Claude, Anthropic's AI assistant, is used by millions of people for research, writing and buying decisions, and it is especially popular with professionals and developers. When someone asks Claude to recommend a project management tool, compare agencies or explain a product category, the brands it names get a quiet but valuable endorsement. The brands it leaves out never enter the conversation.
Getting mentioned by Claude is not about a secret trick. Claude draws on two sources of knowledge: what it learned during training and what it finds through live web search. Each has its own crawlers, its own timeline and its own levers. Most "AI visibility" advice blurs the two, which leads to wasted effort.
This guide explains how Claude finds information, which Anthropic crawlers you need to allow, why Brave Search matters, and a practical step-by-step plan to improve how often, and how accurately, Claude mentions your brand.
Key Takeaways
- Claude answers from its training data and, when web search is used, from live search results with citations.
- Anthropic runs three separate bots: ClaudeBot (training), Claude-SearchBot (search indexing) and Claude-User (fetching pages for a user's question). Blocking the wrong one can remove you from Claude's answers.
- Anthropic's subprocessor list names Brave Search as a web search provider, so being well indexed in Brave is worth checking.
- Consistent, well-sourced mentions of your brand across trusted third-party sites influence both training data and search results.
- Test priority prompts in Claude with web search on and off, and track accuracy as well as mentions.
How Claude finds information about brands
1. Training data
Like other large language models, Claude is trained on a large body of text, including publicly available web content. ClaudeBot is the crawler Anthropic uses to collect web data for model development. What Claude "knows" without searching reflects that training snapshot, so it can be months out of date and is shaped by how widely and consistently your brand was described on the web.
2. Live web search
Anthropic added web search to Claude in March 2025, with direct citations to its sources, and has since expanded it to more users and to its developer platform. When search is used, Claude runs one or more queries, reads results and writes an answer with links. Anthropic's API announcement describes the citation behaviour plainly.
“Every web-sourced response includes citations to source materials, enabling users to verify information directly.”
— Anthropic, Introducing web search on the Anthropic API
Which index sits behind those searches? In March 2025 TechCrunch reported that Anthropic had added Brave Search to its subprocessor list. Later tracking of Anthropic's Trust Center found Brave still listed under web search in mid-2026, alongside a second provider, turbopuffer, added in May 2026 whose exact role is not public. The practical takeaway: you cannot control Claude's search backend, but you can make sure your key pages are discoverable in independent indexes such as Brave, not just Google.
Anthropic's crawlers and what blocking each one does
Anthropic documents three user agents. They are frequently confused, and blocking them all by default is one of the most common reasons a brand disappears from Claude's answers.
| User agent | Purpose (per Anthropic) | If you block it | Our usual recommendation |
|---|---|---|---|
| ClaudeBot | Collects web content that may help improve Anthropic's generative AI models | Future pages are excluded from model training data | A business choice: allow if you want Claude's built-in knowledge to include you |
| Claude-SearchBot | Navigates the web to improve search result quality for users | Your content is not indexed for Claude's search, which may reduce visibility | Allow if you want to appear in Claude's search answers |
| Claude-User | Fetches pages when a Claude user asks a question | Claude cannot retrieve your pages in response to user queries | Allow for public marketing pages |
“Disabling Claude-SearchBot on your site prevents our system from indexing your content for search optimization…”
— Claude Help Center, Does Anthropic crawl data from the web, and how can site owners block the crawler?
Check your CDN and firewall as well as robots.txt. Some security settings block AI user agents at the network level without any robots.txt rule. Our guide to AI crawlers covers how to set rules for every major AI bot.
Step-by-step: how to get your brand mentioned by Claude
- Benchmark first. Write 15–30 prompts your buyers would ask, such as "best [category] for [audience]", "[your brand] vs [competitor]" and "is [your brand] good for [use case]". Run them in Claude with web search on and off, and record mentions, position, accuracy and cited URLs.
- Open the right doors. Allow Claude-SearchBot and Claude-User in robots.txt and your firewall. Decide deliberately about ClaudeBot.
- Check independent indexes. Search your brand and key pages in Brave Search and Bing. If important pages are missing, fix crawlability, internal links and sitemaps.
- Make your own pages quotable. State what you do, who it is for, where you operate and what makes you different in plain sentences near the top of key pages. Avoid burying facts in images or scripts.
- Publish comparison and use-case content. Claude often answers "which is best for" questions. Honest comparison pages and use-case guides give it material to cite.
