Is My Brand Visible in AI Search? How to Check, Engine by Engine

An AI engine can treat your brand in three different ways. It can name you in the answer. It can link to your site as a source. It can state facts about you, which may be right or wrong. A useful check records all three, because each one fails for a different reason.

Visibility is also not one number. ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews, Google AI Mode and Microsoft Copilot draw on different indexes and models. Otterly's 2026 experiment showed how far apart they can sit: after Google deindexed its test sites, Microsoft Copilot kept citing them thousands of times while ChatGPT, Gemini, AI Overviews and AI Mode almost never did (Otterly). A blended score would have hidden that.

So the honest question is narrower: for these buyer questions, on this engine, with retrieval on or off, does the answer name you and get your facts right?

Every answer passes through up to four stages. Knowing which one failed tells you what to fix.

1. Parametric. What the model learned in training, frozen at a cutoff. OpenAI documented, as of August 2026, a February 16, 2026 cutoff for the GPT-5.6 family and August 31, 2025 for the Instant path ChatGPT also uses. Nothing you publish today reaches this layer.
2. Access. Whether the engine's crawler may fetch and store your page at all. This is set by your robots.txt, your firewall or CDN, and your rendering.
3. Retrieval. Whether your page enters the model's context for a given question. Engines rewrite the question first: OpenAI says ChatGPT search "rewrites your query into one or more targeted queries" before sending it to search providers (OpenAI Help).
4. Generation. Which retrieved pages the model uses and cites. OpenAI's API exposes retrieved sources separately from citations, and the retrieved set is often larger, so a missing citation does not prove your page was ignored.

Retrieval does not fire on every question. ChatGPT searches at the model's discretion, Claude.ai web search is a user toggle, and Google AI Overviews often do not appear at all. That is why every check must record retrieval state.

In your control

• Per-bot crawler permission. OpenAI's rules are independent per bot: OAI-SearchBot controls ChatGPT search, while GPTBot governs training only (OpenAI crawlers).
• Network reachability. A firewall or CDN can return 403 to a crawler that robots.txt allows.
• Rendering. Vercel found ChatGPT and Claude crawlers fetch JavaScript files but do not execute them, so they "can't read client-side rendered content" (Vercel).
• Index and snippet eligibility. Google says a page must be "indexed and eligible to be shown in Google Search with a snippet" to appear as a supporting link in AI Overviews or AI Mode (Google).
• Plain facts on fetchable pages. A specific price, number or one-line definition in the HTML gives an engine a sentence it can quote.

Not in your control

• What the model learned before its cutoff.
• Which queries the engine actually runs after rewriting yours.
• Where your page sits in the model's context and whether it gets cited.
• Brand size. Large brands dominate citations across engines, and no on-page change alters that.

The levers that hold up are access levers, and they work by removing blocks. No markup, phrasing or page structure has credible evidence of getting you cited.

This takes an afternoon. The one-afternoon AI visibility check has a fuller worksheet.

1. Write 5 to 10 fixed buyer questions. Use questions a buyer asks before they know you: "best payroll software for a 20-person agency", not your brand name. Add one or two brand questions ("what does [brand] cost?") to test facts. Say you run a dental clinic in Austin: "best dentist for Invisalign in Austin" is a buyer question.
2. Freeze the wording. Copy each question exactly every time. Small wording changes produce different answers.
3. Test logged out, in a clean session. OpenAI says ChatGPT may use saved memories when rewriting a search query, so logged-in results are not reproducible.
4. Run every question on every engine. Do not assume one engine predicts another.
5. Record retrieval state for each answer. If the answer shows sources or links, retrieval ran. If it shows none, mark it unknown, not "no search".
6. Log four things per answer: named (yes or no), linked (yes or no), facts correct (list any wrong ones), and which sites were cited instead of you.
7. Save the date and the raw answer text. You will need it to compare later.

For a quick first snapshot, the free checker on this site runs up to three buyer questions and scores the answers. It is one point in time, not tracking, so keep your own log for comparison.

Work from access outward. An engine cannot cite what it cannot fetch.

