Guide
ChatGPT Brand Rankings: See Who Gets Recommended
ChatGPT does not publish one permanent brand leaderboard. The useful ranking is prompt-level: who appears, where they appear, which pages are cited, and whether a person clicks through.
3
OpenAI agents with different search, training, and user-fetch roles
5
AI engines on paid monitoring, reported separately
Weekly
Perplexity monitoring on Free while the workspace stays active
$49
Launch price for all-five-engine monitoring and developer access
What do ChatGPT brand rankings actually measure?
A ChatGPT brand ranking is not a universal list stored somewhere inside the product. It is the result of a specific question at a specific time. Change the wording, context, location, or search behavior and the answer can change. The defensible unit is therefore a repeated prompt result, not a claim that a brand is permanently number one in ChatGPT.
For a buyer question such as “best AI search monitoring tool for a SaaS team,” record whether the brand appears, its position among named options, the pages cited, the competitors included, and any referral visit. Those observations answer different questions and should never be collapsed into one vague visibility claim.
The answer to track
For each buyer prompt: did ChatGPT name the brand, where did it place the brand, which URL did it cite, which competitors appeared, and did a person visit the cited page?
Which OpenAI crawler controls ChatGPT Search visibility?
OpenAI documents three separate agents. OAI-SearchBot is the automatic crawler for ChatGPT Search. GPTBot is the separate control for content that may be used in model training. ChatGPT-User handles some visits triggered by a person. Treating all three as one “ChatGPT crawler” produces incorrect robots.txt advice and misleading analytics.
| Agent | Purpose | What OpenAI documents | Practical action |
|---|---|---|---|
| OAI-SearchBot | Automatic search discovery | OpenAI says pages opted out of this crawler will not appear in ChatGPT Search answers, although a title and link can still appear as a navigational result. | Allow the crawler on intended public paths and allow OpenAI's published search-crawler IP ranges. |
| GPTBot | Potential model training | OpenAI says disallowing GPTBot indicates that crawled content should not be used to train its generative foundation models. | Set this policy independently from search discovery. Allowing Search does not require allowing training. |
| ChatGPT-User | Some user-triggered fetches | OpenAI says this agent may visit a page after a user request and is not used to determine whether content may appear in Search. | Treat a visit as fetch evidence only. It does not prove that the page was cited or the brand was recommended. |
Allowing OAI-SearchBot makes a page eligible for ChatGPT Search discovery. It does not guarantee indexing, retrieval, citation, position, or recommendation. OpenAI also says a robots.txt change can take about 24 hours to propagate through its search systems.
Measure these five ChatGPT visibility signals separately
| Signal | Question it answers | Evidence boundary |
|---|---|---|
| Mention | Was the brand named in the answer? | A mention is not automatically a citation or recommendation. |
| Answer position | Where did the brand appear among named options? | Position is prompt-specific, not a permanent global rank. |
| Citation | Did the answer link to a page on the brand's domain? | A citation can occur without a favorable recommendation. |
| Competitor set | Which alternatives appeared in the same answer? | Co-occurrence does not reveal why the model selected them. |
| Referral | Did a person click through from ChatGPT? | This is the strongest proof that answer visibility produced a visit. |
This separation matters because a crawler request is not an answer outcome. A page can be fetched and never cited. A brand can be mentioned without its site being linked. A cited page can receive no visit. Report the funnel honestly from eligibility to retrieval to answer evidence to referral. A separate five-engine source-divergence study of 62 complete query groups found brand-mention agreement ranged from 85.5% to 98.4%, while citation-domain overlap was 0.197 or lower for nine of ten engine pairs. A ChatGPT ranking should not stand in for another engine's evidence.
What can you do to improve ChatGPT brand visibility?
OpenAI publishes a Search eligibility rule, not a ranking formula. The workflow below separates that documented requirement from content and authority hypotheses that you must test against your own prompt set.
1. Make the intended page eligible
Evidence: OpenAI documents OAI-SearchBot access and its published IP ranges as the controls for Search discovery.
Test: Verify the target URL is allowed and returns a successful response to the documented search crawler.
2. Answer one buyer question precisely
Evidence: A product page is easier to evaluate when it states the category, use case, price, limits, and differentiators in self-contained language.
Test: Run the same buyer prompt before and after the page change. Do not claim causation from one changed answer.
3. Keep product facts consistent
Evidence: Conflicting names, prices, and plan limits make it harder for any retrieval system or buyer to reconcile what is current.
Test: Compare the visible page, metadata, structured data, documentation, and third-party profiles for contradictions.
4. Earn independent corroboration
Evidence: Reviews, reporting, research, directories, and community discussions can provide sources beyond the manufacturer's own claim.
Test: Record which third-party URLs are actually cited for the target prompt, then pursue relevant, editorially honest inclusion.
