Skip to content
Start trialSee plans

Playbook · 10 min read

How to get ChatGPT to recommend your business: a practical playbook

By Adgeeks by Accoladesigns · Published

The short answer

ChatGPT recommends businesses it can find, understand and verify. To get recommended: let OpenAI's search crawler (OAI-SearchBot) read your site, publish clear pages for each service and location, keep your business facts identical everywhere, back your claims with specific evidence, earn mentions and reviews on sites AI assistants trust, and then track a fixed set of customer prompts to see whether you are named. Nobody can guarantee a recommendation, but these steps give the model the evidence it needs to choose you.

Key takeaways

  • ChatGPT search relies on OpenAI's OAI-SearchBot; if robots.txt blocks it, your site will not be shown in ChatGPT search answers.
  • Blocking GPTBot only opts you out of model training; it is a separate decision from search visibility.
  • Assistants recommend what they can corroborate, so your name, services, locations and proof must read the same on every site that mentions you.
  • Specific evidence (numbers, case studies, credentials, sources) gets quoted; vague claims get ignored.
  • Track 20 to 50 real customer prompts across assistants every week, and change one thing at a time.

How ChatGPT decides who to recommend

When someone asks ChatGPT for a recommendation, it answers from two places: what the model learned during training, and, increasingly, what it finds by searching the web at the moment of the question. The second part matters most for businesses, because it is current, it can be influenced, and it comes with sources.

OpenAI runs separate crawlers for these jobs. OAI-SearchBot finds pages for ChatGPT's search features. GPTBot collects content that may be used to train future models. ChatGPT-User fetches a page when a user's question needs it. OpenAI states that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and that blocking GPTBot signals your content should not be used for training (OpenAI crawler documentation). The two are independent: you can stay out of training and still be visible in search.

Other assistants work in a similar way. Perplexity, Gemini, Copilot, Claude and Google's AI Overviews all retrieve web sources and then write an answer from them. The playbook below therefore works across all of them, not just ChatGPT.

Step 1: Let the right crawlers in

Open yoursite.com/robots.txt and check that search and AI crawlers are not blocked from your public pages. At a minimum, allow Googlebot, Bingbot, OAI-SearchBot, ChatGPT-User, PerplexityBot and ClaudeBot. Keep private areas (logins, carts, dashboards) disallowed for everyone. Also check that your CDN or firewall is not silently challenging these bots; a page that returns a CAPTCHA to a crawler is as good as invisible.

Step 2: Make every service and location its own clear page

Assistants match questions to pages. If you offer five services in three cities and have one generic "Services" page, you give the model nothing specific to cite. Create one page per core service, and one per location you genuinely serve, each opening with a plain answer: what it is, who it is for, where, and what it costs or how pricing works.

  • Use the words customers use ("root canal treatment in Andheri", not "endodontic solutions").
  • Put the answer in the first paragraph, then the detail.
  • Add an FAQ with the real follow-up questions: price, timing, guarantees, what to bring.
  • Link related pages together so crawlers and readers can move between them.

Step 3: Make your business facts identical everywhere

Before an assistant recommends you, it effectively asks: is this business real, and is it what it says it is? It answers that by comparing sources. If your website, Google Business Profile, LinkedIn page, directory listings and press mentions disagree on your name, services, phone number or the areas you serve, the model has less confidence and is more likely to name someone else.

  1. Write a single "brand facts" sheet: legal and trading name, one-line description, services, locations served, founding year, phone, email, website, social profiles.
  2. Update every profile you control to match it exactly.
  3. Add Organization (or LocalBusiness) structured data to your site that matches the visible facts, including links to your official profiles in sameAs.
  4. Fix or claim listings on the directories that rank for your category.

Step 4: Give the model evidence it can quote

Generative engines prefer content they can lift with confidence. The research that coined the term GEO found that adding citations, quotations from credible sources and statistics increased visibility in generative answers by up to 40%, and up to 37% on Perplexity specifically, while keyword stuffing made things worse (GEO: Generative Engine Optimization).

