How to Get Your Business Recommended by Claude: 2026 Guide

TL;DR

There is no Claude Business Profile to claim. When Claude searches the web, it cites the pages it used, so your job is to be findable and easy to verify.
Anthropic runs three bots. Blocking ClaudeBot (training) is a separate choice from blocking Claude-SearchBot and Claude-User, which affect search visibility.
Anthropic has not officially named its search provider, but Brave Search is on its subprocessor list, and Brave will not crawl pages Googlebot cannot reach.
After access, the work is plain: a page that answers the buyer's question, the same facts on every listing and directory, and reviews that describe real jobs.
Nobody can promise Claude will name you, or when. Check by hand each month and log what you see.

How do I get my business recommended by Claude?

You can't claim or buy a spot. When Claude searches the web, it cites the pages it used. Make your business easy to find and easy to verify. Let Anthropic's search bots in. Keep your site crawlable by Googlebot, since Brave follows those rules. Publish a page that answers the buyer's question. Keep your facts identical everywhere. Then check monthly.

No one can promise Claude will name you, or when. That includes me. The rest of this guide separates what is documented, what is only reported, and what is my opinion.

How Claude finds a business

Claude answers from two places. One is what the model learned in training. The other is live web search. You can only influence the second one on your own schedule. Training data changes when Anthropic trains new models.

What happens when Claude searches

Anthropic's help center explains the basics. When a question benefits from current information, Claude uses a search tool to ground its answer in live web content, and every response includes citations so you can check the sources (Anthropic Help Center).

The developer docs say the same thing from the builder's side. The web search tool gives Claude real-time web content beyond its knowledge cutoff, and the response includes citations for sources drawn from search results (Claude Platform Docs).

That matters for one simple reason. Claude can only cite a page it found. Thin, missing or conflicting facts give it nothing to point to.

Claude also runs inside other apps

Many companies build Claude into their own products through the API or cloud platforms. Google Cloud's documentation says that turning on this feature tells the platform to let Claude send search queries to a third-party search service provider selected by Anthropic (Google Cloud). So the same groundwork helps you there too, even when the customer never opens claude.ai.

Why two people get different answers

Search behavior depends on the account and the version. In the new Claude experience there is no web search toggle, and Claude searches the web when it helps. On Team or Enterprise plans, an Owner or Primary Owner must first enable web search for the workspace in Organization settings, under Capabilities.

Peekl, one of the top-ranking guides, also notes that Claude tends to be more careful than other assistants, may give fewer names, or may ask where you are before answering (Peekl). And answers shift from run to run. A BrightLocal study of local AI searches, as relayed on X, found only 20 to 33% of results overlapped on a repeated search. That study covered other AI tools, not Claude, but the lesson carries over: one answer is a snapshot, not a ranking.

Step 1: Let the right Anthropic bots in

Short answer: allow Claude-SearchBot and Claude-User. ClaudeBot is a separate decision about training.

Anthropic says it uses different robots to gather public web data for model development, to search the web, and to retrieve content at users' direction, and it separates them so site owners have a choice (Anthropic crawler page).

Bot	What it does	If you block it
ClaudeBot	Collects content that could help train models	Signals your future content should stay out of training
Claude-User	Fetches pages when people ask Claude questions	May reduce visibility in user-directed search
Claude-SearchBot	Indexes content to improve search results	May reduce visibility and accuracy in search results

Those descriptions come straight from Anthropic. Disabling Claude-User stops Anthropic from retrieving your content in response to a user query, which may reduce your visibility for user-directed web search. Disabling Claude-SearchBot stops Anthropic from indexing your content for search, which may reduce your visibility and accuracy in search results.

The point most owners miss: these are separate switches. As Search Engine Journal put it, blocking ClaudeBot stops training data collection but does nothing about Claude-SearchBot or Claude-User (Search Engine Journal).

The robots.txt pattern if you only want out of training

If you want to opt out of training but stay findable, block only ClaudeBot:

User-agent: ClaudeBot
Disallow: /

Anthropic's instructions say this goes in the robots.txt file in your top-level directory, repeated for every subdomain you want to exclude . Anthropic says its bots honor robots.txt directives and support the non-standard Crawl-delay extension. Do not add the same rule for Claude-SearchBot or Claude-User unless you mean to step out of Claude's search answers.

Should you block ClaudeBot?

