An SEO score is a derived metric: an audit tool runs its checks on your pages, subtracts points for what fails and weights the result by severity. Google does not publish an SEO score for any site, and no score predicts where you will rank. Read it the way you read any composite number on a dashboard: good for catching regressions on your own site, poor for comparing two tools.
The three inputs behind every SEO score
Every score comes from the same three ingredients, and changing any of them moves the number:
- Checks. Which tests run. Lighthouse's SEO category is a short list of pass or fail audits on one page, such as the status code, meta description, canonical, robots.txt and link text. Ahrefs' free Site Audit looks for more than 170 issue types and Semrush's Site Audit runs more than 140 checks across a crawl.
- Weights. How much each failure costs. One tool treats a missing H1 as serious, another barely counts it.
- Scope. Which pages get tested. A score for your home page says nothing about the blog template.
Because all three differ between vendors, the same site can look healthy in one tool and mediocre in another on the same day. Neither tool is lying. They are measuring different things.
A worked example
seogeoaeo.ai publishes its formula, which makes it a good one to walk through. Each check starts at 100. A high-priority fix costs 25 points, a medium one 10 and a low one 2, and a fix that repeats across several pages costs more. An area score is the average of its checks, and a check that returned no result is left out instead of counting as a pass. While any high-priority fix is open, the score is capped at 89.
Say four checks feed one area score for your home page. Robots and on-page come back clean at 100 each. Structured data has three low fixes and lands at 94. The indexing check finds a noindex tag on the home page, a high-priority fix, and drops to 75. The plain average is 92, which would read as green. The cap holds it at 89, out of the green, until the noindex is gone. That cap is the point: a severe blocker should never hide behind clean results elsewhere.
How to use a score without being misled
- Keep one tool and one page set. Compare this week's number with last week's from the same tool on the same URLs, never across tools.
- Watch the delta after a deploy. A sudden drop after a release behaves like a failing test: something in that diff broke a tag, a redirect or a robots rule.
- Read the fixes, not the number. A score is only useful when each lost point maps to a change you can make.
- Separate "not checked" from "passed." A tool that counts a failed check as a pass inflates the score.
What no SEO score can tell you
Rankings depend on competition, links and search intent, which a page check cannot see. Traffic lives in Google Search Console, which is free and shows clicks, impressions and average position. Content quality is invisible to tag checks, so a thin page with perfect markup can score well. And pages outside the tool's scope are simply unmeasured.
Getting a score you can trace
The seogeoaeo.ai website check reports SEO readiness, AEO readiness, AI access and an overall score from 0 to 100, with the same bands as PageSpeed Insights, and a fix list behind every lost point. The home page preview is free without an account; the full home page check costs 67 credits. Run it again after your fixes and it reports the percent of issues fixed and the score change. Coding agents can run it over MCP and loop until the list is clean.
For a free single-page number, the Lighthouse SEO score in PageSpeed Insights does the job. For rankings and backlinks, look elsewhere; the seogeoaeo.ai check does not measure them.