# How to get cited by Claude

> Claude answers some questions from memory and looks others up live. Here is what Anthropic says about when it searches, which of its crawlers to let in, and how to make your pages easy to quote.

Updated September 28, 2026 · AI Syndicate team · https://www.aisyndicate.com/blog/how-to-get-cited-by-claude/

**Quick answer:** **Claude** searches the web when a question depends on current facts, including facts about specific businesses and products, and it cites every web source it uses. To be one of those sources, allow **Claude-SearchBot** and **Claude-User** in robots.txt, make sure normal search crawlers can reach your pages, and write short, factual sentences Claude can quote.

## When does Claude search the web instead of answering from memory?

Claude has two ways to answer. It can answer from what it learned in training, or it can run a web search and cite what it finds. Anthropic's documentation says which one happens. [Claude Platform Docs](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool)

| Claude usually searches for | Claude usually answers from memory for |
| --- | --- |
| Recent events, news or announcements | Established facts, math and science basics |
| Current prices, rates, scores or statistics | Creative writing and brainstorming |
| Specific organizations, people or products that might have changed | Content already pasted into the conversation |
| Anything the user explicitly asks it to look up | Greetings and ordinary conversation |

The row that matters for a business is the third one. A question like *"who's the best roofer near me"* or *"is this software worth it"* is about specific organizations and products, so it is the kind of question Claude looks up. Anthropic also says that when Claude uses the web, it "provides direct citations so you can easily fact check sources." [Claude blog](https://claude.com/blog/web-search) Web search has been on every Claude plan, free included, worldwide since May 27, 2025.

So there are two things to work on: being findable in the live search, which this guide covers, and being known to the model before anyone searches. Training data is slower to change and harder to steer.

## Where do Claude's search results come from?

Anthropic's documentation doesn't name a search provider. But on March 19, 2025, the day before web search launched, **Brave Search** was added to Anthropic's public list of subprocessors, and the web search tool included a setting called `BraveSearchParams`. [Simon Willison](https://simonwillison.net/2025/Mar/21/anthropic-use-brave/) Anthropic also runs its own search crawler, **Claude-SearchBot**, which it says "navigates the web to improve search result quality for users." [Claude Help Center](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler)

Our reading: Claude's live answers draw on an index that Brave and Anthropic's own crawler help build. Anthropic hasn't said that in so many words, so treat it as a strong clue, not a rule. Three things Brave does say about its crawler are worth knowing, because they are easy to get wrong: [Brave Search](https://search.brave.com/help/brave-search-crawler)

- **It follows your Googlebot rules.** Brave doesn't use a separate user agent. If a page is "not crawlable by Googlebot," Brave won't crawl it either, so a robots.txt rule meant for Google also shuts out Brave.
- **robots.txt doesn't remove a page from Brave.** To get a page out of Brave's index, use a `noindex` tag. Brave has to fetch the page again before the change applies.
- **You can ask for a re-fetch.** Brave's [Submit URL](https://search.brave.com/submit-url) page lets site owners ask it to fetch a page again, which helps after an update.

## Which Anthropic crawlers should I allow?

Anthropic runs three bots. Each one has its own robots.txt name, and each does a different job. [Claude Help Center](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler)

| Bot | What Anthropic says it does | What blocking it costs you |
| --- | --- | --- |
| `Claude-SearchBot` | Crawls pages to improve search results | Anthropic says it may reduce your "visibility and accuracy" in search results |
| `Claude-User` | Visits a page when a Claude user's question needs it | Anthropic says it may reduce your visibility in user-directed web search |
| `ClaudeBot` | Collects content that may be used to train models | Your future pages are left out of training data. This is a business choice, not a mistake |

If you want Claude to cite you, **allow `Claude-SearchBot` and `Claude-User`**. Whether to allow `ClaudeBot` is up to you. Allowing it helps Claude know your business without searching, and blocking it keeps your content out of training. Anthropic says all its bots follow robots.txt, including the non-standard `Crawl-delay` setting. It also warns that blocking IP addresses "may not work correctly or persistently guarantee an opt-out," so robots.txt is the switch to use. We sort every AI crawler by what it does in [Which AI crawlers to allow](/blog/ai-crawlers-robots-txt/).

> **robots.txt is only the first gate.** A firewall rule, a CDN bot setting or a login wall can block Claude just as effectively, and nothing warns you when it happens. Check what a crawler actually receives, not only what your robots.txt says.

