ChatGPT Visibility Factors, Graded by Evidence

Each factor below sits at one stage of how ChatGPT answers. Access decides whether OpenAI can fetch your page. Retrieval decides whether it enters the model's context for a query. Generation decides what gets cited. Parametric knowledge is what the model knows with search off, and nobody can change that this quarter.

The pattern is plain. The levers that hold up remove blocks. Nothing additive has credible evidence behind it.

Server-side rendering

Vercel's crawler study found that ChatGPT's crawler fetched JavaScript files on 11.50% of requests but did not execute them, so it can't read client-side rendered content. Independent testing in June 2026 re-confirmed that GPTBot, OAI-SearchBot and ChatGPT-User do not render JavaScript. If your pricing or product names load only after hydration, they may be invisible to ChatGPT. Fixing this doesn't make you more likely to be chosen. It decides whether your content exists as a candidate at all.

OAI-SearchBot permission

OpenAI's crawlers page says sites opted out of OAI-SearchBot "will not be shown in ChatGPT search answers, though can still appear as navigational links." Allowing it buys eligibility, not ranking. A minimal rule set that allows search while declining training looks like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

Note that ChatGPT-User, the agent for user-initiated fetches, works differently. OpenAI says robots.txt rules "may not apply" to it.

Network reachability

OpenAI's help centre states eligibility as a conjunction: allow OAI-SearchBot and confirm your host or CDN allows traffic from OpenAI's published searchbot IP addresses. A firewall that returns 403 to OAI-SearchBot removes you silently, whatever robots.txt says. OpenAI also warns against relying on IP-only allowlisting, so validate on user agent plus verified-bot settings. Cloudflare changed its AI crawler defaults on 2026-09-15, which is a reason to re-test even if you changed nothing. The comparison of rewrites and access fixes explains why these checks come first.

A 2026 critical survey of 45 GEO studies (arXiv 2607.14035) found topical relevance and context position to be the most reproducible levers. A separate 252,000-trial factorial agreed. Two caveats keep this at moderate. Both measured content already placed in the model's context, not live ChatGPT retrieval. And context position is something no publisher can set.

Relevance is not keyword coverage. OpenAI says ChatGPT search "rewrites your query into one or more targeted queries," so you can't see the queries that actually retrieve you. What you can control is stating specific facts plainly on a fetchable page: a price, a number, a one-line definition. That gives the engine a sentence it can quote, and it is the only positive experimental signal for on-page content. Our guide to LLM citation optimization covers what those studies did and did not test.

OpenAI says ChatGPT search "sometimes partners with other search providers" and has named Microsoft. That language is hedged, and the provider list traces to 2024-era documentation. Nobody has published an end-to-end effect size from Bing indexing to ChatGPT citations.

Otterly's 2026 experiment with 2,000 AI-written posts adds a warning. Its sites stayed in Bing and drew Copilot citations, yet ranking in Bing produced effectively zero ChatGPT citations, even with ChatGPT's crawlers visiting. That is one pair of sites on a low-stakes topic. It still shows that "ChatGPT uses Bing" is not a strategy. Keep your Bing index healthy for Copilot's own reasons, and treat any ChatGPT benefit as unmeasured.

• Schema. Ahrefs tracked 1,885 pages that added schema against matched controls and found no major uplift on any platform. ChatGPT citations moved 2.2%, and AI Overviews citations fell a statistically significant 4.6%. Every treated page was already heavily cited, so whether schema helps a new page enter the set is untested. Existing markup does no harm. It just isn't a lever. See does schema markup help AI citations.
• llms.txt. OpenAI does not document reading it, and neither does any other engine. Google says Search ignores it. More in what llms.txt is.
• Chunking and AI-friendly style. Google rejects both and says there is no ideal page length. The chunking research measures segmentation the retrieval operator performs inside its own index.
• Allowing GPTBot. GPTBot governs training only. OpenAI says "Each setting is independent of the others" and claims no search benefit for GPTBot.

If these tasks sit on your roadmap, retire them deliberately and move the hours to access checks.

OpenAI's full ranking disclosure is that ChatGPT ranks results "using multiple factors" and that placement is not guaranteed. Nothing is enumerated. Three more gaps matter for anyone measuring ChatGPT, as of October 2026:

• Search is discretionary. The model decides per query whether to search. Without visible citations, you don't know if an answer was grounded.
• Knowledge cutoffs differ by route. OpenAI lists a Feb 16, 2026 cutoff for the GPT-5.6 family and Aug 31, 2025 for the chat-latest Instant path. Content published after a cutoff can't reach answers given with search off.
• Memory changes queries. Saved memories can shape how ChatGPT rewrites a search, so logged-in tests aren't reproducible.

No published study measures what most teams want: whether a change moves ChatGPT citations for their brand.

1. Fetch your key pages with JavaScript disabled and confirm pricing, product names and definitions are in the raw HTML.
2. Read robots.txt per agent. Confirm OAI-SearchBot is allowed, separately from GPTBot.
3. Request a page as OAI-SearchBot and confirm a 200, not a 403 or challenge page. The AI Crawler Checker shows which crawlers your firewall turns away.
4. Look for accidental noindex, nosnippet or max-snippet:0 directives inherited from a template.
5. Ask buyer questions and record whether ChatGPT names you correctly. The free checker on this site gives a snapshot for three questions.

None of this guarantees a citation. It removes the blocks you control.