AI search guide
How to Rank in ChatGPT: A Practical Guide to AI Search Visibility
How ChatGPT, Perplexity, Google AI Overviews and Claude pick sources, the work that gets a business cited, and how to measure it.
By Chad Westleigh · Updated
How to rank in ChatGPT: what ranking means in an AI answer
Ranking in ChatGPT means being one of the sources an AI answer cites, or one of the brands it names, for the questions your buyers ask. There is no fixed position one. OpenAI says ChatGPT ranks search results on multiple factors and that placement isn't assured, and answers vary between runs, so visibility is a share of prompts, not a rank.
The practice goes by several names: answer engine optimization (AEO), generative engine optimization (GEO) and AI SEO. All describe the same work: make your pages easy for AI systems to fetch, quote and trust, on top of ordinary SEO foundations.
How ChatGPT, Perplexity, Google AI Overviews and Claude choose sources
From each company's own documentation, as of October 2026. None publishes a ranking formula.
| Engine | Where sources come from | Crawler to allow | What the vendor says |
|---|---|---|---|
| ChatGPT search | OpenAI's search crawler plus third-party search providers | OAI-SearchBot | Blocked sites won't be shown in ChatGPT search answers, except as navigational links |
| Google AI Overviews and AI Mode | Google's search index, often via query fan-out across subtopics | Googlebot | Pages must be indexed and snippet-eligible; no special schema or AI text file required |
| Perplexity | Its own index, plus pages fetched live for a question | PerplexityBot | Allow PerplexityBot and its published IP ranges to appear in results |
| Claude | Anthropic's search index, plus pages retrieved for a user's question | Claude-SearchBot, Claude-User | Blocking them may reduce your visibility in Claude's answers |
Query fan-out and Bing: two details that change the work
ChatGPT search and Google's AI features both split one question into several narrower searches. OpenAI describes ChatGPT rewriting a prompt into targeted queries, then sending follow-ups; Google calls it query fan-out. Pages that answer the sub-questions buyers ask, such as cost, timeline, alternatives and risks, get more chances to be retrieved.
The second detail is Bing. OpenAI doesn't name its search partners, but its help page points to Microsoft's privacy statement for how they handle queries. Verify your site in Bing Webmaster Tools, submit your sitemap and consider IndexNow, which Bing supports.
Technical access: server-side rendering comes first
AI crawlers can only cite what they can read, and the safe assumption is that they read only the HTML your server returns. Googlebot renders JavaScript; the crawler documentation from OpenAI, Anthropic and Perplexity doesn't say their bots do. If prices, service descriptions or FAQ answers appear only after client-side JavaScript runs, server-render them.
- Fetch key pages with a crawler user agent and confirm the H1, copy and structured data are in the raw HTML
- Check that your CDN or bot protection doesn't block AI crawlers; OpenAI asks sites to allow its published search-bot IP addresses
robots.txt for AI crawlers: OAI-SearchBot, GPTBot, ClaudeBot and PerplexityBot
Search crawlers decide whether you can be cited; training crawlers decide whether your content trains models. OpenAI says each of its settings is independent.
| User agent | Company | What it does | If you block it |
|---|---|---|---|
| OAI-SearchBot | OpenAI | Surfaces sites in ChatGPT search | Not shown in ChatGPT search answers, except as navigational links |
| GPTBot | OpenAI | Collects content for model training | Excluded from training; search is set separately |
| ChatGPT-User | OpenAI | Visits pages for actions a user starts | robots.txt may not apply to user-requested visits |
| Claude-SearchBot | Anthropic | Indexes content for Claude's search | May reduce visibility in Claude's search results |
| ClaudeBot | Anthropic | Collects content for possible training | Future content excluded from training |
| PerplexityBot | Perplexity | Surfaces and links sites in Perplexity; not used for training | Perplexity recommends allowing it to appear in results |
| Googlebot | Crawls for Search, including AI Overviews and AI Mode | Removed from Search and its AI features | |
| Google-Extended | Controls use in Gemini training and grounding | No effect on Google Search inclusion or ranking |
llms.txt: cheap to add, limited adoption
llms.txt is a proposal, not a standard AI search engines have committed to. Jeremy Howard published it in September 2024: a Markdown file at /llms.txt that summarizes a site for language models. None of the crawler documentation from OpenAI, Anthropic, Perplexity or Google we reviewed says it is used to choose sources, and Google says you don't need AI text files to appear in AI Overviews or AI Mode.
We publish one on operatorintelligence.ai because it's quick and harmless. Treat it as a tidy summary, not a ranking factor.