This part is my opinion, not Anthropic guidance. For most small and mid-size businesses, I'd leave all three open. Training data is part of what Claude "knows" when it does not search. A business that wants to be recommended usually wants to be known. Publishers with paid content may weigh this differently, and that's a fair choice.

Check for accidental blocks today

Open yoursite.com/robots.txt in a browser.
Look for User-agent: * followed by Disallow: /. That blocks every well-behaved bot.
Look for named blocks on Claude-SearchBot, Claude-User or Googlebot.
Ask your web person whether a security plugin, firewall or CDN "block AI bots" setting is turned on. Those can block bots no matter what robots.txt says.

I walk through this in more detail in how to allow AI crawlers in robots.txt.

Step 2: Make sure Brave can reach your pages

Short answer: Anthropic hasn't named its search provider, but the evidence points to Brave, and Brave only crawls what Googlebot can crawl.

What is known about Claude's search provider

Here is the evidence, from strongest to weakest:

Google Cloud's docs: Claude's web search uses a third-party search provider , chosen by Anthropic.
TechCrunch reporting: Anthropic added "Brave Search" to its subprocessor list, programmer Simon Willison reported that at least one search in Claude and Brave returned identical citations, and he found a parameter called "BraveSearchParams" in Claude's web search function (TechCrunch).
A newer wrinkle: one third-party reference site reports that as of September 2, 2026, Anthropic names two subprocessors under Web Search: Brave Search and TurboPuffer (SEO Handbook). I could not confirm that on Anthropic's page, so check it yourself before repeating it.

My read: treat Brave as strongly evidenced, not officially confirmed. Being present in Brave costs you almost nothing and is the best-supported lever available.

Brave follows Googlebot's rules

This is the detail none of the top three guides cover. Brave's crawler does not advertise its own user agent, to avoid being shut out by sites that only allow Google. Brave's help page says: " if a domain or page is not crawlable by Googlebot, then Brave Search's bot will not crawl it either " (Brave).

So a robots.txt rule that blocks Googlebot quietly blocks Brave too. If you've ever restricted Googlebot to save crawl load or hide a staging site, recheck it.

robots.txt, noindex and re-fetching

Tool	What it does in Brave
robots.txt	Controls crawling, not indexing
noindex tag	Removes a page from the index after a re-fetch
Submit URL	Asks Brave to fetch a page again

Brave says robots.txt is not used to keep a page out of its index, a site owner delists a page with a robots noindex directive, and Brave has to re-fetch the page before the change applies. You can request that at Brave's submit URL page.

Two practical checks follow from this. First, make sure no page you care about carries a stray noindex tag. Second, don't combine a robots.txt block with noindex. Google warns that if a page is blocked by robots.txt, the crawler never sees the noindex rule (Google Search Central).

A re-fetch request asks Brave to look again. It does not make Claude cite you.

Quick Brave check

Go to search.brave.com and search your exact business name.
Search your trade plus your city or your product category.
Note whether your site, your listings and your reviews show up.
If your site is missing, fix crawl access first, then submit your key URLs.

What the vendor studies say

These are third-party tests, not Anthropic data. Profound, a vendor, reported an overlap of 86.7% (13 out of 15 results) between Claude's cited results and Brave's top non-sponsored results , as quoted by Keywords Everywhere. That same tracker calls it directional evidence about the integration at that time, and notes Anthropic never confirmed it beyond the subprocessor listing . Fifteen results is a small sample. Other write-ups quote different Profound figures, so read the original before repeating any number.

Step 3: Publish the page that answers the question

Short answer: Claude cites pages. Give it one that answers the exact question your customer asks.

Jamout's guide makes a good point here: write the page that matches the actual sentence a customer would say out loud (Jamout). There's some outside data behind it, all from vendors:

In one B2B client case covering 15,000 answers across five AI engines, 87% of answers that mentioned the client cited the client's own site. That is one client, not a rule.
BrightLocal's study of about 200,000 local AI searches (ChatGPT, AI Mode and AI Overviews, not Claude) found individual business sites got 42% of citations, as relayed on X.
The academic GEO paper (arXiv 2311.09735) found that adding statistics, quotations and cited sources to content improved its visibility in generative engine answers in the authors' tests. It did not test Claude specifically.

What I'd do (my view):

Put one plain sentence near the top of your home page: what you do, who you serve, where. "Commercial roof repair for warehouses in three states" beats "Quality you can trust."
Give each core service its own page. Answer the main question in the first two sentences.
Add specific facts: service area, hours, price ranges if you publish them, licenses, and what you don't do.
Add a short FAQ built from questions customers actually ask you on the phone.
Keep a visible "last updated" date and make it true.

