Claude SEO: how to assess whether your business is being discovered

A buyer uses Claude to research a problem you solve. Will they encounter your business? If they do, will the information be accurate and useful?

Those are better starting questions than “how do we rank number one in Claude?” A conversational answer is not a fixed search-results page, and a single response is not a dependable visibility report.

Claude SEO should begin with a clear assessment of access, relevant content and observed answers. This guide provides a practical audit method based on Anthropic’s documentation and our approach to search strategy. It does not claim that Re-Imagine has run a controlled Claude ranking experiment.

Understand what Claude search is doing

Anthropic describes web search as a way for Claude to retrieve current information and provide source citations. Availability and behaviour can depend on the experience and settings used. Read Anthropic’s web search guidance.

For a visibility check, record whether search occurred and which sources were shown. Do not combine observations from different modes or conversations without noting the difference.

Also distinguish a mention from a recommendation. Your company might be named in a list, compared with another provider or cited as a source for an explanation. Those are different forms of visibility.

Separate training controls from search access

Anthropic documents three different agents: ClaudeBot relates to potential training data, Claude-SearchBot supports search, and Claude-User supports user-directed access. Their purposes differ. Check Anthropic’s crawler documentation.

A technical review should consider the access your business intends to allow, together with robots.txt and hosting protections. It should not casually change training preferences to chase visibility.

Being accessible does not guarantee inclusion. Treat access as one diagnostic check, alongside whether the page contains relevant information and credible evidence.

Build a question set around the work you want to win

Start with customer conversations and search research. Choose questions that reflect an actual decision: requirements, suitability, trade-offs, implementation or supplier selection.

For an illustrative event venue, relevant questions could concern capacity, accessibility, accommodation and transport. The useful questions depend on the events the venue can host and the organisers it wants to reach.

Keep branded questions separate. A business appearing when its exact name is supplied does not establish that an unfamiliar customer can discover it.

Our SEO, AEO and GEO comparison explains how this work fits alongside existing search activity.

Record enough context to interpret the result

For every check, save the exact prompt, date, visible mode or search conditions, answer, sources and any relevant conversation history. Record whether your business appears, what is said about it and whether the cited URLs support those claims.

Repeat the checks under comparable conditions. Keep new prompt variations separate from the original monitoring set, so changes in the test do not masquerade as changes in your visibility.

If Claude does not cite a page, do not assume it has rejected that page for a specific reason. The absence is an observation, not a diagnosis. The next step is investigation.

Inspect what your website actually tells a buyer

Review priority service pages against the questions in your set. Could a new customer determine who the service suits, what is included, what implementation requires and which claims are supported?

Check for specific, verified facts. A page saying “flexible packages for every event” is less useful to an organiser than an accurate explanation of room configurations, capacity and booking requirements.

Where information is missing, work with the person who knows the answer. AI can help structure the material, but it should not fill factual gaps with plausible guesses. Our content marketing service connects that expert input with useful page content.

An illustrative audit decision

Suppose a business supports a specialist integration, but its website only says it “works with your existing systems”. A relevant buyer cannot verify the integration from that statement.

The first action is to confirm the capability and publish a useful explanation: supported versions, limitations, setup requirements and who to contact. Then monitor the same question set again.

This is a sensible content improvement. It is not proof that a particular phrase will make Claude recommend the business. Publishing should be justified by the customer need even before a monitoring result changes.

Choose measures that will guide the next decision

Track observed appearances and citations within the defined set, together with factual accuracy. Where available, review identifiable referrals and useful actions on the website. Do not translate an appearance percentage into a claim about all Claude users.

Keep platform research separate from evidence drawn from other assistants. For example, the Microsoft Copilot results in our Serviceware case study are not evidence of Claude performance.

Our measurement guide explains how to report these boundaries clearly.

Make your next step specific

A practical Claude visibility plan should leave you with a defined question set, a record of observed answers, verified technical findings and prioritised content improvements. It should explain what remains uncertain.

Our AI SEO service helps turn that investigation into a wider search plan. Tell us which customer questions matter to your business.

About the author

Molly Sishton, Founder of Re-Imagine Marketing. Molly combines marketing strategy, SEO and content expertise with hands-on experience of developing AI technology. Re-Imagine has worked with AI since 2021 and developed Sighted, its own enrichment and visibility technology. Meet the Re-Imagine team.

Recognition across our founding team and ventures: Connie Sishton featured among Wales Business Insider’s 25 Rising Stars of Welsh Business in 2025. Molly and Connie’s venture SO-ME won Best Female-Led Startup at Pitch Black Malta. Sighted was selected for Tech Nation’s Breakout 50.