43 checks on one page, against Google’s own documentation.
Paste a URL. The page is fetched as served, the parts Google documents are pulled out of it, and a language model scores each of 43 checks and says why. You get a reading in about a minute, with the working shown check by check.
The analysis is model-generated. It is a heuristic reading of one page, not a measurement, and the section below states exactly what it cannot see before you run it rather than after.
How it scores.
The rubric is the Google Search Central compliance rubric: 43 checks grouped into six dimensions, drawn from Google’s published documentation on Search Essentials, helpful content, crawling and indexing, ranking systems, search appearance and page experience.
Each check scores 2 for a pass, 1 for a partial and 0 for a fail, so the raw total is 86. That is normalised to 100 and banded: 90 and above reads as compliant, 75 to 89 as minor gaps, 55 to 74 as significant issues, below 55 as a critical failure. Every check returns one sentence of reasoning, and most return the evidence from your page that it was read from.
What this tool cannot see.
A free tool that overstates itself costs more credibility than it earns. Four limits apply to every run, and they apply before you paste a URL rather than being discovered in the result.
The disclosure for a model-generated reading.
The same disclosure runs on every figure we publish. Here it says that the instrument is a language model, which is the reason nothing this tool returns is presented as a measured number.
- instrument
- A language model, prompted with Google Search Central's published guidance. Not Caul, and not a measurement.
- what was measured
- Compliance of one URL against 43 documented checks in six dimensions, scored 0, 1 or 2 each, out of 86 and normalised to 100.
- how
- The URL is fetched once as served, along with its robots.txt. Title, meta, canonical, headings, links, JSON-LD, image alt coverage and a text sample are extracted and passed to the model, which returns a score, a sentence and usually a quoted piece of evidence per check.
- over what window
- One run, at the moment you press the button. Nothing is stored and nothing is monitored.
- what this cannot tell you
- A model's reading is not a measurement, and two runs of the same URL can disagree. It sees one page as served, with no JavaScript executed and nothing behind a login. It has no access to your rankings, traffic or Search Console, so it cannot say what any finding costs. It is a Google-documentation check and carries no claim about whether AI answer engines name you.
The measured version is the free report.
This tool reads one page against documentation. The audit measures something else entirely: how often the engines name your firm across a fixed prompt set, against competitors you name, with the coverage, the run count and the capture window attached to every figure. It is free, delivered on a call, and you keep the data whatever you decide.
Free. The $1,500 is what this sweep is priced at when it is sold, not a survey of anyone else’s rates.