One cold question about your category, put to Claude, GPT and Gemini — no hints, no site context. You see each engine's answer excerpt, and the likely reason when you're absent.
No account. One bounded check, never the full crawl or a site-wide score.
Named in 1 of 3 answers, in this sample.
“What are the best options for freight tracking for small importers? Name specific products, companies or sites.”
Derived from your homepage’s own words; no engine was told about Northwind.
For small importers, Harbourline and Tidewell come up most. Harbourline covers ocean and air freight in one view.
Not namedTidewell is a common pick for small import teams, with Portcall for teams that ship mostly by sea.
Not namedNorthwind sends customs alerts and port arrival times for small importers.
NamedAsking Claude, GPT and Gemini…
One bounded check. It returns or it fails, and this page moves to the result when it returns.
Three engines answer one question. Who they name can change by run.
When someone asks an AI assistant "what are the best options for X" and your category is X, either you're in the answer or you aren't — and no dashboard you own will tell you. This check derives one category question from your homepage's own words, asks Claude, GPT and Gemini the same question cold — no site context, no hints — and detects your name in each answer deterministically. We never ask a model whether it mentioned you; models answer that question agreeably.
Unprompted recall is the strictest test of AI visibility: retrieval can be won page by page, but being named cold means the engines' models already associate your brand with your category. It's the AI-era analogue of branded search — and the question this whole category of tooling ultimately sells on.
The product's Recall stage tracks a curated set of category prompts, samples each engine repeatedly, and reports a rate with a Wilson confidence interval — withheld entirely when the sample is too noisy to mean anything. One question, once, is the free taste of that measurement.
One category question is derived from your homepage's own words — deterministically, with no model formulating it — and then put to Claude, GPT and Gemini cold: no hints, no site context, no retrieval. You see each engine's answer excerpt and whether your name appeared in it.
By matching the text ourselves, deterministically. We never ask a model whether it mentioned you — the question is leading, the answer is unreliable, and it doubles the cost of the run for a worse result.
It is reported as unmeasured, never as a miss. A provider outage is not evidence that AI does not know you, and collapsing the two would be the exact error this tool exists to avoid.
No, and the page should not be read that way. It is a single cold sample from three engines — an indication, not a rate. Measuring recall properly means many prompts, repeated samples, and a confidence interval printed beside the number, which is what the product's Recall stage does.
The result gives the likely reason where there is one — most often that the engine's crawler is blocked from your site, which the tool checks. Absence has other causes too: a young site, a crowded category, or simply nothing citable about you in the model's training data. The tool reports the reason it can evidence, not the one that sounds best.
Unblocking removes a known obstacle. It does not create a citation, and nothing here promises one. Being reachable is the first of five separate questions, not a shortcut past the other four.
Each one is its own free tool: one URL, one check, no account.
| Kind | Tool | Question | Run it |
|---|---|---|---|
| Reach and be understood | |||
| AI Crawler Access Checkdeterministic · one fetch | Which AI crawlers can reach my site? | Check crawler access | |
| llms.txt Creatordeterministic · bounded fetch | What should my llms.txt say? | Create my llms.txt | |
| Sitemap llms.txtdeterministic · sitemap locs | Can my sitemap become an llms.txt? | Create llms.txt from sitemap | |
| AI robots.txt Policydeterministic · compose | Which AI crawlers should my robots.txt allow? | Compose my robots.txt | |
| How AI Sees Your Sitemodel-assisted · one read | What does my homepage tell AI about me? | See the AI read | |
| llms.txt Graderdeterministic · bounded fetch | Is my llms.txt a briefing or a stub? | Grade my llms.txt | |
| Build trust | |||
| E-E-A-T Quick Checkmodel-assisted · one read | Does my site show visible trust signals? | Check trust signals | |
| Kind | Tool | Question | Run it |
|---|---|---|---|
| Publish correctly | |||
| Schema Checkdeterministic · one fetch | Is my structured data valid? | Check my schema | |
| FAQ JSON-LDdeterministic · one fetch | Can my FAQ become FAQPage markup? | Build JSON-LD | |
| SERP Previewdeterministic · one fetch | What will my listing look like in search and social? | Preview my listing | |
| Canonical Inspectordeterministic · bounded fetch | Do my redirects agree with my canonical tag? | Inspect my canonical | |
| Page Speeddeterministic · PageSpeed | How fast is my page for real mobile visitors? | Check page speed | |
| Link Graphdeterministic · bounded crawl | Which pages are orphaned or buried deep? | Map my links | |
| Create and convert | |||
| sameAs Packdeterministic · homepage + about | Can I mark up the profiles my page already links to? | Build sameAs JSON-LD | |
| HTML to Markdowndeterministic · one fetch | How does my page read to an LLM? | Convert this page | |
The Free Site Audit puts top issues, ranked keywords, closest competitors, authority and AI readiness in one dated report. No account.