How to Get ChatGPT to Recommend Your Website

Updated 2026-10-24 · By the GEO Checker team

ChatGPT recommends websites it can reach, understand and trust. There is no paid placement and no secret switch: you make your pages easy for OpenAI’s search crawler to fetch, write answers worth quoting, and build the outside signals that make your brand a safe recommendation. This guide walks through each step in order, with examples.

How ChatGPT decides what to recommend

When a user asks ChatGPT something like “what is the best tool to check if my site blocks AI crawlers?”, two different systems can be involved. The first is the model’s trained knowledge: what it learned from public web text up to its training cutoff. The second is ChatGPT Search: a live retrieval step where ChatGPT searches the web, reads a handful of pages and cites them with links. For a new or small website, the second path is by far the most realistic way to be recommended, because it reflects your site as it is today.

OpenAI documents three separate user-agents. OAI-SearchBot crawls pages so they can appear in ChatGPT Search results. GPTBot collects data that may be used to train future models. ChatGPT-User fetches a page when a user (or a custom GPT) asks ChatGPT to open a specific URL. These are controlled independently in robots.txt, which means you can opt out of training while staying visible in search. OpenAI also states that it uses third-party search providers, so being indexed by Bing still helps.

A recommendation therefore needs three things: the page must be retrievable (crawlers are allowed and not blocked by a firewall), relevant and quotable (it directly answers the question in clear text), and credible (other sites, reviews and directories mention you, so the model is confident naming you).

Step by step

1. Allow OAI-SearchBot and ChatGPT-User in robots.txt. Open yourdomain.com/robots.txt and make sure no rule disallows OAI-SearchBot or ChatGPT-User for the pages you want recommended. A common mistake is a blanket “User-agent: * / Disallow: /” on a staging template that went live, or an “AI blocker” plugin that lists every AI bot. Run the AI crawler checker to see which of nine AI bots your rules allow.
2. Make sure your CDN or firewall is not silently blocking them. robots.txt is only half the story. Cloudflare “Block AI bots”, Bot Fight Mode, rate limits and custom WAF rules can return 403 or a challenge page to AI crawlers even when robots.txt says Allow. The free AI visibility checker fetches your page with each crawler’s user-agent and compares the result with a normal browser, so you can spot a block in 20 seconds.
3. Get indexed by Bing and submit a sitemap. Create a free Bing Webmaster Tools account, verify your site and submit sitemap.xml. Bing’s index feeds Microsoft Copilot and is one of the search sources ChatGPT can draw on. Keep the sitemap limited to pages you actually want found and use real last-modified dates.
4. Write pages that answer one question clearly. For each question your buyers ask, have one page whose title and first paragraph answer it directly. Start with a definition sentence (“X is a … that …”), then give concrete numbers, steps, a comparison table and a short FAQ. AI answers quote compact, factual sentences; marketing slogans are rarely cited.
5. Add structured data and an llms.txt file. Add Organization, WebSite and, where relevant, Product, SoftwareApplication, FAQPage and HowTo JSON-LD so machines can identify your brand, prices and answers unambiguously. Then publish an llms.txt summary of your site; validate it with the llms.txt checker. Online stores should also run the AI shopping agent readiness check for product feeds and policy schema.
6. Earn mentions on sites ChatGPT already trusts. Get listed in relevant directories, comparison articles, review sites and community discussions (Reddit, Hacker News, niche forums) under your own name. Do it honestly: real launches, real answers to real questions. Buying links or posting spam violates search engine policies and tends to hurt rather than help.
7. Measure crawls and mentions, then iterate. Confirm that OAI-SearchBot actually visits your important pages with AI crawler analytics, and keep a fixed list of 6–10 real buyer questions that you ask ChatGPT every week. Record whether you are cited and which competitors appear. Single answers vary a lot, so look at the trend over several weeks.

Example: a robots.txt that keeps you visible in ChatGPT

This configuration allows ChatGPT Search and user-initiated fetches, while opting out of model training. It is a common, reasonable default for businesses that want recommendations but are cautious about training use:

User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: GPTBot
Disallow: /

User-agent: *
Allow: /
Disallow: /checkout/

Sitemap: https://example.com/sitemap.xml

Changes to robots.txt are usually picked up by OpenAI within about a day, according to its crawler documentation. If you are unsure whether to keep GPTBot, read should you block GPTBot in robots.txt.

Example: rewriting a page so it is easier to cite

Before: “We empower teams to unlock next-generation visibility across the AI landscape.” Nothing in that sentence can be quoted as an answer.

After: “GEO Checker is a free tool that tests whether 9 AI crawlers, including OAI-SearchBot and PerplexityBot, can access a web page. A check takes about 20 seconds and needs no sign-up.” This version contains a definition, a number, named entities and a clear benefit, which are exactly the pieces an AI answer reuses.

Common reasons ChatGPT ignores a site

Be patient. New sites typically need weeks, not days, before they show up in AI answers, and nobody can guarantee a recommendation. What you can control is removing every technical obstacle and publishing the clearest answer on the web for your topic.