5. Measure the complete outcome
Evidence: Crawler access, a live fetch, a citation, a brand mention, answer position, and a referral describe different events.
Test: Keep those events separate in reporting and compare them by engine and landing page.
What does the public evidence not establish?
OpenAI does not publish numeric weights for brand mentions, schema markup, review ratings, freshness, domain age, or content depth in ChatGPT recommendations. It also does not publish a standard number of brands per response or a guaranteed relationship between a Google position and a ChatGPT recommendation.
Research about generative-engine citations can generate useful hypotheses, but a result from a separate model, prompt set, or intervention is not a ChatGPT ranking weight. For example, a study showing that an instruction to cite sources changed answer visibility does not prove that adding schema markup raises ChatGPT citations by the same percentage. Keep the study's intervention, sample, engine, and outcome attached to the claim.
Use this claim standard
“We observed X across this named sample and time window” is evidence. “ChatGPT gives X a 35% ranking weight” is not, unless OpenAI publishes that rule.
How does Foglift track ChatGPT brand rankings?
Foglift separates page readiness from answer outcomes. A Technical Audit checks crawl access, entity clarity, headings, FAQs, depth, citations, topical coverage, and structured data on one URL. AI Visibility then runs saved buyer prompts and stores mentions, answer position, cited URLs, competitors, sentiment, and response history. Sensor analytics separately records crawler activity and recognized AI referrals.
The public AI Brand Checker returns one live Perplexity result with no signup. It is not a free ChatGPT or multi-engine check. Free workspaces receive weekly Perplexity monitoring while active. Launch costs $49 per month, starts with 20 prompts on a weekly schedule across all five engines, permits a daily cadence, and adds REST API, CLI, and MCP access. Growth costs $129 per month and permits twice-daily monitoring.
That combination gives a team one auditable loop: diagnose the page, monitor the same prompts across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overview, inspect the cited sources and competitors, ship one change, then compare answer evidence and referrals without treating crawler fetches as visibility.
Sources and methodology
- OpenAI, Overview of OpenAI Crawlers, accessed September 25, 2026. Defines the independent OAI-SearchBot, GPTBot, and ChatGPT-User roles plus the published crawler IP ranges.
- OpenAI, Publishers and Developers FAQ, accessed September 25, 2026. Documents OAI-SearchBot access for summaries and snippets and the
utm_source=chatgpt.comreferral parameter. - Foglift, ChatGPT vs. Google AI Overview Citation Divergence, 2026. Compares citation sets by engine rather than treating AI search as one channel.
- Foglift, How We Score AI Visibility, 2026. Publishes the evidence window, denominator rules, tier formula, and error handling used in monitoring.
Frequently asked questions
Does ChatGPT recommend the same brands every time?
No. Answers can change with prompt wording, conversation context, location, date, and whether web search is used. Measure a fixed prompt set repeatedly instead of treating one answer as a permanent ranking.
Which OpenAI crawler controls ChatGPT Search visibility?
OAI-SearchBot controls automatic crawling for ChatGPT Search. GPTBot is a separate potential-training control. ChatGPT-User performs some user-triggered fetches and does not determine whether content may appear in Search.
Does blocking GPTBot remove a brand from ChatGPT Search?
No. GPTBot controls potential training use, not ChatGPT Search inclusion. Manage automatic Search discovery with OAI-SearchBot. The two controls are independent, so a site can allow OAI-SearchBot while disallowing GPTBot.
How many brands does ChatGPT usually recommend?
OpenAI does not publish a standard number. The count changes with the question and response format, so track whether the brand appears, its answer position, cited URLs, and the competitors included for each prompt.
Does schema markup guarantee a ChatGPT recommendation?
No. Keep structured data accurate and consistent with visible content, but OpenAI does not publish schema markup as a guaranteed ChatGPT recommendation factor. Search eligibility, retrieval, citation, and recommendation are separate outcomes.
How do I improve brand ranking in ChatGPT?
First allow OAI-SearchBot on the pages you want eligible for ChatGPT Search. Then publish current, verifiable product facts that answer the buyer's exact question, earn independent corroboration, and measure a fixed prompt set before and after each change.
How often should I check ChatGPT brand rankings?
Use a consistent weekly baseline first. Keep the prompts, brand aliases, engines, and scoring rules fixed. Increase cadence only when launches, pricing changes, or an active incident require faster feedback.
Fundamentals: Learn about GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) (the two frameworks for optimizing your content for AI search engines).
Related reading
How ChatGPT Ranks Websites
Separate documented Search eligibility from testable optimization hypotheses.
Track ChatGPT Recommendations
Build a repeatable prompt panel and measure recommendation changes over time.
AI Search Monitoring
Track mentions, cited URLs, competitors, sentiment, and answer history across five engines.
How We Score AI Visibility
Read the exact evidence window, denominator rules, tiers, and scoring formula.