, update the dateModified schema property when you make substantive revisions. For time-sensitive topics, publish updates rather than relying on stale articles.
3. Use clear, scannable structure. ChatGPT Search extracts paragraph-length snippets cleanly when content is broken into well-headed sections.
4. Establish author authority. Visible bylines, author bio pages, and consistent attribution across publications reinforce the kind of expert recognition that ChatGPT weights.
5. Protect the run-up to your H1. The stored snippet anchors on the H1 and stops just past 200 characters, so mark up a real H1, keep kickers, dates, breadcrumbs and first-image alt text out of the space above it, and put your clearest sentence immediately after it.

How do you measure ChatGPT Search visibility?

There is no direct equivalent of Google Search Console for ChatGPT Search. Measurement requires:

• Manual sampling. Periodically query ChatGPT for representative searches in your niche and check which sources are cited.
• Referrer tracking. Some clicks from ChatGPT Search appear in analytics with chat.openai.com or chatgpt.com as the referrer. The volume is typically small relative to organic search but is growing. As of May 2026, Google Analytics 4 added an ‘AI Assistant’ default channel group, with Google’s documentation explicitly naming ChatGPT, Gemini, Claude, Deepseek, Copilot, and Grok as covered platforms.¹³ Perplexity is captured automatically as well, confirmed by Google Analytics in a LinkedIn post in June 2026 though not yet added to the written documentation.¹⁴ Google’s custom channel group guide covers how to define a channel for any platform still landing in Referral.¹⁵
• Brand mention monitoring. Tools like Mention, Brandwatch, or Google Alerts can pick up brand references in cited content.

The data is fragmentary. Treat ChatGPT Search visibility as a directional indicator rather than a precisely measurable channel.

Frequently asked questions

Is ChatGPT Search the same as ChatGPT plus web browsing?

Effectively yes, in current implementations. The “search” capability and the older “browse” capability have converged into a single web-retrieval feature available across ChatGPT tiers.

Does ChatGPT Search use Google’s index?

Not as its own infrastructure, but Google results do reach it indirectly. The underlying search infrastructure has been Bing-based, with OpenAI also operating its own crawler (OAI-SearchBot) for direct retrieval and indexing. Alongside those, the 2026 traffic teardowns found results arriving through third-party data providers, and one of those pipelines scrapes Google: it is the only route on which a meta description is used, which it is for roughly a third of the results it supplies.⁹ So Bing indexing remains the main path in, and being absent from Google is not a clean way to stay out of ChatGPT Search.

Do citations from ChatGPT Search affect Google rankings?

Indirectly, at most. Being cited in ChatGPT Search does not directly influence Google’s ranking algorithms. The brand visibility and any resulting referral traffic may have second-order effects on engagement metrics and brand search volume, both of which are part of Google’s quality signals.

Footnotes

1. 2026 AEO/GEO Benchmarks Report — Conductor ↩

2. ChatGPT Search Is Citing Fewer Sites, Data Shows — Search Engine Journal ↩

3. GPT-5.5 cites brand sites 47% of the time; GPT-5.4 did 57% — Writesonic. 50 prompts, 150 conversations, one US ChatGPT Plus account, 27 April 2026. ↩

4. GPT-5.5 Instant: smarter, clearer, and more personalized — OpenAI ↩

5. ChatGPT Tripled Its Fan-Out Queries and Looks For Authoritative Sources — Nectiv. Roughly 4,000 prompts re-run against GPT-5.6 Sol, published 13 August 2026. ↩

6. Reddit citations are dropping in ChatGPT — Promptwatch ↩

7. 87% of SearchGPT Citations Match Bing’s Top Results — Seer Interactive ↩

8. ChatGPT citations change when hidden search pipelines switch — Search Engine Land ↩

9. What ChatGPT pulls, what it shows, what it cites — Resoneo ↩ ↩² ↩³ ↩⁴

10. Announcing Microsoft Web IQ — Bing Search Blog ↩

11. Microsoft Web IQ (Microsoft names “AI platforms like Copilot and OpenAI”), with Jordi Ribas, President of Search and AI, quoted in Microsoft releases Web IQ, powered by Bing but designed for how AI-agents search — Search Engine Land ↩

12. ChatGPT’s Search Index Serves Small Sites Too, Data Shows — Search Engine Journal ↩

13. What’s new in Google Analytics: AI Assistant channel — Google Analytics ↩

14. Google Analytics AI Assistant traffic measurement roll-out — Google Analytics LinkedIn ↩

15. Custom channel groups — Google Analytics ↩