Our commercial real estate software gets mentioned in Claude answers about lease negotiation, but Claude never links to our docs—it just cites us by name. How do we get Claude to attribute answers to our actual content?

Claude pulls from its training data cutoff, not live crawls, so it may know your brand exists without having indexed your current documentation. The gap between brand mention and content attribution usually means your docs either weren't in Claude's training corpus or weren't marked as authoritative through structured data signals that influenced earlier training.

Start by ensuring your knowledge base pages use schema.org/Article with author, datePublished, and isBasedOn markup that clearly identifies your company as the source. Add a robots.txt rule that allows AI crawlers (check Anthropic's crawler IP ranges if they publish them). More immediately, create a public llms.txt file at yourdomain.com/llms.txt that lists your primary knowledge base URLs and a brief description of what your company does—this signals to newer AI systems that these are your canonical resources.

Test by asking Claude the same lease negotiation question in a new conversation and note whether it cites your domain. If it still doesn't, reach out to Anthropic's support channel for AI visibility; some vendors are beginning to accept direct URL submissions for citation preference. Track mentions weekly using your existing brand monitoring tool and note the date each mention starts including links—that lag tells you when the new signals started flowing through Claude's training or retrieval pipeline.