Does Claude cite you?
Not a score someone modeled: we ask Claude real questions, store the answers, and count who it cited. If the answer is no, the audit says no, with the sample behind it.
Free, no sign-up. One crawl of your public pages by our self-identifying bot. Nothing is written to your site.
We keep the anonymous result to build an industry benchmark. No personal data is collected, and your individual result is never published.
Nothing here chases you. No follow-up sequence, no “strategy session” booked on your behalf, and no email field on this form, so there is nothing we could send you even if we wanted to. You get the whole result, including every failure we find. If you want a human to read it with you, you have to come and ask for one, and that offer sells nothing.
How Claude decides who gets cited
- Claude answers with web search on and cites the pages behind its claims. Business buyers increasingly ask it for vendor shortlists, and those shortlists carry links.
- Anthropic's training crawler is ClaudeBot; live search fetching is separate. As with the other engines, the audit reports your robots.txt treatment of each token rather than assuming.
- Claude sampling costs roughly 8x what the other engines cost us per prompt, which is why its full-depth sampling lives on the Agency tier: a real cost constraint, stated instead of hidden.
What we measure on Claude
We sample Claude with search enabled on buyer-shaped prompts, store every answer, and count cited domains. Pro measures Claude at reduced sampling depth to control its cost; Agency samples it at full depth, so its confidence interval is the narrower one.
Claude citations are measured on Pro (shallow sampling) and at full depth on Agency; the depth difference and why are stated on the pricing page. See plans
Measurement IntegrityVerified⌄
We measure your visibility. We never manufacture it. Each guarantee below is enforced in code, not just promised:
- Self-identifying crawler. Our crawler names itself (SEOAIO, seoaio.ai/bot) so you can exclude it from your analytics.
- No manufactured citations. A citation is counted only when the AI engine returns it. We never inject your URL to inflate a score.
- Elicited-probe method, disclosed. Answer Share is measured by asking engines a fixed, disclosed prompt set — never dressed up as organic traffic.
- Measured vs Modeled, always labeled. Every number is stamped measured, modeled, or mixed. We never show a single Measured number over a modeled part.
- Honest confidence bands. Answer Share ships with a 95% confidence interval (Wilson) and its sample size. A small sample shows a wide band, not false precision.
- No engagement manufacturing. We measure your visibility. We never pay to seed prompts or manufacture the metric we report.
Claude visibility FAQ
Why is Claude treated differently from the other engines?
Cost, and we say so. Claude sampling costs us roughly 8x per prompt, so Pro samples it shallowly and Agency at full depth. The interval that difference produces is printed on the numbers rather than smoothed over.
Does Claude actually cite websites?
With search enabled, yes: answers carry source links, and we store the answers so you can read the exact sentence that cited you or a competitor.
Can I see the answers behind my Claude numbers?
Yes. Every sampled answer is stored verbatim with its date and prompt. A number you cannot open is an estimate; ours open.
Also measured: Perplexity · Google AI Overviews · ChatGPT · Gemini (not yet)
One audit covers every engine we measure.
Free, no sign-up, and honest about what it could not measure. Run it and read the exact answers behind your numbers.