It is one score out of 10, nine category scores, two AI-search readiness scores, and a ranked list of fixes that each quote the number behind them. Below is a real report for braydensautomations.com, scored 9.6 out of 10, with every section explained.
Your meta description runs to 181 characters, so search results will cut it off around 160. Rewrite it to 70–160 characters covering what you offer, what sets it apart, and the next step.
Your title is 67 characters, so search results truncate it at around 60. Lead with your main offering and audience, and put the brand name last.
Real output from a real audit. Not an illustration.
A single number out of 10, and a label from Poor to Excellent. It is a weighted mean of the categories below it, not a separate judgement, so it always reconciles with the breakdown. Measured categories like performance and SEO carry more weight than subjective ones like design, which is what keeps repeat audits of an unchanged site landing in the same place.
Design, UX, performance, mobile, SEO, accessibility, trust signals, content and CTA effectiveness, each out of 10 with the evidence that produced it. A category with nothing to measure is reported as not measured rather than scored zero, because a missing signal and a failure are different things.
Two scores out of 10 for AI search. AEO is how ready the page is to be quoted directly by assistants like ChatGPT and Perplexity; GEO is how well AI search can identify and cite the brand. Each comes with the individual signals that passed, partially passed or failed.
Every finding names the measurement behind it, what to change, why it matters and how to confirm the fix worked. The ranking is by how much fixing it would move the score, so the top of the list is the shortest path to a better number.
Run one on a site you are pitching. The free account includes one full audit every 7 days, and no card is needed.
Run a free auditAn overall score out of 10, nine category scores with the evidence behind each, AEO and GEO readiness scores for AI search, and a prioritised list of fixes that each quote the measurement that produced them.
One screen of scores plus a prioritised fix list, typically six to twelve findings depending on the page. It is also downloadable as Markdown, and as a white-label PDF on paid plans.
Yes. Growth and Enterprise plans export it as a PDF carrying your agency name, logo and accent colour, with no mention of AI Website Critic anywhere in it.
Yes. Scores come from a fixed, published rubric applied to measured Lighthouse and page data, not from a model's opinion, so auditing an unchanged page twice gives the same numbers.