Sooner or later every business owner tries it. You open ChatGPT, ask it to recommend someone who does what you do in your city, and read the reply with a slightly sinking feeling. Three names, none of them yours. The good news is that this is not random and it is not rigged. The assistant is working from what it can find, and you have more influence over that than you might think.
The short answer
To be recommended by ChatGPT, a business needs to be reachable, understandable and vouched for. Reachable means its crawlers are allowed to read your site. Understandable means your pages state plainly who you are, what you do and for whom, with the answer first. Vouched for means other trusted sites, such as review platforms, directories and publications, mention you. Nobody can guarantee a recommendation, but these three things decide the odds.
How ChatGPT decides who to mention
An AI assistant has two sources of knowledge. The first is training: everything it absorbed up to a certain date. The second is live search. When a question needs current or local information, ChatGPT searches the web, reads a handful of pages and writes its reply from them, usually with links to the sources.
For most small businesses, live search is where the opportunity lies. You cannot change what a model learned last year. You can change what it finds today. OpenAI's own documentation is clear on the first requirement: sites that block its search crawler, OAI-SearchBot, will not be shown in ChatGPT's search answers.
Eight steps that improve your chances
- 1
Make sure the door is open
Check your robots.txt for anything that blocks OAI-SearchBot, GPTBot, ClaudeBot or PerplexityBot. Then check your CDN or security service. Cloudflare, for example, has AI crawler settings that can block these bots at the network level whatever your robots.txt says, and since September 2026 one of those settings can block search crawlers too. Many site owners have no idea theirs is switched on.
- 2
Say exactly who you are
Somewhere on your site, in plain text, state what the business is, what it offers, where it is, who it serves and how to contact it. An About page that reads like a mission statement gives an AI nothing to work with. A paragraph of facts does.
- 3
Keep those facts identical everywhere
Your name, address, phone number and description should match on your website, your Google Business Profile, your social pages and every directory. An assistant that finds three different versions has less reason to trust any of them.
- 4
Answer buyers' questions on your own site
Think about what people ask before they hire someone like you, and give each question its own page or section. One 2026 preprint, not yet peer reviewed and written by the founder of an AI-visibility company, found that brands were mentioned far more often when one of their own pages was among the sources the assistant had read.
- 5
Put the answer first
Do not make the reader, or the machine, wait. An analysis of more than 18,000 ChatGPT citations found that 44% came from the first 30% of the page. State the answer in the opening lines, then explain.
- 6
Use real numbers, real sources and real quotes
The original academic study on generative engine optimization tested nine ways of rewriting content. Adding quotations and statistics raised visibility the most, with citing sources close behind. On the study's main measure, stuffing in keywords made things worse. The catch is that the figures have to be true, which is why we never use a number we cannot back up.
- 7
Get mentioned where AI looks
Assistants lean on review platforms, industry directories, LinkedIn, YouTube, Reddit and news sites. Research by Ahrefs across 75,000 brands found that mentions of a brand around the web correlated with AI visibility much more strongly than backlinks. Claim your listings, ask customers for reviews on the platforms that matter in your field, and publish where your industry reads.
- 8
Keep it current
Assistants prefer fresh information for anything time-sensitive. Review your key pages regularly and update them when something changes. Changing the date without changing the content does not count, and Google explicitly warns against it.
What does not work
- An llms.txt file on its own. It is harmless and takes a few minutes, so there is no reason not to have one. But Google has said its search ignores the file, and no major AI company has confirmed that it relies on one.
- Schema markup as a magic switch. Structured data is worth having for search engines. A controlled test in 2026 found no meaningful lift in AI citations from adding it.
- Keyword stuffing. On the main measure in the academic study above, it scored below doing nothing.
- Hidden instructions to AI. Text written for machines and hidden from people is spam, and Google's spam policies now explicitly cover attempts to manipulate AI responses.
- Fake reviews and paid mentions. They are against the rules of every major platform, and they tend to be found.
How to measure it honestly
This is where many reports mislead. AI answers are not stable. When SparkToro ran the same prompts through AI tools thousands of times, the same list of brands came back less than 1% of the time. A screenshot showing you were recommended once proves very little, and so does one showing you were not.
What does work is frequency. We keep a fixed list of questions a customer might ask, put each to the main assistants several times, and record three things: whether the business is mentioned, how it is described, and which sources the assistant cites. Repeated monthly, that shows a real trend. The cited sources are the most useful part, because they tell you where the assistant is getting its picture of your industry.
Two reports from the search engines help as well. Google Search Console now shows impressions from Google's own AI features, and Bing Webmaster Tools has an AI Performance report covering Microsoft Copilot.
A plan for the first 30 days
- 1.Week 1. Ask ChatGPT, Gemini and Perplexity ten questions your customers would ask, three times each. Record the answers and the sources.
- 2.Week 1. Check robots.txt and your CDN settings for blocked crawlers. Verify your site in Google Search Console and Bing Webmaster Tools.
- 3.Week 2. Rewrite your About page and main service pages so they state the facts plainly, with the answer first.
- 4.Week 2. Make your name, address, phone and description match on every profile and listing.
- 5.Week 3. Write three pages that answer the questions where competitors were cited and you were not.
- 6.Week 4. Claim or complete your profiles on the review sites and directories that appeared as sources. Ask recent customers for reviews.
- 7.After that. Repeat the ten questions every month and compare.
Questions people ask
Can I pay ChatGPT to recommend my business?
No. ChatGPT does show ads in many countries, but OpenAI says they are clearly labeled and kept separate from the answer. The recommendations inside an answer cannot be bought.
How long does it take to show up in ChatGPT?
Changes that affect live search, such as unblocking a crawler or publishing a clearer page, can show within weeks. Building up mentions and reviews takes months. What a model learned in training only changes when the model is updated.
Does ChatGPT use Google or Bing to search?
OpenAI does not fully disclose its search sources. It runs its own crawler, OAI-SearchBot, and also draws on search partners. The safe approach is to be indexed everywhere: in Google, in Bing, and open to OAI-SearchBot.
Does this work for Gemini, Perplexity and Claude as well?
Largely, yes. Each uses different sources, but all of them reward businesses that are easy to reach, clearly described and mentioned by others. Work done for one usually helps with the rest.
Will a brand-new business ever be recommended?
It is harder without reviews or mentions, but not impossible. Specific questions have fewer good answers, so a new business with a clear page on a narrow topic can be cited where a vague, established competitor is not.
Want to know what AI says about you?
An AI visibility audit is the first step of our GEO service. We put your customers' questions to the main assistants, record what they say and which sources they trust, and set out what to change.Sources
- OpenAI: Overview of OpenAI crawlers (OAI-SearchBot, GPTBot, ChatGPT-User)
- Cloudflare: Stay discoverable in search while disallowing AI training (September 2026)
- Aggarwal et al.: GEO: Generative Engine Optimization (KDD 2024)
- Search Engine Journal: The science of how AI pays attention (analysis of 18,012 ChatGPT citations, February 2026)
- Ahrefs: AI Overview brand visibility correlations, 75,000 brands (May 2025)
- Ahrefs: Does schema markup increase AI citations? (May 2026)
- Search Engine Journal: AI recommendations change with nearly every query (SparkToro study, January 2026)
- Brand mentions and own-domain citations in AI answers (arXiv preprint, September 2026; data from an AI-visibility vendor)
- Google Search Central: AI optimization guide
- OpenAI: Testing ads in ChatGPT (February 2026)