If you serve a specific industry, my guides for accountants and veterinarians show what these pages look like in practice.

Step 4: Get named on the pages Claude reads

Short answer: your own site isn't enough. Claude often cites review sites, directories and comparison pages.

AirOps, a vendor, studied 380,000 AI answers across B2B categories. It reported that ChatGPT, Perplexity and Claude sent 52 to 72% of citations to off-site product and comparison pages and 16 to 21% to review platforms. The same post says Claude rarely cited Reddit or YouTube. That's B2B data, so local results may differ.

In October 2026 spot checks our team ran on "best agency" style prompts, Claude cited directories and listicles such as Clutch, DesignRush and The Manifest. Those were snapshots, and results change between runs.

Get listed on the directories and review sites your industry actually uses.
Ask to be included in honest roundups and "best of" lists in your field.
Keep the same name, phone and description on every profile.
Never buy, fake or script reviews.

If Claude keeps naming a rival, why AI recommends your competitor walks through how to read the cited sources and close the gap.

Step 5: Keep your facts identical everywhere

Short answer: when Claude searches, it reads what the web says about you. Make every source agree.

AI Listings, another top guide, argues that if the web cannot agree on who you are, what you do, or where you operate, Claude tends to hold back (AI Listings). That's their observation, not an Anthropic statement, but it matches common sense. A wrong phone number on three directories confuses any reader, human or machine.

Place	What to check
Your website	Name, phone, address or service area, hours
Google Business Profile	Claimed, correct category and hours
Bing Places	Same facts as Google
Apple Business Connect	Same facts as Google
Industry directories	No old phones, addresses or names
Review sites	Claimed, same description

BrightLocal's local study, as relayed on X, found Yelp was the second-biggest source for local AI answers. It recommended Bing Places and identical name, address and phone details. That study looked at other AI tools, so treat it as a direction, not a Claude rule.

If you sell nationally and have no storefront, the same idea applies to your company name, one-line description, founder name and social profiles. Setup help: Apple Business Connect setup.

Step 6: Add structured data, and keep llms.txt in perspective

Short answer: both are cheap hygiene. Neither is proven to make Claude recommend you.

Structured data is a labeled copy of your business facts in your page code. Google's LocalBusiness documentation is built on schema.org/LocalBusiness. It lists name and address as required. It recommends properties like telephone, url and openingHoursSpecification. Use the most specific business type that fits. National brands without a storefront can use Organization instead.

Now the honest part. Ahrefs, a vendor, reported across its studies that schema had no meaningful impact on ChatGPT citations. I still add it, because it keeps your facts consistent and machine-readable. Just don't expect it to move answers by itself. Details: LocalBusiness schema for AI search.

llms.txt is similar. Ahrefs log data across 137,000 domains showed 97% of valid llms.txt files got zero requests in May. One developer team says its logs show Claude fetching llms.txt, which is one anecdote. Add it if it takes ten minutes. Don't pay anyone to build a strategy around it. More in llms.txt for small business.

Step 7: Earn reviews that describe the real work

Short answer: ask for reviews that name the job, not just the feeling.

This is my view, not a documented rule. "Great service!" tells a reader nothing. "They replaced our 40-gallon water heater the same day" tells a reader, and any search tool, what you do. Jamout suggests asking for a review that mentions what you hired them for and what changed . I agree. Reply to reviews in your own words. Don't script customers, and don't buy reviews.

How to check if Claude recommends you

Short answer: ask Claude the way a customer would, read the cited sources, and log the results monthly.

Open Claude. On older interfaces, make sure web search is on. On Team or Enterprise, confirm an owner has enabled it.
Write five prompts in your customers' words. Example: "Who's a good commercial electrician in Denver?" or "Best bookkeeping service for restaurants."
Include a location when it matters. Claude may ask for one anyway.
Read the cited sources, not only the names. They show which pages Claude used.
Log the date, prompt, whether you were named, who was named, and every cited URL.
Fix what the sources reveal. If a directory shows your old phone number, fix that first.
Repeat with the same prompts next month and compare.

Track each AI tool separately. One study, relayed on X, found only 11% source overlap between ChatGPT and Perplexity. Winning in one tool tells you little about another. My full method is in how to check if AI recommends your business, and the AI visibility audit checklist turns it into a worksheet.