1. robots.txt, per bot. Check that retrieval bots are allowed: OAI-SearchBot, PerplexityBot, Claude-SearchBot, Meta-WebIndexer, Googlebot and bingbot. The most common error is a blanket AI block that also removes ChatGPT search eligibility. A minimal allow looks like this:

User-agent: OAI-SearchBot
Allow: /

2. Firewall and CDN. OpenAI says eligibility needs the crawler allowed and the host or CDN letting its published IP addresses through. From September 15, 2026, Cloudflare blocks Training and Agent crawlers by default on ad-displaying pages for new domains and existing free-tier customers, while Search crawlers stay allowed. Check what your settings now say. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.
3. Rendering. View your page source with JavaScript off. If your pricing, product names or key claims only appear after scripts run, most AI crawlers cannot see them. Server-render that content.
4. Suppression controls. Search your templates for noindex, nosnippet, max-snippet:0 and data-nosnippet. These keep content out of AI answers, and they often arrive by accident.
5. Index status, engine by engine. Check Google Search Console and Bing Webmaster Tools separately. Healthy in one tells you nothing about the other.
6. Facts on your own site. Make sure your price, founding year and product names match across pages. Contradictions give engines conflicting sentences.
7. The sites cited instead of you. When engines cite review sites, directories or comparison pages, being named there is associated with appearing in answers. That is an association, not a guaranteed effect.

For ChatGPT specifically, the ChatGPT visibility checklist walks through each gate.

• Treating one zero as "AI doesn't know us". A missing mention may be a retrieval-off answer from old training data, or a blocked crawler. See what a zero AI visibility score means.
• Assuming strong SEO carries over. Ahrefs found that 37.9% of URLs cited in AI Overviews also appeared in the top 10 results (March 2026), so most citations came from elsewhere. More on whether strong SEO means AI visibility.
• Allowing GPTBot and expecting ChatGPT search results. GPTBot governs training. Search uses OAI-SearchBot.
• Tracking a "rank" in AI answers. Google exposes AI impressions only, so no position data exists to check a rank against.
• Relying on analytics alone. Google AI Overviews send no referral tag, so traffic reports undercount AI exposure. See what Google Analytics can and cannot tell you.
• Buying shortcuts. Schema, llms.txt, "AI-friendly" rewriting and mass-published content have no credible evidence of producing citations. Hidden text or prompt injection aimed at AI crawlers is manipulation, and it carries real risk.

Answers move when engines change models, when retrieval behaves differently, and when your access changes without you touching anything. Re-run the same fixed prompts on a set schedule, and again after any site migration, CDN change or robots.txt edit. Compare like with like: same prompts, same logged-out setup, retrieval state noted.

The re-check cadence guide sets out a practical rhythm. When you want to compare yourself with competitors over time, measuring AI share of voice builds on the same log.

Why does ChatGPT give a different answer each time I ask about my brand? Answers vary because ChatGPT decides per question whether to search, rewrites your question into its own search queries, and may route to models with different training cutoffs. Logged-in sessions can also draw on memory. Test logged out, keep the wording fixed, and record whether sources appeared. Compare several runs rather than trusting one.

Does blocking GPTBot remove my brand from ChatGPT search? No. OpenAI states that each crawler setting is independent. GPTBot controls whether content may be used for training. OAI-SearchBot controls whether your site can appear in ChatGPT search answers. You can block one and allow the other. Check both lines in your robots.txt, then confirm your firewall lets OAI-SearchBot through.

Can AI search cite my site if Google has not indexed it? Some engines can, if they use a different index. In Otterly's 2026 test, Microsoft Copilot, which grounds on Bing, kept citing sites after Google removed them, while Google's AI surfaces almost never did. Google's AI Overviews and AI Mode require the page to be indexed and snippet-eligible in Google Search.

How long after a fix will AI answers change? Nobody can give a reliable timeline. Answers with live retrieval can reflect a fix once crawlers refetch and reindex your pages. Answers without retrieval come from training data, which only changes with a new model generation. Re-run your fixed prompts on a schedule and note retrieval state. See whether [a new model release changes brand answers](/questions/new-ai-model-changes-brand-answers).

Should I add schema markup to show up in AI answers? Not as a visibility tactic. Google says structured data is not required for generative AI search, and Ahrefs' matched test found AI Overview citations fell 4.6% on pages after adding schema. Existing markup does no harm, so there is no need to remove it. Put your effort into access, rendering and accurate facts instead.