For a business, "evidence" looks like:

  • Named case studies with specific results and timeframes (see how we present ours on Results).
  • Credentials, certifications, awards and memberships, with links to where they can be verified.
  • Clear pricing, or a clear explanation of how pricing works.
  • Original data: survey results, benchmarks, before-and-after numbers.
  • Expert authorship: who wrote the page and why they are qualified.

Step 5: Earn mentions on sites assistants already trust

Look at what the assistants cite today. Ask ChatGPT, Perplexity and Gemini the questions your customers ask, and note the sources they link: industry directories, review platforms, "best of" round-ups, news sites, Reddit threads. Those are the places your business needs to appear. For agencies that might be Clutch, GoodFirms and DesignRush; for a restaurant, Google reviews, Zomato and local food guides; for software, G2 and Capterra.

Reviews deserve special attention. Assistants frequently summarise sentiment ("praised for fast response times"), so a steady flow of recent, detailed reviews on the platforms that matter in your category shapes not only whether you are recommended but how you are described.

Step 6: Write the answers AI is looking for

Many recommendation prompts are comparisons: "X vs Y", "best X for Y", "is X worth it". If the only comparison content about your category is written by competitors, that is what the assistant reads. Publish honest comparison and buying-guide content on your own site, answer the objections customers raise, and be specific about who you are and are not a good fit for. Assistants are more willing to recommend a business that clearly states its niche.

Step 7: Track prompts, not just keywords

Build a list of 20 to 50 prompts that real customers would type, across awareness ("how do I..."), comparison ("best... in...") and brand ("is [your name] good?"). Run them regularly in each assistant you care about and record four things: whether you are mentioned, your position in the answer, the sentiment, and the sources cited. Because AI answers vary from run to run, look at trends over weeks, not single responses.

This is the core of our GEO service: every client gets tracked prompts across up to seven AI engines, with mentions, citations and share of voice in a live portal.

What not to do

  • Do not stuff pages with "best [service] in [city]" phrases; research shows it lowers visibility in generative answers.
  • Do not publish fake reviews or invented statistics; assistants cross-check, and platforms remove them.
  • Do not hide key information in images, PDFs or JavaScript-only widgets that crawlers may not read.
  • Do not expect a single trick (a file, a tag, a prompt) to fix it. Recommendation is earned from consistent evidence over time.

Checklist

  1. robots.txt allows OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Googlebot and Bingbot on public pages.
  2. One clear, answer-first page for each service and each location you serve.
  3. A brand facts sheet, mirrored on every profile and in Organization structured data.
  4. Evidence on every key page: numbers, case studies, credentials, sources.
  5. Listings and reviews on the platforms assistants cite in your category.
  6. Honest comparison and buying-guide content.
  7. A tracked prompt set, reviewed weekly.

Want to know where you stand first? Our free AI Visibility Audit checks whether ChatGPT, Gemini, Perplexity and Google's AI features mention you, who they name instead, and what to fix first.

FAQ

Questions, answered.

Can I pay to be recommended by ChatGPT?

Organic answers in ChatGPT cannot be bought. What you can do is make your business easy to find, understand and verify, which is what the assistant uses to decide who to mention.

Should I block GPTBot?

It depends on whether you want your content used for model training. Blocking GPTBot does not remove you from ChatGPT search; OAI-SearchBot controls that. Many businesses allow OAI-SearchBot and make a separate decision about GPTBot.

Does an llms.txt file help ChatGPT recommend me?

llms.txt is an optional proposal for giving AI systems a clean summary of your site. It is cheap to add and can help tools that read it, but no major assistant has said it uses it for ranking, and Google states no special files are needed for its AI features. Treat it as a nice-to-have, not a strategy.

How quickly can I get recommended?

Crawler access and fixes to facts can take effect within weeks as pages are re-crawled. Earning third-party mentions and reviews, which most strongly influence recommendations, typically takes a few months of consistent work.

Next step

Want this done for you?

Every Adgeeks plan covers SEO, AEO and GEO with a live portal. Start with a 7-day full refund, or get a free AI Visibility Audit first.