What you control and what you don't

You control	You don't control
Which Anthropic bots you allow	When Claude decides to search
Googlebot and Brave crawl access	What Brave ranks or indexes
Clear, specific pages	Which sources Claude cites
Matching listings and profiles	Run-to-run answer changes
Honest reviews and mentions	Anthropic's training schedule

What nobody can promise

No one can promise Claude will recommend your business. No one can promise when, either. Anthropic decides how Claude searches. The search provider decides what it indexes. Answers change between runs.

What can be promised is the work: open crawler access, pages that answer real questions, matching facts everywhere, and an honest monthly check. If someone guarantees you a Claude mention, ask what they are actually selling. My guide on how to choose an AI search optimization agency lists the questions I'd ask.

Where Get Found with AI fits

I run Get Found with AI, a national service that helps businesses get found, cited and recommended in ChatGPT, Claude, Gemini, Grok, Perplexity and Google's AI Overviews and AI Mode. The steps above are the same ones I work through with clients: crawler access, Brave and Bing presence, answer pages, listings and a logged monthly check. You can see the process on how it works and the scope on services. Costs are covered in what AI search optimization costs.

I can't promise any placement, and I'll say so up front. You can do every step in this guide yourself. If you'd rather have help, start by booking a call.

Working on other tools too? See ChatGPT, Gemini, Perplexity, Grok and Google AI Overviews, or start with what AI search visibility is.

How I put this together

On October 22, 2026, I checked the top three US results for "how to get your business recommended by Claude" and read each one in full:

Jamout: a practical nine-step checklist across ChatGPT, Claude and Google AI. It's strong on one-sentence positioning, answer pages, FAQs and specific reviews. It doesn't explain Claude's web search, Anthropic's three bots or Brave.
AI Listings: the most Claude-specific of the three, and it correctly says to allow Claude-SearchBot and Claude-User. It's written for a UK audience. Some claims, like how Claude evaluates passages, are stated as fact without a primary source. It also skips Brave's Googlebot rule.
Peekl: focused on checking whether Claude names you, with a useful note that Claude gives shorter, more cautious lists. It's light on fixes and crawler detail.

This guide covers every question those three answer. It adds what they leave out: Anthropic's own crawler table and robots.txt pattern, the Google Cloud disclosure about Claude's search provider, Brave's crawler and noindex rules, the 2026 removal of the web search toggle, labeled vendor data, and a repeatable monthly log. Every factual claim links to its source. Where something is my opinion, I say so. The Get Found with AI team reviewed it.

Frequently asked questions

Is there a Claude Business Profile I can claim?

No. Claude has no business listing like Google Business Profile. When it searches the web, it reads public pages and cites them, so your website, listings, directories and reviews are what it can find.

What search engine does Claude use?

Anthropic has not officially named one. Google Cloud's documentation says Claude's web search sends queries to a third-party provider selected by Anthropic. TechCrunch reported in March 2025 that Anthropic added Brave Search to its subprocessor list, and independent tests found matching citations. Treat Brave as strongly evidenced, not confirmed.

Should I block ClaudeBot if I want Claude to recommend my business?

That is a training data choice. Blocking ClaudeBot signals that your future content should stay out of training, and it does not block Claude-SearchBot or Claude-User. Do not block those two if you want to show up when Claude searches. My view is that most small businesses are better off leaving all three open.

Does submitting my site to Brave make Claude cite me?

No. Brave's URL submission asks Brave to fetch a page again. It can help a fixed or new page get picked up, but nothing documented says it makes Claude cite you.

Do schema markup or an llms.txt file help Claude recommend me?

Treat both as hygiene, not levers. Structured data puts your facts in a documented, machine-readable format, but an Ahrefs study found no meaningful citation lift from schema. Ahrefs log data also showed most llms.txt files got zero requests. Both are cheap, so do them, but spend your real effort elsewhere.

How often should I check whether Claude names my business?

Once a month is enough for most businesses. Use the same prompts each time, read the cited sources, and log the results. Answers vary from run to run, so look for patterns across several months rather than reacting to one answer.

Can anyone guarantee Claude will recommend my business?

No. Anthropic controls how Claude searches and the search provider controls its index. What someone can promise is the work: open crawler access, clear pages, matching listings and an honest monthly check. If someone guarantees a Claude mention, ask what they are actually selling.