What Moz Spam Score actually measures
Moz Spam Score is not a penalty counter and it never reads Google's mind. It is a probability. Moz studied a large sample of domains that Google had penalized or de-indexed, isolated the structural features those domains shared, and built a model that returns a percentage from 1 to 100. A score of 30 means that roughly 30 percent of sites carrying a similar feature footprint were found penalized or banned in Moz's reference set, not that your site is 30 percent of the way to a manual action.
That distinction matters because the number gets read backwards constantly. A high Spam Score flags correlation, not causation. Plenty of legitimate sites score high because they share surface traits with spam: a brand-new domain, a thin link profile, an exotic TLD. The metric describes company kept, not guilt proven, and Moz has been explicit about that framing in its documentation since the 2019 rebuild.
Moz's own walkthrough is the fastest way to anchor the definition:
The original 2015 version exposed 17 binary flags. The current model collapses those signals into a single predictive percentage, which is cleaner to consume but easier to misread as a verdict. Treat it as what it is: a triage signal that tells you where to look, never a reason on its own to reject a domain.
How the metric works in 2026
The engine behind the score is a feature-correlation model. The inputs are the kind of things Moz can observe at scale: the ratio of followed to nofollowed links, the relationship between MozTrust and MozRank, anchor diversity, the proportion of low-quality referring domains, TLD distribution, and on-page markers like thin or templated content. None of these is decisive alone. The model weighs them together and outputs the percentage.
This Whiteboard Friday digs into why the metric exists and how to apply it:
In 2026 the honest position is that Spam Score is a useful but aging input. Google has spent a decade getting better at ignoring spammy links rather than penalizing the target, a shift the company described when it folded the link spam work into its core systems via SpamBrain (Google Search Central, 2022 link spam update). The practical consequence: a high Spam Score on a domain pointing at you is far less likely to drag you down today than it was in the Penguin era. The metric still earns its place as a screening tool, but the stakes attached to any single reading have dropped.
You can read the score in Moz Link Explorer, inside Moz Pro campaigns, or through several third-party checkers that pull the Moz API. Operationally we never act on the headline number. We open the referring-domain list, sort by Spam Score, and inspect the tail by hand. The threshold question, what counts as high, has no universal answer: Moz historically treated 1 to 30 percent as low, 31 to 60 as moderate, 61 plus as high, but those bands are guidance, not a kill switch.
One operational detail trips people up: the score is not real time. Moz refreshes it when its link index updates, so if you remove or disavow links, the change only surfaces after the next index cycle. Do not judge a cleanup by checking the number the next morning.
Where it fits in a netlinking operation
In a live netlinking workflow Spam Score is a filter at the prospecting stage, nothing more. When you evaluate a publisher before placing a link, a high score is a prompt to investigate, not an automatic veto. We pair it with a manual read of the link profile, real traffic data, and the editorial quality of the pages, because the score alone cannot tell a young legitimate site from a recycled spam domain.
This is where owning your inventory changes the calculus. When you run a network of media in-house, as we do at Stringer, the toxicity question is settled upstream: you control the link profiles of the domains you place from, so you are not reverse-engineering a stranger's Spam Score after the fact. For anyone buying placements externally the discipline is the same as ours internally, vet the source before the link exists. You can browse the catalogue of media we operate, with each domain's metrics visible up front rather than discovering a problem after purchase.
When the goal is to source a backlink straight from a vetted publisher, Spam Score belongs in the same bucket as the count and quality of referring domains: one coordinate in a multi-dimensional read, never the verdict by itself.
Spam Score against the other toxicity metrics
Every tool models trust and toxicity differently, so the numbers are not interchangeable. Moz Spam Score is a penalty-probability percentage. Majestic's Trust Flow approaches the same question from the opposite side, scoring proximity to a seed set of trusted sites, so a low Trust Flow hints at weak trust rather than active spam. Semrush does not fold toxicity into its Authority Score at all; it surfaces a separate Toxicity Score inside the Backlink Audit, which is the closer analogue to what Moz reports.
Ahrefs deliberately publishes no toxicity metric. Its Domain Rating measure of link popularity answers strength, not cleanliness, and the team has argued publicly that a generic toxicity figure does more harm than good because it nudges people to disavow links Google already ignores. That stance is worth taking seriously. Comparing a Moz Spam Score of 40 to a Trust Flow of 15 or a Semrush Toxicity Score of 60 is meaningless arithmetic: different reference sets, different definitions, different sampling. Use each within its own tool and never average them.
Which one should you trust? None as an absolute, all as relative signals within their own ecosystem. In practice we lean on Moz Spam Score for a quick first pass because the percentage framing is intuitive, then corroborate with a manual look at the link's context and the publisher's real audience. When three tools disagree about a domain, that disagreement is itself information: it usually means the domain sits in a grey zone where human judgment, not a metric, makes the call.
Reducing a high Spam Score without overreacting
The standard advice is a sequence: audit the backlink profile, identify toxic links, disavow them, improve thin on-page content. Each step is reasonable, but the order of importance is backwards in most tutorials. The first question is not how to lower the number, it is whether the number reflects anything real.
A practical fix-it walkthrough for anyone facing a genuinely high score:
Here is the senior stance: disavowing links to chase a lower Moz Spam Score is, in 2026, mostly wasted effort. Google's own guidance is that the disavow tool is for cases where you have manipulative links and either a manual action or a credible expectation of one (Google Search Central, disavow documentation). Outside that, the algorithm discounts spammy links rather than counting them against you. Lowering a vanity metric by disavowing aggressively can even strip out links that were passing modest value. If your Spam Score is high purely because the domain is new, thin, or on an unusual TLD, the fix is to build a real site and earn real links, not to file a disavow.
Where the score does point at a genuine problem, a paid-for link from a known PBN, a footer-wide sitewide, a hacked-injection pattern, then audit and disavow with intent. The cleanup that actually moves the needle is upstream: stop acquiring the kind of links that raise the score in the first place. A campaign planned around editorial placements rather than raw volume never builds the profile that lights up a toxicity checker.
Common mistakes and the myths behind them
The most expensive mistake we see is treating Spam Score as a penalty itself. A high score does not guarantee a Google penalty; it estimates a probability from shared features, and correlation is not a manual action. Sites with scores in the high bands rank fine every day when the underlying signals are benign.
The second myth conflates Spam Score with Domain Authority. They are separate models answering separate questions: link-volume and trust style metrics measure strength, Spam Score estimates risk. A domain can carry a high DA and a high Spam Score at the same time, which happens often on aged domains with messy histories.
The third myth is that only a paid Moz Pro seat can reveal the score. It surfaces in free Link Explorer queries and through third-party checkers built on the Moz API, so cost is not the barrier. The real failure mode is letting any single third-party number drive an irreversible decision like disavowing. The number is an input. The judgment stays with you.