How to get your business recommended by ChatGPT and other AI assistants

Last updated 16 October 2026

By the FocalScore team at Astrolabix LLC. First published 9 October 2026. Sources last checked October 2026.

The short answer

An AI assistant can only recommend a business it can find, read and believe. To give yourself the best chance: let the assistants' search crawlers into your site, say plainly on your pages what you do, for whom and where, keep your name, address and phone number identical everywhere they appear, and earn genuine mentions and reviews on other sites. Nobody can guarantee that an assistant will name you, and anyone who promises it is guessing.

Where an assistant's answer comes from

When someone asks ChatGPT, Claude, Perplexity or Google's AI "who's the best electrician near me?", the answer comes from two places: what the model learned in training, and pages it looks up at the moment of the question. For a local business, the lookup is the part you can influence, and each company runs separate crawlers for each job:

OpenAI uses OAI-SearchBot to find sites for ChatGPT's search answers and GPTBot to collect training data. Its crawler documentation says that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and that blocking GPTBot is a training opt-out, controlled separately.
Anthropic runs Claude-SearchBot (search quality), Claude-User (fetches a page when someone asks Claude a question) and ClaudeBot (training). Anthropic says blocking either of the first two may reduce your site's visibility in Claude's answers.
Perplexity uses PerplexityBot to surface and link websites in its answers, and says it is not used to collect content for training AI models.
Google's AI Overviews and AI Mode link to pages from Google Search. Google says a page must be indexed and eligible to show in Search with a snippet to be used as a supporting link, and that there are no additional requirements or special optimisations for it.

The crawlers by name, as each company documents them. The names are what you write in robots.txt or allow in a firewall rule:

User agent	Company	What it feeds	If you block it
`OAI-SearchBot`	OpenAI	ChatGPT's search answers	OpenAI says your site won't be shown in ChatGPT search answers.
`GPTBot`	OpenAI	Training OpenAI's models	A training opt-out, controlled separately from search answers.
`Claude-SearchBot`	Anthropic	The quality of Claude's search results	Anthropic says it may reduce your visibility in Claude's answers.
`Claude-User`	Anthropic	Pages Claude fetches when someone asks it a question	Anthropic says it may reduce your visibility in Claude's answers.
`ClaudeBot`	Anthropic	Training Anthropic's models	A training opt-out.
`PerplexityBot`	Perplexity	Finding and linking websites in Perplexity's answers; Perplexity says it does not collect training data	The crawler Perplexity uses to surface and link sites can't read your pages.
`Googlebot`	Google	Google Search, which AI Overviews and AI Mode link to	Google says a page must be indexed and eligible for a snippet to be a supporting link.
`Google-Extended`	Google	Training and grounding Gemini models	Google says it does not affect your inclusion or ranking in Google Search.
`(none: Brave follows Googlebot)`	Brave	Brave Search	Brave says that if Googlebot can't crawl a page, its crawler won't either.

OpenAI and Perplexity also run fetchers that visit a page only when a person asks the assistant about it (ChatGPT-User and Perplexity-User). Their pages linked above explain how those differ from the crawlers in the table.

Brave Search is the exception to naming a crawler. Brave says its crawler does not use a user agent of its own, and that it will not crawl a page Googlebot is not allowed to crawl. So there is no Brave line to write in robots.txt: your Googlebot rules decide it.

So the work is mostly the work of being a clear, findable, trustworthy website. There is no secret file or tag that makes an assistant pick you.

1. Let the answer crawlers in

If your robots.txt or your firewall turns away OAI-SearchBot, Claude-SearchBot or PerplexityBot, those assistants can't read your pages, however good they are. This happens more than owners expect: a "block AI" snippet copied from a forum, a CDN setting or a security plugin can shut out the crawlers that answer questions along with the ones that train models.

Blocking the training crawlers (GPTBot, ClaudeBot, Google-Extended) is a legitimate choice and does not remove you from answers. Blocking the search and user crawlers does. Run the free AI crawler checker to see which ones your site lets in, and read why AI assistants can't see your website for each kind of block and its fix.

