Short answer
To get ChatGPT to recommend your business, it first has to be able to find and read your pages when it searches: allow its search crawler (OAI-SearchBot), be properly indexed in the classic search engines, then publish a clear page that answers the exact question being asked, with facts that match what the rest of the web says about you. Nobody can guarantee a spot in an AI answer, and no single file does the job. This guide gives the method in 9 steps, engine by engine, and how to measure the result honestly.
The short answer: three conditions, in this order
An AI answer engine only recommends what it could read and considers reliable. For a small or mid-sized business, it comes down to three conditions. If the first one fails, the other two do not matter.
This is the hands-on companion to our GEO guide, which explains how GEO differs from SEO and covers the five levers. Here we move to execution: what to fix, in what order, and how to check it.
| Condition | What it means in practice | How to check it |
|---|---|---|
| 1. Readable | AI search crawlers and classic search crawlers can open your public pages, and the important content is in text | robots.txt, CDN or firewall settings, a curl test |
| 2. Indexed and understood | Your pages are in Google’s and Bing’s indexes, and your business is described the same way everywhere (name, city, services, phone) | Google Search Console, Bing Webmaster Tools, a search for your own name |
| 3. Citable | One page answers the question your customers ask, with numbers, steps and sources | Ask the engines the question and see which pages they cite |
Where each engine gets its answers
Each engine runs its own crawlers and uses its own sources. robots.txt rules apply per user-agent: allowing one says nothing about another.
According to OpenAI’s help center, ChatGPT search uses third-party search providers as well as content provided directly by partners, and often rewrites the question into one or more more targeted queries. OpenAI’s crawler documentation states that sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links.
Google, for its part, says there are no additional requirements to appear in AI Overviews or AI Mode: a page must be indexed and eligible to be shown in Google Search with a snippet, and no new machine-readable files, AI text files or special schema.org markup are needed.
| Engine | Where sources come from | Crawler(s) to allow | Where to measure |
|---|---|---|---|
| ChatGPT (search) | Third-party search providers, partners, and sites open to OAI-SearchBot | OAI-SearchBot (search); ChatGPT-User (user-initiated visits). GPTBot is for training: blocking it does not remove you from search | Referrals from chatgpt.com in your analytics |
| Google AI Overviews and AI Mode | Google’s search index | Googlebot. Google-Extended does not affect Google Search | Search Console Performance report, “Web” search type (blended with other traffic) |
| Gemini app | Google, with grounding on the web | Google-Extended governs use for Gemini training and grounding | Referrals from gemini.google.com |
| Microsoft Copilot and Bing AI summaries | Bing’s index | Bingbot | Bing Webmaster Tools AI Performance report (citations, cited pages, grounding queries) |
| Perplexity | Its own index plus on-demand visits | PerplexityBot (not used for model training, per Perplexity); Perplexity-User | Referrals from perplexity.ai |
| Claude | Claude’s web search | Claude-SearchBot and Claude-User; ClaudeBot is for training | Referrals from claude.ai |
Sources: official documentation from OpenAI, Google, Microsoft Bing, Perplexity and Anthropic, checked on October 2, 2026 (links at the bottom). Crawler names change from time to time: re-read those pages before editing your robots.txt. Our AI crawlers guide covers robots.txt in depth.
The method in 9 steps
Steps 1 to 4 take a few days and fix most cases where a business is missing from AI answers. Steps 5 to 8 are the long-term work. Step 9 tells you whether it is working.
- 1. Write down the questions. List 20 to 30 real questions, phrased the way your customers would ask ChatGPT: “Which agency can rebuild my website for under $1,000?”, “Which roofer near me answers on weekends?”. Pull them from your emails, calls and quote requests.
- 2. Measure the starting point. Ask those questions in ChatGPT (with search on), Gemini or Google AI Mode, Copilot, Perplexity and Claude. Record who gets named, in what position, and above all which pages are cited as sources: those are the pages to match or to get mentioned on.
