Severity
Three levels, which determine both urgency and the score impact:
Work errors first. Notices frequently need no action at all — a
noindex on a thank-you page is correct.
Meta tags
Duplicate title and description issues depend on crawl coverage — both copies must be crawled for the duplication to be detected. A narrow crawl under-reports them.
Links
Broken links are reported with the HTTP status they returned, and redirect chains with the number of hops — enough to triage without opening each URL.
Orphan pages need a broad crawl to detect reliably: a page can look orphaned simply because the crawl never reached whatever links to it.
Performance
Each issue reports its actual measurement, so you can see how far past the threshold a page is.
These are crawl-time measurements from a single request. For field-quality performance data, use Web vitals.
Images
Indexing
The two notices are deliberately low severity because they’re usually intentional. Read them as a checklist — “did I mean to exclude these?” — rather than as defects.
They’re also the issues to check first when a page mysteriously won’t rank.
Content
The 300-word threshold is a blunt rule. A well-targeted product or contact page under 300 words isn’t a problem; a category page padded to 400 words still is.
Security
Both are errors, and HTTPS also carries a +5 score bonus — fixing it moves the score twice.
Schema
A notice, not an error, because schema isn’t a ranking factor in itself. It does affect how your result is displayed, which affects clicks.
Working through a list
- Errors first, largest deduction and real damage.
- Sitewide patterns before one-offs. An issue on every page is one template fix; the score moves far more per unit of work.
- Then the +5 bonuses — HTTPS,
robots.txt, sitemap. Three quick wins worth 15 points between them. - Re-run the audit. The score belongs to the crawl that produced it and won’t update on its own.
- Compare the two crawls to confirm what actually got fixed.