How to get ChatGPT to recommend your business (without tricks)

Why ChatGPT recommends some firms and not yours

Most owners assume ChatGPT simply knows which firms are good. For questions about local or current businesses, that is mostly not how it works. When search is involved, the assistant runs one or more web searches, reads a handful of the pages that come back, and writes an answer grounded in them. If your firm is not among the pages it retrieves, or those pages do not clearly say what you do and where, you will not be named, however good your work is.

Google describes the same pattern for AI Overviews and AI Mode. Its guide to generative AI features says they are rooted in the core Search ranking and quality systems, use retrieval-augmented generation to pull pages from the Search index, and use query fan-out: a set of related searches the model issues to gather more results. A single question such as best residential architect near Palo Alto can quietly become several searches about ADU specialists, permitting experience and client reviews.

The practical consequence is simple. Being recommended by AI is downstream of three things: being findable in search, being clearly understood, and being vouched for by others. There is no separate back door.

How ChatGPT search, Google AI Overviews and Perplexity pick sources

Each tool retrieves differently, but all of them start from a search index and then apply model judgment. This is what you can and cannot influence:

Why your own website alone rarely wins

A 2025 study by Chen et al. found that AI search engines strongly favor earned media (third-party coverage) over brand-owned content, and that niche brands have to overcome a bias toward big, well-known brands. For a twelve-person firm, that means best-firm-in-my-city questions are often answered from directories, local press, professional association listings, award pages, review platforms and roundup articles rather than from firm homepages.

Your site still matters. It is where a model verifies the details (services, service area, project types, credentials) and where interested clients land. But it is rarely enough evidence on its own.

The stakes are rising. Pew Research Center found that when a Google AI summary appeared, users clicked a traditional search result in 8% of visits, compared with 15% when no summary appeared, and clicked a link inside the summary in only 1% of visits. Being named in the answer itself now matters more, because fewer people click through to compare firms.

What demonstrably helps

None of these are tricks. They are the same things that make a firm easy for a human researcher to shortlist, done thoroughly.

• Be indexable everywhere that matters. Google says a page must be indexed and eligible to show a snippet to appear in its AI features. Check that robots.txt does not block OAI-SearchBot, which OpenAI's crawler documentation says is used to surface sites in ChatGPT search, or PerplexityBot. GPTBot is OpenAI's separate training crawler, so you can decline training and still allow search. Verify your site in Google Search Console and Bing Webmaster Tools.
• State clear entity facts in plain HTML. Your exact firm name, what you do, who you serve, which cities or regions you cover, licenses, principals and contact details. Use the same wording on your homepage, About page, footer, Google Business Profile, LinkedIn and directory listings. Inconsistent facts make models hedge or confuse you with a similarly named firm.
• Build specific service and location pages that deserve to exist. A page on seismic retrofit engineering for San Francisco soft-story buildings, with real scope, process, project examples and the questions clients actually ask, beats a generic Services page. Do not mass-produce near-identical city pages; Google treats that as scaled content abuse.
• Earn third-party mentions honestly. Project features in trade or local press, award submissions, professional association and chamber listings, talks, case studies published by clients, and collaborators who link to you. These are the earned sources AI answers lean on.
• Keep reviews and profiles real and current. Google says Business Profiles help businesses be visible in AI responses. Keep Google Business Profile, Bing Places and the directories your clients actually use accurate, and ask satisfied clients for reviews.
• Publish citable facts. Specific, checkable statements such as typical permit timelines in your jurisdiction, project types and sizes, and credentials, with sources where relevant. In the GEO benchmark by Aggarwal et al. (KDD 2024), adding citations, quotations and statistics improved source visibility in generative answers by up to roughly 40% on their metric. That is a lab result, not a guarantee, but it matches common sense: specific claims are easier to cite than adjectives.
• Use structured data as support, not magic. Organization, ProfessionalService or Service JSON-LD that matches your visible content helps disambiguation and rich results. Google states that no special schema is required for its AI features.

What is hype, and what can backfire

• llms.txt as a ranking lever. llms.txt is a proposal for a plain-text site summary aimed at language models. Google says Search ignores it, and its effect elsewhere is unproven. It is cheap and optional; trugoods.net publishes one, but we do not count it as a visibility lever. See our llms.txt guide.
• Rewriting everything into AI-friendly chunks. Google says there is no need to chunk or rewrite content for AI. The C-SEO Bench study (2025) found that most conversational SEO content tricks were ineffective or harmful in its tests, traditional ranking dominated, and any gains shrank as more sites adopted them.
• Guaranteed ChatGPT ranking offers. Nobody outside OpenAI or Google controls these systems. Google explicitly warns against tools that claim access to internal ranking or AI metrics.
• Manipulation. Fake or purchased reviews, paid best-firms listicles, hidden prompt text aimed at AI, cloaking (showing bots different content from people) and mass-generated pages. These violate search spam policies, can get pages dropped from the index, and destroy trust with the very clients you want if discovered.
• Chasing inauthentic mentions. Google says seeking inauthentic mentions across the web is not as helpful as it might seem, because its generative features depend on both quality ranking and spam systems.