Answer-first content: write passages AI can quote
AI engines retrieve passages, not whole pages, so each section should open with a direct answer that stands on its own. The pattern we use:
- A 40 to 70 word summary at the top that answers the main question and names entities in full
- Question-shaped headings with the answer in the first sentence below
- Specific facts with sources. In the GEO paper by Aggarwal et al. (2023), adding citations, quotations and statistics raised visibility in generative answers by roughly 30 to 40% on the authors' benchmark; keyword stuffing did little.
- Tables for comparisons, since X vs Y prompts are common
- Dates on time-sensitive claims, so engines can judge freshness
- Separate pages for the sub-questions fan-out searches look for
Entity consistency: make your brand easy to describe
AI systems describe your business from everything they can find, so the same facts need to appear everywhere. Use one name, one category description and the same founders, offerings and proof on your site, LinkedIn, directory profiles and press bios. Add Organization schema with sameAs links, and Person schema for founders and authors.
Our own name is shared by unrelated companies, so we write Operator Intelligence, the AI enablement consultancy, rather than an acronym alone. When an AI answer gets a fact wrong, the cause is usually an outdated profile or page.
Third-party mentions: listicles, reviews and Reddit
For many buyer prompts, AI engines cite third-party lists, reviews and forum threads rather than brand sites. Check the citations in your own prompt panel; earning a place on those pages is much of how to get cited by ChatGPT.
What works without gaming anything:
- Listicle inclusion: send roundup authors a one-page proof sheet with named results; don't pay for placement
- Review profiles on your category's directories, with reviews from real clients
- Reddit and community answers under your own name, where you have expertise
- Podcasts, guest articles and partner pages
- Original data others want to cite
This is slow, and no one can promise a citation from it.
Structured data's role in AI search
Structured data helps engines parse a page, but it isn't a ticket into AI answers. Google says no special schema.org markup is needed for AI Overviews or AI Mode, and that structured data should match the visible text. Use Organization and Person markup for entity facts, Product markup for ecommerce and Article markup with authors and dates for guides. Our agentic commerce guide covers product data for AI shopping agents, and jewelry SEO and AI search shows a vertical version.
How to measure AI visibility with a prompt panel
Rank trackers don't show whether ChatGPT recommends you. A prompt panel does.
1. Write the prompts
25 to 50 real buyer prompts: categories, use cases, comparisons and best-X-for-Y questions from sales calls and search data.
2. Run them monthly
In ChatGPT, Perplexity, Gemini, Google AI Overviews or AI Mode, and Claude, several times each, because answers vary.
3. Log what matters
Whether you appear, how you're described, which competitors are named and which sources are cited.
4. Turn gaps into work
A missing page, a wrong fact, an outdated profile or a roundup worth pitching.
5. Report the trend
Share of prompts where you appear, by engine and topic, against your baseline.
Track AI referrals with a GA4 channel
GA4 doesn't separate AI-assistant traffic by default; it lands in Referral or Direct. Create a custom channel group (Admin, Data display, Channel groups) with an AI assistants channel matching sources such as chatgpt.com, perplexity.ai, claude.ai, gemini.google.com and copilot.microsoft.com, ordered above Referral, because GA4 assigns traffic to the first matching channel.
Caveats: some AI traffic arrives as Direct, so treat the channel as a floor; AI Overviews and AI Mode clicks count in Search Console's overall Web traffic; and influence exceeds clicks, so watch branded search.
How we rebuilt operatorintelligence.ai for AI search
We applied this guide to our own site in October 2026. Operator Intelligence moved operatorintelligence.ai from a client-rendered single-page app to server-side rendering, so crawlers such as GPTBot and ClaudeBot now get the same full HTML as a browser. Service and resource pages open with an answer-first summary and self-contained sections. Our robots.txt allows search and AI crawlers, we publish llms.txt, and we pair our name with our category everywhere.
We aren't publishing ranking or traffic results from the rebuild yet. For DTC brands, our Shopify SEO and AI search (GEO) service applies the same method, starting with an AI visibility audit.
AI visibility audit checklist
Run through this before you write anything new.
- Key pages return full content in raw HTML, without JavaScript
- robots.txt and CDN allow OAI-SearchBot, Claude-SearchBot, PerplexityBot and Googlebot
- Training crawlers are allowed or blocked on purpose
- The site is verified in Google Search Console and Bing Webmaster Tools
- Each money page opens with a direct, self-contained answer
- Brand facts match across the site and profiles, and schema matches visible text
- A baseline prompt panel and a GA4 AI assistants channel are in place
Frequently asked questions
Can you pay to rank in ChatGPT?
Not in its organic answers. OpenAI says ChatGPT ranks results on multiple factors and that placement isn't assured. ChatGPT does carry ads, whose landing pages OpenAI's OAI-AdsBot checks, but ads are separate from cited sources.
What is an AEO agency?
