SaaSClerk · free tool · no signup, no card
AEO Scanner
When someone asks an AI assistant for a tool like yours, the answer comes from text the assistant could read. If the crawler is blocked, or the page never says what it is in a form a model can quote, the answer is built from your competitors instead. This scan reports what is actually readable on your site today.
It checks seven machine-readable things and asks one language model one question. Every line in the report is tied to a response we received while you waited — no industry averages, no scores invented from a database.
What each check measures, and what it does not
- Reachable — The home page answers at all. A scan of a dead site measures nothing else.
- AI crawlers — Eight named crawler user agents checked against robots.txt: is each one blocked site-wide?
- llms.txt — The emerging plain-text map for language models, served at /llms.txt.
- Sitemap — Declared in robots.txt, or served at /sitemap.xml.
- Structured data — JSON-LD blocks on the home page and the @type names they declare.
- Title and description — Length, in the range a search or answer surface can display without truncation.
- One h1 — The page states its own subject exactly once.
- One live model — Asked, verbatim: “Do you know <brand>?” The answer is printed unedited.
It does not query ChatGPT, Perplexity or Google AI Overviews directly: there is no vendor API behind this page, and an unmeasured engine is reported as not measured rather than replaced with a guess. It also does not measure rankings in classic search — different surface, different data source.
Why robots.txt is the first check
A single wildcard Disallow: / added to quiet one crawler takes every crawler with it, named ones included. That is the most common self-inflicted wound we see, and it is invisible on the rendered site: the page looks fine in a browser and is empty to a model. The scan names each blocked crawler instead of returning a single pass or fail.
FAQ
Does the scan store what I type?
The domain and the pass count are logged as an event without your email. If you leave an email for the monitoring plan, the domain, score and failed checks are stored so a person can reply to you.
What is the $149 plan, exactly?
Weekly re-runs of the same checks, a diff of what changed per check, and three prioritised page fixes each Monday. It is a reservation at this stage: the form collects the email, and the checkout link follows. No card is charged on this page.
Why only one model?
Because we can measure one honestly. Every additional engine has to be measured with its own key and its own error surface; bolting on numbers we cannot re-derive would turn a small true report into a big untrue one.
Next step
The scan tells you what a model can read. The other half is knowing what you actually pay for and who holds it — the unused seat calculator sizes that, and a free audit builds the inventory. Plans and limits are on the pricing page.
Built by SaaSClerk. Related: read-only subscription scanning and the Shadow IT risk calculator.