- 3. Open the door to crawlers. In robots.txt, allow at least OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Bingbot on public pages, and close only private areas. Also check your CDN, firewall and security plugins: Google reminds site owners that crawling must be allowed in robots.txt and by any CDN or hosting infrastructure.
- 4. Get indexed. Add the site to Google Search Console and Bing Webmaster Tools, submit your sitemap, and use IndexNow to report every new or updated page (a protocol supported by Bing, Yandex, Naver, Seznam and Yep, among others). Microsoft recommends IndexNow so that AI systems reference the most current version of a page.
- 5. Make your business impossible to confuse. Same name, same one-sentence description, same city, same phone on your site, Google Business Profile, Bing Places, directories and social profiles. Add Organization structured data with a stable @id and sameAs links to your profiles. If a company shares your name, say plainly that you are not it.
- 6. Publish one page per important question. First sentence: the answer. Then a table or list, real price ranges, steps, pitfalls, an FAQ and an updated date. Structured data must match the visible text.
- 7. Earn mentions elsewhere. Industry directories, your professional body, the chamber of commerce, local press, partners, reviews from real customers. It is the slowest lever; it cannot be bought or faked.
- 8. Add an llms.txt file (optional). A Markdown summary at the root of your site. It takes an hour and does no harm, but Google states plainly that it is not needed for its AI features. See our llms.txt guide.
- 9. Measure every month. Ask the same question list to the same engines, check Bing’s AI Performance report, Search Console, and visits from chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai.
What a “citable” page contains
AI engines reuse pages that answer cleanly. Microsoft, introducing its AI Performance report, advises clear headings, tables, FAQ sections, claims backed by current data, and accurate information; Google asks for helpful, people-first content with important content available in text.
| Element | Weak version | Citable version |
|---|---|---|
| First sentence | “Every project is unique.” | “A 5-page website redesign typically costs between $X and $Y in 2026.” |
| Pricing | “Contact us for a quote.” | A published range, before tax, with what makes it move |
| Proof | Adjectives (“best”, “leading”) | Named, linked sources, dates, conditions |
| Structure | One long block of text | Question-shaped headings, a table, numbered steps, an FAQ |
| Service area | Missing or vague | Cities and regions named; remote or on site, stated clearly |
| Freshness | No date | A visible updated date, numbers reviewed at least yearly |
Calling yourself “the best” does not make a model think so: it cross-checks what you say against what others say about you.
Pitfalls to avoid
- Paying for a “guaranteed spot in ChatGPT”. Nobody outside the AI companies controls the answers. A serious agency promises work and measurement, not a rank.
- Blocking without knowing it. A 2023 rule written to “keep AI out”, a CDN setting or a security plugin can shut out OAI-SearchBot or PerplexityBot. Blocking GPTBot (training) is a defensible choice; blocking OAI-SearchBot removes you from ChatGPT search.
- Publishing dozens of near-identical pages (one per city, one per keyword variant). Google calls generating many pages primarily to manipulate rankings “scaled content abuse”, whether the content is made by AI or by people.
- Hiding text aimed at AI crawlers (white-on-white text, hidden instructions). That is hidden text under Google’s spam policies and can get the page dropped.
- Inventing reviews, clients or statistics. Engines cross-check sources; one false fact gives them a reason to doubt everything else.
- Trusting a single test. Answers vary with wording, session and location. Run the same list several times and watch the trend over weeks.
- Letting your site contradict itself. An old price on a forgotten page, a business listing under another name: the model may repeat the wrong version.
How long it takes and what it costs
Technical fixes (crawlers, indexing, structured data, llms.txt) take a few days. In our experience, consistent citations generally take 3 to 6 months after that, and a clear effect on inbound leads 6 to 12 months; that is an estimate, not a guarantee, and third-party mentions are built continuously.