An AEO agency works on answer engine optimization: getting a business cited and described accurately in AI answers. A good one also fixes the technical SEO underneath and measures prompt-level visibility. Our AI search (GEO) service is our version for Shopify brands.
Does blocking GPTBot remove me from ChatGPT search?
No. OpenAI says each crawler setting is independent: GPTBot governs training and OAI-SearchBot governs ChatGPT search. You can block one and allow the other.
How long does it take to show up in ChatGPT?
OpenAI says its search systems take about 24 hours to adjust after a robots.txt change. Content fixes can be picked up within weeks of being crawled; third-party mentions take months, and no one can promise a citation.
What is an AI visibility audit?
A check of whether AI engines can read your site, what they say about you and which sources they cite instead. It combines a rendering and crawler review, an entity and schema check and a baseline prompt panel, ending in a prioritized fix list. Ask us to scope one.
Find out what AI search says about your business
We'll run your buyers' real prompts across ChatGPT, Perplexity, Gemini and Google, show where you appear and which sources get cited, and hand you a fix list. Or start with the free AI readiness assessment.
Get an AI visibility auditSources
- OpenAI, Overview of OpenAI crawlers (developer docs) (2026): OAI-SearchBot surfaces websites in ChatGPT search; sites opted out won't be shown in ChatGPT search answers but can appear as navigational links; GPTBot is for training foundation models; ChatGPT-User is for user-initiated actions and robots.txt may not apply; OAI-AdsBot validates landing pages submitted as ads on ChatGPT; each setting is independent; about 24 hours for search systems to adjust after a robots.txt update
- OpenAI Help Center, Searching the web with ChatGPT (2026): ChatGPT ranks search results using multiple factors and placement is not assured; to be eligible, allow OAI-SearchBot and OpenAI's published searchbot IP addresses; ChatGPT search sometimes partners with other search providers, rewriting the query into one or more targeted queries and sending follow-up queries; links to Microsoft privacy statement and Shopify privacy policy for how search providers process queries
- Google Search Central, AI features and your website (2025): To appear as a supporting link a page must be indexed and eligible to be shown with a snippet; no additional requirements or special optimizations; no new machine-readable files, AI text files or markup needed; no special schema.org structured data; AI features may use query fan-out; structured data should match visible text; traffic included in Search Console Performance report under Web; last updated 2025-12-10
- Google Search Central, Google's common crawlers (2026): Google-Extended does not impact a site's inclusion in Google Search nor is it used as a ranking signal; it manages use for Gemini training and grounding
- Anthropic Privacy Center (2026): ClaudeBot collects content for possible model training (blocking excludes future content from training datasets); Claude-User retrieves content in response to user queries (blocking prevents retrieval); Claude-SearchBot indexes content for search (blocking may reduce visibility and accuracy in search results); robots.txt recommended over IP blocking
- Perplexity docs, Perplexity crawlers (2026): PerplexityBot surfaces and links websites in Perplexity search results and is not used to crawl content for AI foundation models; Perplexity recommends allowing PerplexityBot and its published IP ranges; Perplexity-User supports user actions and generally ignores robots.txt
- llmstxt.org (2024): llms.txt proposed by Jeremy Howard, published September 3, 2024, as a /llms.txt Markdown file to help LLMs use a website; described as a proposal open for community input
- IndexNow.org (2026): IndexNow lets site owners notify search engines of added, updated or deleted URLs; supported by Microsoft Bing, Naver, Seznam.cz, Yandex and Yep
- arXiv / KDD 2024 (2023): GEO paper (Aggarwal et al., arXiv Nov 2023, KDD 2024): Cite Sources, Quotation Addition and Statistics Addition achieved 30-40% relative improvement in visibility on GEO-bench; keyword stuffing gave little to no improvement
- Google Analytics Help, Custom channel groups (2026): GA4 custom channel groups: created in Admin > Data display > Channel groups; traffic goes to the first channel whose definition it matches in the current order; can be applied to reports retroactively; Editor role or above required
- Google Search Central, Understand JavaScript SEO basics (2026): Googlebot renders pages with headless Chromium and executes JavaScript, then indexes the rendered HTML; server-side or pre-rendering is still recommended because it makes sites faster for users and crawlers and not all bots can run JavaScript
- Operator Intelligence (internal site records and live checks) (2026): operatorintelligence.ai migrated from a Vite single-page app to TanStack Start server-side rendering on 2026-10-16; all user agents (browser, Googlebot, GPTBot, ClaudeBot) receive identical full HTML; live robots.txt allows * and named AI crawlers (Bytespider disallowed); live /llms.txt returns 200
Get the Operator Intelligence briefing
A short, practical note every two weeks on what is working with AI in real businesses. No spam; unsubscribe anytime.