Domain Spam Score: What triggers it and what the Moz signal predicts

What is Domain Spam Score

Domain Spam Score is a Moz metric expressed as a percentage from 0 to 100 percent that estimates how closely a domain resembles the sites Google has penalized or banned. The reading is a predictive risk proxy generated by a machine-learning model, not a record of any Google sanction.

Moz launched Spam Score in 2015 with 17 subdomain flags, then expanded the model to 27 features and shifted the output from a flag count to the 1-100 percentage.

A Spam Score of 8 percent describes a clean-looking profile; a Spam Score of 70 percent describes a profile that shares characteristics with the penalized cohort.

Moz publishes Spam Score through Link Explorer and the Moz API.

The Spam Score value appears in Moz Link Explorer alongside Domain Authority and Page Authority. The same value is exposed through the Moz API.

Charles Floate has documented spam-signal screening across PBN A/B testing, and the pattern is consistent:

• A domain carrying inflated link volume with a rising Spam Score reads as a manipulated profile.
• A domain with an organic profile holds a low Spam Score over time.

The metric sits in the same Moz toolset as Domain Authority because both read the Mozscape link graph.

Spam Score describes resemblance, not a confirmed sanction.

A high Spam Score names a statistical resemblance to penalized sites. The model has no access to Google manual-action records, so the percentage predicts risk instead of reporting an applied penalty.

Practitioners reading the score treat a high value as a prompt to investigate the link profile and on-page signals, not as proof that Google has acted.

The distinction defines how the SEO industry reads the metric across the Domain Authority & Metrics toolset.

What Spam Score measures: the penalized-cohort percentage

Spam Score measures the percentage of sites with features similar to the target domain that Google has penalized or banned. A machine-learning model trained on millions of penalized and healthy sites computes the value.

The model identified the features that penalized sites share disproportionately. Each domain is scored by how strongly its profile matches that penalized cohort.

A reading of 40 percent states that 40 percent of sites sharing the domain’s feature set landed in the penalized group.

The model is trained on penalized sites, not on Google ranking factors.

Moz trained the Spam Score model on a dataset of sites Google penalized or banned, contrasted against healthy sites. The training target is penalization, so the output predicts penalization risk.

This is the reason the score reads as a risk proxy: it learns the fingerprint of the penalized population and reports how closely a domain matches it.

Cyrus Shepard at Zyppy has covered the resemblance-based reading of link-risk metrics across correlation research through 2022 to 2024.

A percentage reading is comparative, not a guarantee about a single domain.

A Spam Score of 30 percent does not predict the fate of one specific domain. It states that across the population of sites sharing that feature profile, 30 percent were penalized. The reading is probabilistic.

A clean domain can carry an elevated score from a coincidental feature match. A manipulated domain can carry a low score before flags accumulate. For that reason the score is read alongside the Trust Flow to Citation Flow ratio instead of in isolation.

The Spam Score flag categories

Spam Score tracks 27 flags across 4 categories: link-profile signals, domain-level signals, on-page signals, and reputation signals. Link-profile flags carry the strongest predictive weight because manipulated link graphs are the clearest fingerprint of the penalized cohort.

Domain-level and on-page flags add structural and content evidence, and reputation flags compound the reading when historical penalty or blacklist signals are present.

A low MozTrust-to-MozRank ratio is the headline link-profile flag.

A domain accumulating link volume faster than trust shows a MozRank that outpaces its MozTrust.

The gap is the signature of a profile built on quantity instead of editorial validation, and it maps to the same trust-versus-volume divergence that the Trust Flow to Citation Flow ratio exposes at Majestic.

Charles Floate has documented this divergence as a manipulation tell across PBN audits, where inflated volume raises the count while trust-seed proximity refuses to follow.

Domain-level flags catch patterns that pre-date the link profile.

Spam-correlated TLD patterns, an unusual domain length, and numerals inside the domain name are structural features the model weights because penalized sites carried them disproportionately. These flags fire before a single backlink is evaluated.

An aged domain inheriting a clean structural profile starts the Spam Score reading from a stronger baseline than a freshly hand-registered numeric domain on a spam-correlated TLD.

Spam Score tiers: low, medium, and high risk

Moz Spam Score reads across 3 risk tiers: 1 to 30 percent is low risk, 31 to 60 percent is medium risk, and 61 to 100 percent is high risk. Each tier maps to a distinct action.

A low-tier reading clears a domain for normal use. A medium-tier reading prompts a link-profile and on-page investigation. A high-tier reading demands an audit and a cleanup plan before the domain carries a project.

The tiers convert the percentage into an acquisition decision.

The low tier clears a domain; it does not certify quality.

A Spam Score in the 1 to 30 percent band states that the domain resembles the healthy population, which clears the risk question. It does not confirm topical relevance, content depth, or Trust Flow strength.

A low Spam Score paired with a weak Trust Flow still flags a thin profile. The tier answers the penalization-risk question and hands the quality question to the cross-validation metrics.

The high tier prompts an audit, not an automatic rejection.

A reading above 60 percent signals strong resemblance to the penalized cohort, which is the reason the SEO industry treats it as an action threshold.

The flags driving the score are inspectable. A high reading from low link diversity is a different case from a high reading from a spam-correlated TLD.

Aged-domain buyers read the flag breakdown so the audit targets the cause. The SEO Domains screen routes high-tier readings out of the catalogue before they reach a listing.

