How the SEO score is calculated
The formula is five weights and one subtraction. The interesting part is the decision underneath it: a page counts once per area, no matter how many problems it has there.
Every completed crawl gets a number between 0 and 100. It appears on the report cover and on the chart that compares runs. This is what it is made of.
The formula
The score starts at 100 and subtracts, area by area, a share of that area's weight proportional to how many pages it affects:
- Indexing — 25
- Content — 30
- Experience — 20
- Links — 15
- Media — 10
If a quarter of the pages have a content problem, content gives up a quarter of its 30 points. Add the five losses, subtract from 100, round. That is the whole calculation.
Do the math
Say how many pages each area affects and watch the score come out of the same function the product uses.
out of 100
A page counts once per area
This is the decision that shapes the number. A page missing a title, missing a description and missing an H1 has three content problems — and counts as one affected page.
Counting occurrences would make the score answer the wrong question. Three problems on the same page are one page to open and one deploy to ship; one problem on three pages is three. The score is meant to estimate the size of the job, not the length of the list.
What it refuses to do
A run with no pages gets no score at all — not zero. Zero is a measurement, and “we could not reach the site” is not a measurement. It shows as absent, and the run stays out of the comparison chart.
Comparing two runs
Two crawls of the same site rarely cover the same pages: each discovers addresses from the entry URL. A score that moved four points between different samples measured the sample. Use “Re-run” to crawl exactly the previous set — then the difference is about the site.
The report says which of the two situations produced the number you are looking at. It seemed dishonest to show the delta without that.