A 30-day action plan for a small firm

This plan assumes a few hours a week from an owner or office manager, plus whoever maintains your website. Technical fixes are usually picked up within weeks of being crawled; earned mentions and reputation take months, so treat day 30 as a first checkpoint, not a finish line.

1. Days 1-3: Record a baselineWrite 10-15 prompts a real client would type, combining service, location and project type. Run each prompt 5-10 times in ChatGPT with search on, Google AI Mode or AI Overviews, and Perplexity, logged out where possible. Record whether you are named, whether your site is cited, which competitors are named, and which third-party pages are cited.
2. Days 3-5: Check access and indexingReview robots.txt and any CDN bot-blocking settings, run a site: search on Google and Bing, and verify the site in Search Console and Bing Webmaster Tools with a sitemap submitted. Our AI visibility audit walks through each check.
3. Days 6-10: Clean up entity factsWrite one canonical description of the firm: name, services, service area, address, phone, principals, licenses. Update the website, Google Business Profile, Bing Places, LinkedIn and key directories so they match.
4. Days 11-18: Rebuild two or three service pagesChoose your highest-value services. Give each a page with the concrete scope, your process, project examples with photos, credentials and answers to questions clients really ask. Make sure the text is in the HTML, not only rendered by scripts.
5. Days 19-24: Add citable facts and structured dataAdd specific, verifiable facts to those pages. Add Organization and Service JSON-LD that matches the visible content, then validate it with Google's Rich Results Test.
6. Days 25-30: Start earning mentions and reviewsGo back to the third-party pages AI answers cited in your baseline. For each, find the legitimate path in: a directory listing, an association membership, an award entry, a project story pitched to local press. Ask three to five recent clients for an honest review. Then re-run the full prompt panel and compare.

How to measure progress without fooling yourself

AI answers vary from run to run, by location, by logged-in state and by model version. One screenshot proves nothing, good or bad. Track share of answers instead: out of N runs of a prompt, how many name your firm and how many cite one of your pages. A simple spreadsheet with date, tool, prompt, run number, named (yes or no), cited URL and competitors named is enough.

For Google, use the Generative AI performance report in Search Console, which Google recommends for measuring visibility in its AI features, alongside impressions and clicks for your service queries. Add an AI assistant option to your how-did-you-hear-about-us question so you capture referrals that analytics cannot see.

Hold any vendor, including us, to the same standard: a stated method and date. For example, in a 7-query Chinese-language Google check on 2026-10-04 (non-personalized desktop, San Jose) for a Bay Area design studio we worked with, 3 of 7 queries showed the studio on page one (positions 3, 4 and 7) and 2 of 7 AI Overviews cited it. That is a dated snapshot, not proof of lasting results. Details are in our case study.

Where to go next

If you run an architecture or design practice, read AI search for architecture firms for sector-specific examples. If you want AI agents, not just AI answers, to be able to act on your site, see our WebMCP guide. For how we approach technical SEO and generative engine optimization together, see SEO and GEO at TruGoods.

Frequently asked questions

How do I get AI to recognize my business?

Make your basic facts unambiguous and identical everywhere: exact firm name, services, service area, address, principals and contact details on your website, Google Business Profile, Bing Places, LinkedIn and directories. Put those facts in the page HTML, and optionally in Organization or ProfessionalService structured data that matches the visible text. Then ask ChatGPT what your firm does; wrong details usually point to the inconsistent source.

How do I get AI to suggest my business?

There is no way to pay for an organic recommendation inside ChatGPT's answers, and anyone selling guaranteed placement is selling something else. AI tools suggest firms that their searches can retrieve and that independent sources vouch for. Be crawlable and indexed, publish specific service and location pages, and earn real mentions and reviews.

How long does it take to show up in ChatGPT search results?

Technical fixes such as unblocking OAI-SearchBot or getting indexed in Bing can be reflected within weeks once your pages are recrawled. Earned mentions and reviews build over months. Nobody can promise a specific timeline because retrieval and model behavior change.

Should I block GPTBot to protect my content?

That is a business decision. According to OpenAI's documentation, GPTBot is used for model training and OAI-SearchBot is used for ChatGPT search, and they are controlled separately in robots.txt. Blocking GPTBot does not by itself remove you from ChatGPT search, but blocking OAI-SearchBot may.

Does llms.txt help ChatGPT recommend my business?

There is no public evidence that it does. Google says Search ignores llms.txt, and other providers have not documented using it for ranking. It is harmless and cheap to add, but it should not replace indexing, clear facts and earned mentions.

Why does ChatGPT give a different answer every time I ask?

Answers are generated, not looked up, and each run can trigger different searches and weigh sources differently. Location, account state and model updates also change results. That is why you should test the same prompts repeatedly and track how often you appear rather than trusting one result.

Related guides

• Is my website visible to AI? A 30-minute AI visibility audit
• SEO and AI search for architecture firms: how to become the studio ChatGPT and Google can confidently recommend
• What is llms.txt, and is it worth adding to your website?
• WebMCP: how to make your website agent-ready, and how it differs from MCP