AI visibility
How to appear in Claude
Appearing in Claude means it mentions or recommends your brand in its answers. Claude answers from training and, when web search is on, from live pages it retrieves and cites.
How Claude picks and cites its sources
When web search is on and a question needs current information, Claude searches the web and cites the pages it used with links. Otherwise it answers from training alone, so some conversations never touch your site, however good it is.
Anthropic documents three separate crawlers: ClaudeBot collects content that may be used for training, Claude-User fetches a page when a person asks Claude to, and Claude-SearchBot indexes content to improve the quality of search results. Each follows its own robots.txt rule, so blocking one does not block the others.
- Training knowledge: brands described consistently across reputable sources can be recalled without searching.
- Search, when on: relevant and current pages are retrieved and cited inline.
- Verifiability: precise, well-sourced claims are easier to cite than vague marketing copy.
Action plan: get Claude to cite you
Pin down your facts
Write one canonical description of what you do, your product names, category and location, and use it verbatim across your site and profiles.
Publish precise, sourced pages
For each buying question, state what the product does and doesn't do, with figures, dates and named sources rather than superlatives.
Set robots.txt bot by bot
Allow Claude-SearchBot and Claude-User if you want to be retrievable. ClaudeBot is a separate decision about training.
Earn independent references
Get described by industry publications, comparison pages and documentation your category relies on. Independent text is what models tend to recall.
Keep pages machine-readable
Server-rendered HTML, clean headings and Organization structured data. An llms.txt is cheap to add but carries no promise.
Test with search on and off
Ask the same questions both ways to separate what Claude recalls from what it retrieves, then work on whichever is weaker.
What SeoWave measures for Claude
- Mention rate: how often Claude names your brand across your buying questions, run with web search on.
- Cited sources: the domains it links, grouped as your site, competitors, community and review sites, and editorial media.
- Competitors named instead of you, and the tone of each mention.
SeoWave asks each buying question through the engine's API with web search on, so results approximate what people see in the app rather than replicate it. Read trends over weeks, not single answers.
Common mistakes
- Treating the three Claude bots as one. Blocking ClaudeBot affects training only; blocking Claude-SearchBot or Claude-User affects retrieval.
- Assuming Claude always searches. When it answers from memory, only long-run reputation moves the result.
- Copy built on superlatives like "the best platform" with nothing to check. Vague claims are hard to cite.
- Letting facts differ between your site, profiles and press, so it isn't clear which is right.
See if Claude recommends your brand
Run a free scan to see how AI engines answer questions about your brand, and who they name instead.
Scan my site freeFrequently asked questions
Does Claude browse the web?
It can. When web search is on and the question needs current information, Claude searches and cites live pages; otherwise it answers from training.
How do I let Claude access my site?
Manage its three bots in robots.txt: Claude-SearchBot and Claude-User for retrieval, ClaudeBot for training. Keep pages fast and server-rendered, and structure content clearly.
How is this different from optimizing for ChatGPT?
The fundamentals overlap: clarity, corroboration and crawlability. The engines differ in what they retrieve and remember, so measure each one separately.
Does llms.txt make Claude cite me?
Don't count on it. It is a proposed convention, not a guarantee, and Google has said it isn't needed. Treat it as a tidy index of your best pages, not a lever; content and third-party evidence do the work.