## What makes a page easy for Claude to cite?

Anthropic's documentation shows what a Claude citation contains. Each one carries the page's **URL**, its **title**, and up to **150 characters** of the quoted text. Each search result also carries a `page_age`, meaning when the page was last updated, and search results can be localized to the user's city, region or country. [Claude Platform Docs](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool) That suggests four ways to write pages:

1. **Put the key fact in one short sentence.** A quote can be at most 150 characters, so write sentences like *"We install metal roofs in Tampa, Clearwater and St. Petersburg"* that make sense on their own. A key fact spread across a whole paragraph is hard to quote.
2. **Make the title say what the page is.** The title appears with the citation. "Metal Roof Installation in Tampa | Smith Roofing" tells a reader why the link is there. "Home" doesn't.
3. **Keep dates honest.** Claude gets an update date for each result. Update pages when prices, hours or services change, and don't change the date when nothing else did.
4. **Say where you work, in plain text.** Results can be tailored to where the user is. Name the cities you serve in the page text, not only on a map or in an image.

These rules come from the shape of a citation, not from a ranking formula. Anthropic doesn't publish how Claude chooses between results, and neither do we. Pages that follow them are simply easier to quote accurately.

## What should be on my Claude checklist?

| Check | Done when | Source |
| --- | --- | --- |
| `Claude-SearchBot` allowed | robots.txt doesn't block it | Claude Help Center |
| `Claude-User` allowed | robots.txt doesn't block it | Claude Help Center |
| Decided on `ClaudeBot` | You chose to allow or block it on purpose | Claude Help Center |
| Googlebot can reach key pages | No robots.txt rule blocks Google, so Brave isn't blocked either | Brave Search |
| No accidental `noindex` | Pages you want cited don't carry the tag | Brave Search |
| Firewall and CDN don't block AI bots | A test fetch returns the real page, not a challenge or 403 | Claude Help Center |
| Key facts fit in one sentence | Each fact is 150 characters or less | Claude Platform Docs |
| Page titles say what the page is | No "Home" or "Untitled" titles | Claude Platform Docs |
| Update dates are real | They change only when the content changes | Claude Platform Docs |
| Service area written in text | City names appear in the page text | Claude Platform Docs |

## How do I see whether Claude cites my site?

We don't know of a Claude report for site owners like Bing's [AI Performance report for Copilot](/blog/microsoft-copilot-visibility/), and Anthropic's crawler help article doesn't mention one. That leaves two practical signals:

- **Referral visits.** When someone clicks a citation in Claude, the visit can show up in analytics with claude.ai as the referrer. We explain the setup, and its gaps, in [How to track AI search traffic](/blog/track-ai-search-traffic/).
- **Asking the questions yourself.** Analytics only shows clicks. It can't tell you when Claude cited a competitor instead of you. Our [Prompt Simulator](/prompt-simulator/) runs one question across 12 engines, Claude included, and highlights your citation and your position in each answer.

One warning: access breaks without any alert. A new CDN rule, a plugin update or a changed robots.txt can quietly shut Claude out. A one-time check won't catch that, so check on a schedule.

## FAQ

### Does Claude use Google search?

Anthropic doesn't say which search engine Claude uses. Brave Search was added to Anthropic's subprocessor list on March 19, 2025, and the web search tool included a setting called BraveSearchParams. [Simon Willison](https://simonwillison.net/2025/Mar/21/anthropic-use-brave/) Anthropic also runs its own search crawler, Claude-SearchBot.

### If I block ClaudeBot, will Claude stop citing me?

Not by itself. Anthropic says blocking ClaudeBot keeps your future content out of training data. Search and live page visits are handled by different bots, Claude-SearchBot and Claude-User. Blocking those is what Anthropic says can reduce your visibility in search. [Claude Help Center](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler)

### Can free Claude users search the web?

Yes. Anthropic launched web search in March 2025 for paid US users, then made it available globally on all Claude plans on May 27, 2025. [Claude blog](https://claude.com/blog/web-search)

### Can I submit my site to Claude?

Anthropic's crawler help article doesn't mention a way to submit URLs. Brave Search, which appears on Anthropic's subprocessor list, has a Submit URL page where site owners can ask it to fetch a page again. [Brave Search](https://search.brave.com/help/brave-search-crawler)

## Sources

1. [Web search tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool) — Claude Platform Docs. Read Sep 28, 2026.
2. [Does Anthropic crawl data from the web, and how can site owners block the crawler?](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler) — Claude Help Center. Read Sep 28, 2026.
3. [Claude can now search the web](https://claude.com/blog/web-search) — Claude blog. Read Sep 28, 2026.
4. [Anthropic Trust Center: Brave Search added as a subprocessor](https://simonwillison.net/2025/Mar/21/anthropic-use-brave/) — Simon Willison. Read Sep 28, 2026.
5. [Brave Search Crawler](https://search.brave.com/help/brave-search-crawler) — Brave Search. Read Sep 28, 2026.