Which pages a scan reads, and why the report says something is missing

When the report says something is missing and you know you have it, it is almost always on a page the scan did not read.

Six pages

A scan reads at most six pages in full, and the sitemap as a list. The six are chosen in this order:

1. Your homepage, always.
2. Your strongest about-style page.
3. Up to two pages that last scan's advice named, so the advice can be checked.
4. Up to two pages that are new since the last scan.
5. Pages that represent your main topics, judged by how close they are to what the homepage is about.

The report lists exactly which pages it read, under "From the N pages we read". Consecutive re-scans rotate through pages the advice names, so a page missed this time is read next time.

What "not found" means

A finding that says an author bio, a client result or a number is missing means it was not found on those six pages. The report is written to say so ("not detected on the pages we read"), and it is forbidden from claiming something is absent across the whole site. If the thing exists elsewhere, dismiss the recommendation with Already have this and it will not come back. Better still, put it on the pages that get read: the homepage and your about page are always in the sample.

What the whole sitemap is used for

Counts are site-wide: how many pages you have, how many sit in each topic, whether a page the advice tells you to create already exists. Content is not. A page in your sitemap that the scan did not open has not been read.

Videos and things behind JavaScript

A page whose stories are in a video, or whose text only appears after JavaScript runs, cannot be read by the scan, and cannot be read by the AI engines either. The report is told never to treat an unread video as missing content, and never to lower a score for it. If your proof lives in video, the fix is the same one the engines need: put the words on the page.