2. Put specific answers on pages that can be indexed

An assistant quotes and links to pages. It needs a page that answers the question someone actually asked, in words it can lift out. Vague pages give it nothing to work with.

Say what you do in the first lines of the homepage. "Welcome to our website" tells an assistant nothing. "Emergency plumber in Leeds, open 24 hours, no call-out fee" tells it who, what and where in one line.
Give each main service its own page, with what it includes, who it suits, the area you cover and a price or how pricing works.
Answer the questions buyers ask, under headings phrased as those questions: "How much does a boiler service cost?", "Do you cover Harrogate?". Put the answer in the first sentence under the heading.
State facts an assistant can quote: years trading, qualifications and registrations, areas covered, opening hours. Only true ones.
Make sure the pages can be indexed: no noindex tag, nothing behind a login, and the text present in the page itself rather than drawn in by JavaScript afterwards.

Google's guidance on helpful, people-first content asks the same questions an assistant effectively asks: does the page offer original information, and is it clear who wrote it and why?

3. Keep your business details the same everywhere

Anything that reads the web, a search engine or an assistant, sees every place your business is listed, not just your website. If your website says "Smith & Sons Plumbing, Unit 4", your Google Business Profile says "Smith Plumbing Ltd" and a directory has an old phone number, you look like three businesses, or an unreliable one.

Use one business name, one address format and one phone number, everywhere.
Claim and complete your Google Business Profile. Google's own tips for local ranking start with complete, accurate, up-to-date business information, including hours and categories.
Put the same details on your site in the footer and on the contact page, and consider describing them in structured data (an Organization or LocalBusiness block). Google says its AI features need no special markup, but structured data states your facts in a form any machine reads the same way.

4. Be mentioned on other sites

An assistant that searches the web before answering finds other people's pages about you as well as your own. For local results, Google says prominence takes into account information such as how many websites link to your business and how many reviews you have, and that more reviews and positive ratings can help. Useful places to be mentioned:

Review platforms your customers already use, with replies from you.
Trade bodies, accreditation schemes and suppliers' "find an installer" lists.
Local directories, chambers of commerce and local news.

Ask real customers for reviews. Never buy or invent them: in the US, the Federal Trade Commission's rule bans fake reviews and testimonials, including ones attributed to people who don't exist.

What you can't control

Which sources an assistant picks. It decides, and the same question can get a different answer tomorrow, or from a different person.
What it learned in training. A model may describe your business from older pages it learned from, until its search results correct it.
Paying for a place. Google says there is no way to request or pay for a better local ranking. Be wary of anyone selling guaranteed AI recommendations.

What you can control is whether you are readable, clear, consistent and mentioned. That is the whole job.

How to check where you stand

Ask the assistants yourself. Type the question a customer would ("best roofer in Bristol for flat roofs") into ChatGPT, Claude, Perplexity and Google, and note who is named and which pages are cited.
Run the free AI crawler checker to confirm the answer crawlers can reach your site.
Run the free scan below: its AI readiness checks cover the points in this guide, page by page. The optional $49 AI Advanced Report puts real buyer questions to an AI assistant and records whether you were mentioned, and who was recommended instead.

Checklist

OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User and PerplexityBot are allowed in robots.txt.
Your firewall, CDN and security plugins don't block those crawlers.
Your important pages have no noindex tag and their text is in the page without JavaScript.
The homepage says what you do, for whom and where, in its first lines.
Each main service has its own page, with price information and the area you cover.
Common buyer questions are answered under question headings.
Name, address and phone are identical on your site, your Google Business Profile and directories.
You ask real customers for reviews and reply to them.

What the free scan checks for this

These are the FocalScore checks that cover this guide's subject. Each one in your report says what it found, shows the evidence and gives the fix.

AI crawlers allowed
Search crawlers allowed
Indexability
Opening states what you do
Buyer questions covered
Facts an assistant can quote
Entity / structured data for AI
Google Business / NAP
Testimonials / reviews on site

Sources

Last checked October 2026. Documentation, settings and rules change; if one of these pages now says something different, it is right and this guide is out of date. Tell us at info@focalscore.com.