How Spam Score adjusts Domain Authority in DA 2.0

Spam Score was integrated into the Domain Authority calculation in the DA 2.0 retraining of March 2019 as a downward adjustment, so a high Spam Score pulls the published Domain Authority lower. Before DA 2.0, manipulated profiles inflated Domain Authority through link volume without a proportional Spam Score consequence.

After the retraining, the Spam Score adjustment is baked into the model. Profiles with high spam resemblance saw Domain Authority drops as the adjustment compounded.

The integration ended the era of penalty-free volume inflation.

Before March 2019, a profile built on link volume held a high Domain Authority while carrying a separate, ignored Spam Score.

The DA 2.0 retraining tied the two together, so volume tactics that raise the link count now raise the Spam Score and trigger the downward adjustment.

Charles Floate documented the resulting Domain Authority drops on manipulated profiles across PBN A/B testing through 2019 and 2020, where inflated scores corrected within 1 to 2 index refreshes.

A clean Spam Score keeps the Domain Authority the model awards.

A domain with a low Spam Score takes no meaningful downward adjustment, so its Domain Authority reflects the link-graph strength the model measured.

This is the acquisition target: Domain Authority the model will keep, not Domain Authority headed for a Spam-Score correction at the next refresh.

Aged-domain buyers verify the Spam Score alongside Domain Authority so an inflated score does not pass as inherited authority.

Spam Score compared with the Trust Flow to Citation Flow ratio

Practitioners cross-check Spam Score with the Majestic Trust Flow to Citation Flow ratio because both metrics expose the same trust-versus-volume weakness from independent indexes. Citation Flow counts link volume; Trust Flow weights trust-seed proximity.

When Citation Flow runs 3 to 4 times the Trust Flow, the profile is volume-heavy and trust-light, the signature that also drives a rising Spam Score.

Agreement across the two readings confirms the risk; a clean Spam Score paired with a balanced ratio confirms the profile.

A balanced TF:CF ratio corroborates a low Spam Score.

A Trust Flow to Citation Flow ratio near 1:1, or a Trust Flow that holds above half the Citation Flow, describes a profile where trust tracks volume. That balance corroborates a low Spam Score from an independent Majestic index.

Constantin Oesterling has documented the multi-index screening discipline across guest-posting authority audits through 2021 to 2024, where a metric confirmed by a second vendor carries more weight than a single reading.

Anchor distribution is the third reading that completes the cross-check.

A natural anchor profile spreads across three documented bands:

• Branded anchors at 40 to 60 percent.
• Naked URLs and generic anchors at 30 to 40 percent.
• Exact-match commercial anchors under 15 percent.

A profile concentrating exact-match commercial anchors drives both the Spam Score link-profile flags and the Citation-Flow-heavy ratio.

Matt Diggity at Authority Builders has documented the safe anchor distribution across affiliate-site audits, and the SEO Domains screen reads the anchor breakdown alongside Spam Score and the TF:CF ratio.

Why Spam Score is a Moz risk proxy, not a Google penalty

Spam Score is a Moz predictive risk score computed on the Mozscape link graph, and Google does not use it in the ranking algorithm. The score predicts how closely a domain resembles penalized sites. It does not record, trigger, or reflect a Google manual action.

A domain can carry a high Spam Score with no Google penalty applied, and a penalized domain can carry a moderate Spam Score. The two are correlated proxies of link quality, not the same record.

Spam Score predicts risk; it does not detect a sanction.

The Moz model reads public link-graph and on-page features. It has no view into Google manual-action records, so it cannot detect an applied penalty. The percentage forecasts the probability that a domain belongs to the penalized population.

Reading Spam Score as a confirmed penalty overstates the metric; reading it as a resemblance forecast uses it as the SEO industry intends. Charles Floate frames the score as a screening signal instead of a sentence across PBN documentation.

The risk-proxy reading turns Spam Score into an acquisition filter.

A buyer reading Spam Score as a forecast acts on it: a high reading triggers a link-profile and on-page investigation before acquisition, and a low reading clears the penalization-risk question.

Aleyda Solis applies this resemblance-versus-record distinction in enterprise link-risk audits. The filter is the value: the score sorts inventory into investigate-further and cleared-for-acquisition before a buyer spends time on a listing.

How aged-domain buyers screen Spam Score at the SEO Domains marketplace

Aged-domain buyers screen Spam Score at the SEO Domains marketplace by reading the percentage inside the catalogue’s multi-index screen instead of in isolation. Each reading is verified before a listing reaches buyers against the flag breakdown, the Trust Flow to Citation Flow ratio, and the anchor distribution.

A high-tier reading routes the domain out of the catalogue at ingestion. The inventory a buyer browses has already cleared the penalization-risk question.

The catalogue reads Spam Score as one of several cross-validated signals.

SEO Domains catalogue listings surface Spam Score alongside Domain Authority, Domain Rating, Trust Flow, and Citation Flow, so a buyer reads the agreement or disagreement across the set instead of a single percentage.

Constantin Oesterling has documented the same multi-input screening discipline across guest-posting authority audits through 2021 to 2024. The cross-validated read resolves the resemblance question before a buyer commits.

ICANN-accredited transfer preserves the screened profile through ownership change.

SEO Domains operates with ICANN-accredited registrar transfer protocols.

The Mozscape index records link profiles by hostname and propagates the Spam Score inputs to the new owner, and the parallel Majestic index propagates the Trust Flow and Citation Flow that corroborate it.

Buyers acquire the cross-validated profile that the catalogue’s 7-vector inheritance screen verified at inventory ingestion, so a clean Spam Score transfers as inherited authority instead of an unverified number.