Why is my website not showing up in ChatGPT?

A website not showing up in ChatGPT may be blocked from search access, difficult to find or simply not selected for that answer. On 15 October 2026, Cloudflare and OpenAI reported a discovery test that brought previously unknown pages into OpenAI's search index, its searchable collection of web pages. For small businesses seeking unpaid ChatGPT visits, that makes access worth checking before buying more articles.

What changed in October

The pilot covered roughly 1% of Cloudflare's total network and enabled OpenAI to discover and index hundreds of millions of new pages, according to the published results. That describes pages found and stored for search, not customers sent to those websites.

Finding a page, keeping it in a searchable collection and selecting it for an answer are different steps. The pilot does not establish that your site was involved, that it is blocked or that opening access will make ChatGPT recommend you.

Check what ChatGPT actually finds

Start by asking ChatGPT to search the web for your business name and town. Note whether it finds your website, whether it links to the right page and whether the details are correct.

Then try a question a new customer would ask, such as 'emergency plumber in Leeds' or 'accountant for a small cafe in Bristol'. Keep the wording, date and links shown in the answer so you can compare later.

Finding your business by name but omitting it from a general recommendation is different from failing to find it at all. Neither test proves the state of OpenAI's whole searchable collection, so treat these checks as a starting record rather than a verdict. If the information is wrong rather than missing, the page checks in Google AI wrong about my business: what to check are useful here too.

Is my website blocking ChatGPT?

Ask your host, the company that keeps your website online, to check access for OpenAI's search crawler. A crawler is a program that fetches web pages automatically.

Request evidence from its access records for your important service or product pages, rather than a general 'AI allowed' reply. Have the host check:

• Your robots.txt file, a small file on your website that tells search engines and AI companies which pages they may read.
• Security settings that refuse automated visits or show a 'prove you're human' check.
• Password protection or other restrictions on pages that should be public.

Ask whether verified OpenAI search requests receive the page itself, rather than a refusal or a human-verification screen. A page loading for you does not prove that an automated visitor can read it.

Do not switch off all website protection. Ask for a targeted change for the search service and have the host test it while retaining protection against unwanted requests.

You can ask about search access separately from AI training, which uses content to develop an AI system. Cloudflare's 15 September 2026 announcement describes a Disallow AI Training setting intended to preserve search access while refusing training use. Its protection depends on controls and commitments from the companies running those automated visitors, so ask your host which services it covers.

Make your existing pages easier to find

Give each important service a page that a customer can reach through an ordinary website link. A treatment buried in a booking form or a service described only in a downloadable brochure is harder for a visitor to inspect.

For a dental practice, that page should explain the treatment, who it is suitable for, the practice location and how to enquire. For a shop, product pages should state what is being sold and whether it is available. Use actual details, not a paragraph that could describe any competitor.

Link relevant pages from your main menu and from other pages where the link helps a reader. Ask your website supplier to check your sitemap, a file listing pages for search services to find, and confirm it contains the public pages you want discovered. A sitemap is a discovery aid, not registration with ChatGPT.

Remove outdated claims about areas you no longer serve and keep your business name and contact details consistent. The aim is to make an existing useful page accessible and understandable, rather than publish another general article about your industry.

Measure visits and enquiries separately from mentions

Look for visits identified as coming from ChatGPT in whatever visitor reporting your website already has. Treat that as an incomplete record: some visits may arrive without a useful source label.

More importantly, record whether those visitors make an enquiry, book or buy. Ask new customers how they found you, and note changes to your website or hosting settings alongside those records.

An answer that mentions your cafe without a link is different from a visitor who reads the menu. Keep mentions, visits and business enquiries separate so a more flattering answer does not get mistaken for more trade.

More articles will not fix a blocked crawler

More articles will not fix a blocked crawler, and being indexed does not guarantee a recommendation. Check your existing host's access settings before buying more content.

You can do much of this yourself if you can edit your pages and ask your host for help. You do not need a new subscription just to check a security setting, correct an opening time or add a useful link.

It is also reasonable to leave AI-specific work alone for now. If your booking page is broken, your services are unclear or your opening hours are wrong, those problems deserve attention before repeated ChatGPT tests.

If your existing pages are accurate and accessible, and you have no evidence that missing ChatGPT mentions are costing enquiries, monitor rather than invent a publishing target. A missing recommendation is not a diagnosis of a broken website.

Keep useful pages working for search

Your organic traffic is the visitors you get from ordinary, unpaid search results, along with the enquiries those visits can produce. Record ChatGPT visits separately, but remember that these access checks concern the same service pages, product information and contact details that support your wider search presence.

At PageNumberOne, our automated SEO, the work that helps your website appear in unpaid search results, runs every day. On the Page One plan, weekly technical checks and fixes sit alongside work on page titles, headings and content. We also maintain links between related pages, work on local business information and track whether your business appears in AI answers.

That routine addresses the practical part of this news: helping search services reach and understand useful pages, then watching for changes. It cannot override restrictions controlled by your host, so a hosting-level block still needs to be checked with that company.

We track where your pages appear in search and provide a plain-English monthly report, which you can read alongside your own enquiry records. That keeps attention on access, useful information and evidence, rather than treating every announcement as a reason to publish more. We cannot guarantee a place on Google or a recommendation in ChatGPT.