You can do steps 1 to 4 yourself with this guide. If you would rather hand it off, these are our published prices. We work remotely from Québec City for clients in Canada and the United States, in English or French.
| ZeniTech service | Price (CAD, before tax) | What it covers |
|---|---|---|
| GEO audit | $1,500 | Testing ChatGPT, Claude, Gemini and Perplexity on your market, AI crawler access, entity consistency, structured data and llms.txt, action plan |
| GEO only | $1,200/month | For a site that already has solid SEO and wants to be cited by AI engines |
| SEO + GEO package | $2,000/month | Found on Google and recommended by AI engines, one monthly report |
| Express redesign of an existing site | From $599 | When the site itself keeps it from being read or cited |
Prices as displayed on zenitech.dev/en/pricing on October 2, 2026, in Canadian dollars, before taxes.
What ZeniTech does
ZeniTech is a web, app, automation and AI agency based in Québec City, founded by Alexandre Blais. We apply this method to our own sites first: answer-engine crawlers allowed on public pages, structured data linked by a single @id, an llms.txt file, IndexNow on every new page, and sourced, dated guides like this one.
An engagement starts with step 2 on your real questions: we show you who the engines name today, which pages they cite, and what stands between you and them. We do not promise a place in any AI answer; we promise the work, the measurement and a clear report.
Frequently asked questions
How do I get ChatGPT to recommend my business?
Allow its search crawler, OAI-SearchBot, make sure your site is properly indexed by search engines, describe your business the same way everywhere, and publish a page that directly answers the question your customers ask, with numbers and sources. Add mentions on other sites (directories, press, real reviews). Nobody can guarantee the outcome, but these conditions are necessary.
Can I pay to appear in ChatGPT answers?
Be wary of any offer of a “guaranteed spot” in an AI engine’s answers: no agency controls them. What can be done is to make your pages readable, understandable and citable, then measure whether the engines name you more often.
Should I block GPTBot to protect my content?
You can. According to OpenAI, GPTBot is used for model training, while OAI-SearchBot powers ChatGPT search. Blocking GPTBot does not keep you out of ChatGPT search; blocking OAI-SearchBot does. For most small businesses, allowing both on public pages is the simplest choice.
Is an llms.txt file enough to show up in ChatGPT?
No. It is a useful, low-cost convention, but to our knowledge no major AI company has announced that it uses llms.txt to choose its sources, and Google says no special AI text file is needed for AI Overviews. Crawler access, indexing and pages that answer cleanly matter far more.
How long until ChatGPT mentions my business?
Technical fixes take days, and OpenAI says robots.txt changes take about 24 hours to apply to its search crawler. For consistent citations, expect generally 3 to 6 months, and 6 to 12 months for a clear effect on leads; that is our estimate, not a guarantee.
How do I know whether AI engines already mention my business?
Ask ChatGPT, Gemini, Copilot, Perplexity and Claude the questions your customers ask, without naming your business, and record who gets recommended and which pages are cited. Then check the AI Performance report in Bing Webmaster Tools and visits from chatgpt.com or perplexity.ai in your analytics.
Does GEO replace SEO?
No, it builds on it. Google AI Overviews rely on Google’s search index, Copilot on Bing’s, and ChatGPT search on third-party search providers. A poorly indexed site has little chance of being cited by an AI engine.
Do I need a page for every city to be recommended locally?
No. Dozens of near-identical pages can be treated by Google as scaled content abuse. You are better served by one strong page per service that clearly names your service area, accurate Google Business Profile and Bing Places listings, and real local mentions.
Related
Free consultation
Get a fixed price for your project in 30 minutes.
We look at what the project has to achieve, tell you what it costs, and tell you when you do not need it.
Book a call →+1 581-748-7017Sources
- OpenAI: Overview of OpenAI crawlers (OAI-SearchBot, GPTBot, ChatGPT-User)
- OpenAI Help Center: ChatGPT search (where results come from)
- Google Search Central: AI features and your website
- Google Search Central: Google’s common crawlers (Google-Extended)
- Google Search Central: spam policies (scaled content abuse, hidden text)
- Google Search Central: creating helpful, reliable, people-first content
- Google Search Central: Organization structured data
- Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools (February 10, 2026)
- IndexNow: FAQ (participating search engines)
- Perplexity: Perplexity crawlers
- Anthropic: Anthropic crawlers and how to block them
- llmstxt.org: the llms.txt proposal