The audit gives your site a number and a grade. Here is exactly what drives it, so the result is something you can trust and act on.
24
checks per audit
2
scored categories
5
grade bands
Drag the slider. The same thresholds the audit uses, 40, 60, 75 and 90, decide where a score turns into the next letter.
/100
SEO Score
This maps a final score to its grade. The score itself comes from the findings and how widespread each one is.
Solid technical foundation. Minor refinements only.
Healthy, with a few issues worth fixing.
Working, but several problems are holding it back.
Real issues across the site that cost visibility.
Serious problems that need attention before anything else.
On-page content and technical setup are scored separately, so a strong result in one cannot hide a weak result in the other. The overall score never rises above the weaker area.
What a reader and a search engine see on the page: titles, meta descriptions, headings, content depth and image alt text.
Some of the thresholds we measure against
How the site is built and served: HTTPS, indexing, canonicals, sitemap, structured data, broken links and international tags.
Each check has a severity. A problem that removes a page from search costs far more than one that only trims a result's appearance.
Stops a page from ranking or being found at all: no HTTPS, a noindex tag, a missing title, duplicate titles or a missing canonical. While any critical issue is open, the score cannot reach an A.
Does not remove the page, but loses ground to competitors: titles outside the ideal length, missing meta descriptions or H1s, thin content, missing alt text or broken internal links.
Refinements that sharpen the result without changing whether the page ranks: heading order, hreflang details or a description slightly outside the ideal length.
A check that does not apply to a site is neither a pass nor a fail. It is marked N/A and left out of the maths entirely, so it never penalises the score and never inflates it. A whole category can be N/A, in which case it is dropped from the weighted average; when a site's content is only readable after JavaScript runs, the overall score itself becomes N/A rather than a guess.
Content is JavaScript-rendered, not present in the HTML.
Only one page was analysed, nothing to compare against.
The site declares no translations (hreflang).
A small site, where a sitemap or robots.txt brings marginal benefit.
N/A never counts against you. It marks the limits of what a fair, honest read of your site can measure, and those limits are stated on the report itself, next to each skipped check.
AI search engines, and the crawlers behind ChatGPT, Claude, Perplexity and Google's AI answers, read the raw HTML your server returns and do not run JavaScript. Our audit reads a site the exact same way, so what we can see is what they can see. That is how we know whether a site is readable by AI. We say readable, never that you will be cited.
What we check for AI-readiness
AI crawlers do not run JavaScript. If your content only appears after JS runs, they see an empty page.
Your robots.txt lets GPTBot, ClaudeBot, PerplexityBot and Google-Extended read the site.
Schema.org or microdata spell out what a page means, so a machine reads it without guessing.
The title is the first thing an AI uses to name and cite your page.
The meta description is a ready-made summary an AI can quote directly.
Headings tell a parser where each topic starts, so answers quote the right passage.
A thin page gives an AI nothing to summarise or cite.
An llms.txt hands AI models a clean, curated map of your site.
The same issue is weighed by how much of the site it touches. A missing title on one page of fifty barely moves the score; the same fault on every page drags it down hard. One bad page does not sink a good site, and a site-wide fault is never rounded away.
We read the real HTML your server returns, the same way a search engine's crawler does. We report only what that shows, and never invent signals we cannot measure. Content that only appears after JavaScript runs is not read, by us or by AI search. When a site relies on it, we flag it, because that is a real loss of visibility, not a blind spot on our side. A check we cannot run, or a category we cannot see, is marked as not evaluated rather than given a score it did not earn. The free audit reads up to 25 pages; a full-site sweep is part of the paid work.
The score is meant to be checkable, not a black box. Here is what usually comes up.
A single critical issue, like a missing title, no HTTPS or a noindex tag, caps the score no matter how clean the rest is. The more pages it affects, the harder the cap. The findings on your report name the exact issue and the pages it is on.
Yes. Each audit reads your site's current HTML, so fixing an issue and re-analysing raises the score. A re-analysis is a new report, and the previous one is kept so you can see the change.
A check that does not apply to your site is left out of the maths entirely, so it neither helps nor hurts the score. Content that only appears after JavaScript runs, a single-page site, or a site with no declared translations are the common cases.
The free audit reads up to 25 pages, enough to surface the patterns that repeat across a site. A full-site crawl is part of the paid work, where a specialist reviews the findings with you.
No, and it is not meant to. It is a fast, honest read of the technical basics and whether your site is ready for AI-driven search, a clear starting point. The deeper, ongoing work is what our team does with you.
See your score, the exact findings behind it, and a generated llms.txt, in a couple of minutes and at no cost.