How to get your business recommended by ChatGPT

More people now ask an AI assistant for a recommendation before they search Google: “Who’s a good emergency plumber in Austin?”, “What’s the best invoicing tool for freelancers?”, “Where should I get coffee beans shipped from?” If the answer names three businesses and yours is not one of them, you lose that customer without ever knowing they looked.

There is no form to submit and no way to pay ChatGPT to recommend you in its answers. But AI recommendations are not random either. They come from what these systems can read about you and how often trusted sources mention you. This guide covers how that works and what to do about it.

How ChatGPT decides what to say

ChatGPT builds an answer from two sources:

1. What the model already knows. The model was trained on a large amount of public web text up to a cutoff date. Well-known brands that are mentioned often across the web are “remembered”. A small local business usually is not.
2. What it finds when it searches. For questions about current or local things (“best plumber near me”, “open now”, “2026 pricing”), ChatGPT often searches the web, reads a handful of pages and writes an answer with links to its sources. OpenAI has said ChatGPT search uses third-party search providers as well as its own crawler. Microsoft Bing is widely reported to be one of those providers.

The second source is where most businesses can make progress quickly, because it depends on things you can influence: whether your pages can be crawled, whether they rank in the underlying search results, and whether other sites mention you.

Other assistants work in a similar way. Perplexity searches the web for nearly every answer and cites its sources. Google’s AI Overviews and AI Mode draw on Google’s own index, so classic Google SEO feeds directly into them. Gemini, Claude and Copilot each mix model knowledge with live search.

Step 1: Let AI crawlers read your site

This is the step people skip, and it is the one most often broken.

AI companies use separate crawlers for different jobs. For OpenAI:

• OAI-SearchBot finds pages to show in ChatGPT search answers. OpenAI says sites that block it will not be shown in those answers.
• GPTBot collects content for training future models. Blocking it does not stop you appearing in search answers.
• ChatGPT-User fetches a page when a user asks ChatGPT to look at it.

Other companies have similar crawlers, such as PerplexityBot, ClaudeBot and Claude-SearchBot. You can decide separately whether to allow training and whether to allow search. Many businesses block training crawlers but allow search crawlers, because search is what sends customers.

Check two places:

• robots.txt. Look for User-agent: OAI-SearchBot or a blanket User-agent: * with Disallow: /. See our robots.txt entry.
• Your firewall, CDN or security plugin. Many block “unknown bots” by default and return a 403 error or a challenge page, even when robots.txt allows them. This is the most common hidden problem we see.

The free AI crawler checker tests both for the main AI crawlers in one go.

Step 2: Rank in the search results AI assistants use

When ChatGPT searches, it reads pages that rank in the search results it gets back. If your site does not rank for “emergency plumber Austin” in normal search, it is unlikely to be one of the pages read.

That makes ordinary SEO the foundation of AI visibility:

• Pages for each service and location, with clear titles. See our local SEO checklist.
• No crawl or indexing problems. Our SEO audit guide covers the checks.
• Set up Bing Webmaster Tools as well as Google Search Console. Submit your sitemap there and check that Bing has indexed your important pages. Many sites never look at Bing, and some find whole sections missing from its index.

Step 3: Say clearly who you are, what you do and where

AI assistants summarize. If your homepage says “Solutions for every need” and nothing more, there is nothing useful to summarize.

Make sure your site states plainly:

• What the business is (“Coastline Plumbing is a licensed plumbing company in Austin, Texas”).
• What you offer, with a page for each main service or product.
• Who you serve and where (towns, industries, customer types).
• What makes you different, in specific terms: 24/7 service, a particular specialty, a guarantee, years in business. “High quality” and “customer focused” say nothing.
• Prices or price ranges where you can. Questions like “how much does a water heater replacement cost in Austin” are common, and assistants prefer sources that answer them.

Use schema markup for your organization or local business, with name, address, phone, hours, service area and links to your official profiles. It does not guarantee anything, but it removes ambiguity about basic facts. The free schema generator builds it for you.

Step 4: Get mentioned by sources AI assistants trust

When ChatGPT answers “best plumber in Austin”, it often relies on pages that compare or list businesses: local “best of” lists, review sites, news articles, community forums and Reddit threads. For software, it leans on comparison articles, review platforms and discussions.

So the question becomes: where are people already discussing businesses like yours, and are you there?

• Reviews. A steady flow of genuine reviews on Google and the review sites that matter in your industry. Assistants often summarize review sentiment.
• Local and industry lists. Find the “best [service] in [city]” articles that rank in Google and Bing. Some accept suggestions; some are built from review data.
• Press and community. Local news, sponsorships, industry associations and trade publications. These also earn backlinks, which help your normal rankings.
• Consistent details everywhere. If your NAP differs across directories, assistants may give the wrong phone number or merge you with another business.
• Honest participation in communities. Answering questions on Reddit or forums as yourself, without spamming links, gets your name into the discussions assistants read. Fake accounts and planted reviews are against the platforms’ rules and, in the US, can break FTC rules on fake reviews.

Step 5: Write content that answers real questions

Assistants like pages that answer a specific question directly, then give detail. That also happens to be what works in Google.

• Start with the answer, then explain. “A tankless water heater installation in Austin usually costs [your real price range], depending on…” beats three paragraphs of history.
• Use clear headings that match how people ask.
• Show experience: real examples, your own photos, named authors with credentials. See E-E-A-T.
• Keep facts current. An assistant that finds your 2022 price list may quote it.

Step 6: Consider llms.txt, but do not expect much

llms.txt is a proposed file at the root of your site that gives AI tools a short, plain summary of the site and links to key pages. It is cheap to add and harmless, but none of the major AI companies have said their assistants rely on it for answers. Treat it as a nice extra after the steps above, not a fix. The free llms.txt generator creates one in a few minutes.

Step 7: Measure it properly

AI answers change from one run to the next, and between assistants. Asking ChatGPT once and seeing your name (or not) tells you very little.

A better approach:

1. Write down 10 to 30 questions your customers really ask. Mix broad (“best dentist in Columbus”) and specific (“dentist in Dublin Ohio that takes Delta Dental”).
2. Check them regularly across assistants: ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews.
3. Record whether you are mentioned, whether you are cited with a link, and who is mentioned instead. The competitors that keep appearing show you which sources the assistants trust.
4. Look at trends over weeks, not single answers.

RankThat’s AI Visibility tracks this for questions about your services and area across those five assistants and compares you with competitors, so you can see whether the work above is moving anything.

What not to do

• Do not stuff pages with “best [service] in [city]” hoping an assistant repeats it. It reads as spam to people and search engines.
• Do not create fake reviews or fake “best of” lists. It breaks platform rules, can break consumer protection law, and tends to be found out.
• Do not block all AI crawlers by accident through a firewall rule and then wonder why you never appear.
• Do not ignore Google. AI Overviews and AI Mode depend on Google’s index, and most other assistants depend on search engines too. Good SEO remains the base.

A realistic timeline

Fixing crawler access can make a difference as soon as your pages are recrawled. Showing up for broad recommendation questions takes longer, because it depends on reviews, mentions and rankings that build over months. Model knowledge updates only when new models are trained, so being “known” by the model itself is the slowest part of all.

Start with access and clarity, build mentions steadily, and measure across many questions. For software companies, our page on SEO for SaaS covers comparison and alternatives content, which assistants lean on heavily for tool recommendations.