Spam score: what it catches and what it misses
Spam score is an index’s estimate of how manipulated a domain’s inbound links look: link networks, comment spam, patterns of anchor text that no editor would produce. It is a useful third opinion and a poor first one.
What does spam score measure?
Patterns in the domain’s backlink profile associated with manipulation, scored 0–100 by the backlink index. It says the rating was built out of bad company; it does not say the site has no readers, and a site can have a clean profile and still be a hollow domain.
How do we use it?
A score over 50 fires the spam_score quality test and marks the domain suspect; a score of 30 or more takes points off the quality component of the Placement Score without flagging. It is one of five tests, counted separately because they overlap; on its own it is the smallest contributor to the suspect share.
What does it miss?
Expired domains bought with a clean history, networks built carefully enough to look editorial, and every problem that is not about links: thin content, no readers, a subdomain trading on a parent. Those are what the other four tests are for, and why no single score decides anything here. Method on the quality methodology page.
You do not need to filter on it; suspect domains are excluded from plans by default and the reason is named on the row.
Methods: research methodology.