- Strengthen your entity. Use Organization structured data, consistent names and descriptions, and profiles on trusted directories. See our guides to entity SEO and Wikidata and Wikipedia for brands.
- Earn third-party mentions. Independent reviews, industry roundups, podcasts and press coverage feed both training data and search results. Our digital PR guide explains how.
- Correct errors at the source. If Claude repeats something wrong, find where it comes from and fix it there. Follow our process to fix AI brand misinformation.
- Re-test monthly. Search-based answers change quickly; training-based answers change only with new model releases.
Claude vs other assistants: what differs?
The fundamentals overlap with ChatGPT search optimisation and Google's AI features, but the details differ.
| Factor | Claude | ChatGPT | Google AI Mode |
|---|---|---|---|
| Search crawler | Claude-SearchBot | OAI-SearchBot | Googlebot |
| Training crawler | ClaudeBot | GPTBot | Controlled by the Google-Extended token |
| Known search provider | Brave Search listed as a subprocessor | OpenAI's own search systems | Google's index |
| Citations | Inline citations to web sources | Sources panel and inline links | Supporting links in the answer |
| Reporting for site owners | Referrals from claude.ai in analytics | Referrals from chatgpt.com | Included in Search Console Performance report |
For a wider overview, see our guide to AI search engines. If you sell to businesses, our article on B2B AI search covers buyer research patterns in more depth.
What kind of content does Claude tend to cite?
Anthropic does not publish ranking factors for Claude's search. In our testing and in Semrush's September 2026 hands-on review, which found Claude gave accurate, focused recommendations but leaned on a repetitive set of sources, a few patterns stand out:
- pages that answer the exact question in a self-contained paragraph;
- pages with clear dates, specifications and named sources;
- independent comparison and review content rather than only vendor pages;
- official documentation for factual questions about a product.
None of this is unique to Claude, which is good news: one well-executed content and PR programme serves every assistant.
Common mistakes
- Blocking all Anthropic bots together. Blocking ClaudeBot to opt out of training is reasonable; blocking Claude-SearchBot and Claude-User as well removes you from live answers.
- Testing only with search off. Training-based answers lag. Test both modes.
- Assuming Google rankings carry over. Claude's search does not use Google's index, so check independent search engines too.
- Prompt-stuffing your pages. Hidden instructions aimed at AI models are manipulative, can be treated as spam and erode trust.
- Ignoring accuracy. A mention with an outdated price or wrong location can cost you more than no mention.
Related Guides
- AI Crawlers: Should You Block or Allow GPTBot and Others?
- Wikidata and Wikipedia for Brands: Feeding the AI Knowledge Graph
- Content Chunking: Write Passages AI Can Quote
Frequently Asked Questions
Can I pay to be recommended by Claude?
No. Anthropic does not sell placement in Claude's answers. Visibility comes from being crawlable, clearly described on your own site and mentioned by trusted third parties.
Which search engine does Claude use?
Anthropic's subprocessor list names Brave Search under web search, as first reported by TechCrunch in March 2025. A second provider was added in 2026. Anthropic does not publish full details of how results are chosen.
Does blocking ClaudeBot remove me from Claude's answers?
Blocking ClaudeBot opts future pages out of training. To stay visible in Claude's live search answers, keep Claude-SearchBot and Claude-User allowed.
How do I see traffic from Claude?
Look for referrals from claude.ai in GA4. Create a custom channel group for AI assistants so Claude traffic is not lost inside general referrals.
How long until changes show up in Claude?
Search-based answers can change as soon as your pages are re-crawled. Training-based knowledge only updates when Anthropic releases a new model trained on newer data.
Conclusion
Getting your brand mentioned by Claude comes down to two jobs: make sure Anthropic's search agents can reach and understand your pages, and make sure the wider web describes you clearly and consistently. Start with a prompt benchmark and a crawler check, then build entity signals and third-party mentions over time.
Want a Claude and ChatGPT visibility audit? Our AI brand visibility service tracks how assistants describe you and fixes the gaps, as part of our wider generative engine optimisation work. Get a quote to begin.
References
- Anthropic Help Center: Does Anthropic crawl data from the web?
- Anthropic: Introducing web search on the Anthropic API
- SD Times: Anthropic's Claude can now search the web
- TechCrunch: Anthropic appears to be using Brave to power web search for Claude
- Xponent21: Anthropic lists two web-search subprocessors behind Claude
- Semrush: 9 best AI search engines in 2026
