Free Test: What Happens When ChatGPT Reads Your Page

This test answers one narrow question: when a person asks ChatGPT to read a specific page, does the text ChatGPT gets match the text you published?

That request is handled by ChatGPT-User, which OpenAI describes as an agent for actions a user initiates. It is a user-triggered fetcher, not a search crawler. OpenAI's bots page says ChatGPT-User is "not used to determine whether content may appear in Search" and that, because a user starts the action, "robots.txt rules may not apply."

So the test reveals two practical failures:

• Missing content. ChatGPT reads the page but skips facts a visitor can see.
• Blocked fetches. ChatGPT cannot open the page at all, often because a firewall or CDN turns the request away.

Both are fixable on your side. For a fuller picture of the flow, see what happens when ChatGPT opens your URL.

You need a ChatGPT account, a terminal, and ten minutes.

1. Pick a page with hard facts. Choose a pricing, product or about page that states specific values: prices, plan names, a founding year. Write those facts down from the live page in your browser.
2. Check the raw HTML first. Fetch the page without running JavaScript and search for one of the facts:

curl -s https://example.com/pricing | grep -i "49"

If the fact is missing here but visible in your browser, it is added by JavaScript. To mimic the agent more closely, add -A with the ChatGPT-User string published on OpenAI's bots page. 3. Reduce noise in ChatGPT. Use a fresh chat. Turn memory off or use a logged-out session, since OpenAI says saved memories can shape how ChatGPT rewrites a query. 4. Ask for exact facts, not a summary. Try this prompt: "Open https://example.com/pricing and list every plan name and price exactly as written on that page. If you cannot open it, say so and report the error." 5. Confirm it actually fetched. Look for a link or citation to your URL in the answer. No link means ChatGPT may have answered from memory instead. 6. Check your server or CDN logs. Filter for the ChatGPT-User agent around the time of your test. Note the status code: 200, 403, 429, or a challenge page.

Compare ChatGPT's answer line by line with your notes from step 1.

The JavaScript case is common. Vercel's crawler study found ChatGPT's crawlers download JavaScript files but do not execute them, so client-side rendered content is invisible to them. If your site is a single-page app, read whether ChatGPT can read a React website.

The blocked case is often silent. Bot management can mistake legitimate agents for scrapers and return a 403 with no alert to you. Use your logs to verify a real AI crawler hit before you change any rules.

Very heavy pages can also be cut short by fetch limits. If your HTML is large, run the page size crawler limit check.

Be clear about the limits before you act on the result.

• It does not test ChatGPT search inclusion. As of October 2026, OpenAI's documentation says OAI-SearchBot controls whether sites appear in ChatGPT search answers. Each OpenAI agent setting is independent. A page ChatGPT-User can read may still be excluded from search.
• It does not test robots.txt. OpenAI says robots.txt may not apply to ChatGPT-User, so a successful fetch says nothing about your rules for OAI-SearchBot. Check those separately, and make sure crawlers can fetch your robots.txt file itself.
• It does not show whether ChatGPT recommends you. Reading a page on request is different from naming your brand when a buyer asks an open question.
• One run is a snapshot. ChatGPT decides when to browse, and results vary between sessions and models.
• It covers one engine. Perplexity, Claude and Google use different agents and different rules.

The search gate is a conjunction. OpenAI's help article says eligibility requires allowing OAI-SearchBot and confirming your host or CDN allows traffic from OpenAI's published search bot IP addresses. Neither is proven by this test.

Each outcome maps to a specific change you control.

• Missing facts: move prices, plan names and key claims into the HTML your server sends. Server-side rendering or prerendering does this. It does not make ChatGPT more likely to choose you. It makes your content exist in what the agent receives.
• Blocked fetch: review firewall and bot-management rules for OpenAI agents. Validate by user agent plus verified-bot programs rather than IP ranges alone.
• Answered from memory: nothing on your page fixes a training cutoff. Make sure the current facts are stated plainly on a page that can be fetched, so a live fetch has a sentence to quote.

To see your pages the way other crawlers do, try the methods in seeing your site like an AI crawler.

The manual test tells you whether ChatGPT can read one page. It does not tell you whether AI engines mention your brand when buyers ask about your category.

The free checker on this site covers that second question. It runs up to three buyer questions, yours or ones written from your homepage, and scores the answers. Treat it as a snapshot, not continuous tracking. Run the manual fetch test first, fix any access or rendering gaps, then check whether you are being named.

Can ChatGPT open any website I give it? Not always. ChatGPT-User fetches the page on your request, but a firewall, CDN challenge or rate limit can block it. Pages behind logins are also out of reach. If ChatGPT reports an error on a public page, check your server or CDN logs for a 403, 429 or CAPTCHA response to the ChatGPT-User agent at the time of your test.

Is ChatGPT-User the same as OAI-SearchBot? No. OpenAI lists them as separate agents with independent settings. ChatGPT-User handles actions a user starts, like opening a pasted URL, and OpenAI says it is not used to decide search inclusion. OAI-SearchBot crawls for ChatGPT search, and sites that opt out of it are not shown in ChatGPT search answers. Test and configure each one on its own.

Does blocking ChatGPT-User in robots.txt stop this fetch? Not reliably. OpenAI says that because these actions are initiated by a user, robots.txt rules may not apply to ChatGPT-User. If you need to stop user-triggered fetches, that happens at the server, firewall or CDN level. Be careful, because the same rules can accidentally block OAI-SearchBot and remove you from ChatGPT search.

Why did ChatGPT describe my page wrongly without citing it? It probably did not fetch the page. ChatGPT decides when to browse, and without a fetch it answers from training data frozen at a knowledge cutoff. That answer can be months out of date. Rerun the test with a prompt that tells it to open the exact URL and report any error, and check that the reply links to your page.

Does passing this test mean my page will be cited in ChatGPT answers? No. Passing shows that one user-triggered fetch received your content. Search inclusion depends on OAI-SearchBot access, network reachability and whether ChatGPT retrieves your page for a given query. Even then, which sources get cited is decided by the model. The test removes one possible block. It does not promise citations or traffic.