Enter any public page URL for a 30-second check of answer-first structure, FAQ, structured data, AI crawler access and more. No login.
Honesty note: this is a static heuristic check for optimisation reference only โ it does not represent any AI engine's official standard.
Privacy: we only fetch the public page you submit and never store its content.
You give one public page URL
Any page that opens without a login โ it doesn't have to be yours, competitors work too.
We fetch that page, plus robots.txt and llms.txt
Public content only โ no login, we don't run your scripts, and we touch nothing on your server.
Seven checks, scored 0โ100 on the spot
No AI is called at any point, which is why it returns in seconds and costs nobody any quota.
The weights are fixed and public. Publishing them means people can optimize against them โ which is the point: we sell verification, not a secret recipe.
| Check | Weight | Why this much |
|---|---|---|
| robots.txt access for AI crawlers | 25 | The only check that can zero out visibility outright โ block crawlers site-wide and nothing else you do is ever seen. |
| Answer-first structure | 20 | Models lift the passage that already reads like an answer. A conclusion buried eight screens down may as well not exist. |
| Structured data (JSON-LD) | 15 | The machine-readable copy. Same content, but marked up it's far likelier to be attributed to you correctly. |
| Title & meta description | 15 | The oldest signal, and the one least excusable to miss โ it decides how the page gets summarized. |
| FAQ signals | 10 | Q&A shape matches how models lift text, but not every page should have one โ hence the middling weight. |
| Language declaration (html lang) | 10 | One attribute, and it matters most on bilingual sites: get it wrong and engines file your content under the wrong language. |
| llms.txt | 5 | An emerging convention no major engine has publicly committed to reading. Nice to have, not required โ hence the lowest weight. |
Per check: a pass earns full weight, a warning half, a failure zero. One exception โ if robots.txt can't be fetched we treat it as open and award full credit: not fetchable doesn't mean misconfigured (it may be a network blip, or you simply have no such file, and no file means everything is allowed). Docking the heaviest check over one failed fetch would manufacture anxiety, not information.
It measures whether your page is ready โ not whether AI will mention you.
The difference is practical: readiness is entirely yours to control and you can score full marks by fixing things. Whether AI mentions you depends on what everyone else in the category published and how the engine summarizes that day. A 100 guarantees nothing, and a 60 rules nothing out. Anyone claiming to derive ranking outcomes from a page score is selling something they can't deliver.
To find out what AI actually says about your category today, that's the other tool: Run a free checkup โ
How ready one page is for AI engines: whether it's structured clearly, crawlable, and carries machine-readable signals. It's a static check โ it reads that one public page and calls no AI at all, which is why it returns in seconds.
Seven weighted checks, and the weights are fixed and public: robots.txt access for AI crawlers 25, answer-first structure 20, structured data 15, title & meta 15, FAQ signals 10, language declaration 10, llms.txt 5. A pass earns full weight, a warning half, a failure zero. It's deliberately coarse โ it's a heuristic, and dressing it up as precise would be the dishonest part.
Because it's the only check that can zero out visibility outright. The other six make things better; blocking AI crawlers site-wide means nothing else you do will ever be seen. Fix that one first.
No. When we can't fetch it we treat it as open and award full credit. Not fetchable doesn't mean misconfigured โ it may be a network blip, or you simply don't have the file (no file means everything is allowed). Docking the heaviest 25 points over one failed fetch would manufacture anxiety, not information.
No, and this matters. Diagnose measures whether your page is ready, not whether AI will mention you. The first is entirely within your control; the second depends on the whole content landscape. To find out what AI actually says today, run the free checkup โ that's the tool that measures visibility.
We weight it 5 โ the lowest of the seven. It's an emerging convention and no major engine has publicly committed to reading it. Nice to have, not required. If time is short, do answer-first structure and structured data first (35 points combined); the payoff is far better. We publish one ourselves mostly because a GEO company ought to eat its own dog food.
All standalone, all free to finish without an account.
Free checkup
30-second AI-visibility check, no login
Benchmarks
Public, anonymized industry percentiles
Embed widget
One line of code to embed the checkup
