The read ladder
The audit tries these sources in order and uses the first one that returns a real page:
- Read directly: the page as our reader fetches it (it always says who it is, and follows your robots.txt).
- Read after JavaScript: for a page built in the browser, the same page after its scripts ran in our renderer.
- Read through your connector: for your own site with the SEOmise connector, the page sent from inside your site.
- Read from the site's code (approximate): for a page built by JavaScript that our renderer could not open, the text is read from the page's own script files (they are never run): about how many words, the headings, and the title and description the code sets. Only files on your own site are read.
- Measured by Google Lighthouse: Google's PageSpeed Insights tests the page from Google's own servers. It measures the SEO basics (title, description, status, whether search engines may index it, link text, image alt text, language), speed and accessibility.
- From Search Console: when your Search Console is connected, Google's own URL Inspection of the page is added for information.
We never solve a firewall's challenge and never pretend to be a browser or another robot.
When your firewall blocks us
If Cloudflare or another firewall blocks our reader, you still get a grade, from what Google Lighthouse measured, with a card that shows how to let us in. Checks that need your page's HTML (headings, amount of text, sharing tags, structured data) are listed as not measured until we can read it. Let us in (see "Let SEOmise reach a site behind Cloudflare") and the next audit reads every check directly.
"Couldn't read this page" appears only when every source failed, for example when the site is down.
Site scans behind a firewall
When a firewall blocks a scan's start page, or most of its pages, and no connector can read them, Google Lighthouse measures the start page and up to 10 more of the blocked pages. Your site gets a health score from those pages, and the firewall card says so, for example "Site score from 11 pages measured by Google Lighthouse; 230 pages not read directly". The other pages stay "not read", never failed, until you let us in.
The labels
Each check shows its source: read directly, read after JavaScript, read through your connector, read from the site's code (approximate), measured by Google Lighthouse or from Search Console. When the text is drawn by JavaScript, or the page was not read directly, a short note at the top of the report says so in plain words, with the builder when its mark was seen in the page. Under the summary, the report says how many checks were measured, for example "31 of 40 checks measured". A check that is not measured costs nothing.
Not what you needed? Ask Mise or contact support.
