What Is Spam Score in SEO?

Spam Score is a Moz metric, from 1 to 100, that estimates the percentage of sites with similar features that Moz found penalized or banned by Google. Moz rates 1-30% as low, 31-60% as medium and 61-100% as high risk. Google doesn't use it, so treat it as a first-pass filter when vetting links, not a verdict.

The usual moment you meet this number: a site offers you a link, you paste its domain into Moz, and it shows a Spam Score of 38%. Is that a site to avoid, or just a small blog? The answer depends on what the number actually measures.

What the number actually measures

Spam Score is a prediction about a group of sites, applied to one site. A score of 30 doesn't mean a site is 30% spam. It means roughly 30% of sites sharing its profile were penalized.

According to Moz's Spam Score help page, the metric comes from a machine learning model that identified 27 common features among millions of banned or penalized sites. Spam Score reports how closely a given site matches that pattern.

The practical consequence: a Spam Score of 40 tells you this site looks like a group where 4 in 10 got penalized. It tells you nothing about whether this particular site is one of the four or one of the six. That's a useful triage signal and a poor basis for a final decision. Moz says the same thing: because the score is based on correlation rather than causation, it recommends using it to start an investigation, not to disavow by number.

The old version vs the current one

If an SEO article talks about "flags out of 17," it's describing the original version of the metric, not the one Moz shows today.

	Original Spam Score	Current Spam Score
Scale	0-17	1-100 (a percentage)
What it counted	Number of "spam flags" triggered	Modelled % of similar sites penalized
Reading it	"7 out of 17 flags"	"27% of similar sites were penalized"
Features in the model	17 flags	27 features

The old flags (thin content, no contact page, few internal links, an unusual TLD, a very small page count) are still roughly the kind of feature the model learns from. But you can no longer read the score as a flag count, and any guide telling you "4 flags is fine, 8 is bad" is describing a scale that's gone.

What counts as a risky Spam Score

Moz groups scores into three bands, and only the top one is a strong signal on its own:

1-30%: low risk. Most ordinary websites.
31-60%: medium risk. Worth a manual look before you rely on the site.
61-100%: high risk. Treat as a strong signal to walk away.

Two caveats on those bands.

First, small and new sites score high for innocent reasons. A six-page site on a fresh domain with no contact page and few inbound links shares a lot of surface features with disposable spam sites. That's the model doing what it was built to do, and it's why a new, legitimate blog can show a Spam Score in the 30s or 40s while being perfectly clean. Moz itself says that if you haven't had a penalty, a low or medium score on your own site isn't a cause for concern.

Second, Google does not use Spam Score. It's a Moz product built from Moz's index. Google has no access to it and doesn't rank by it. The same goes for Semrush's Toxicity Score, a different metric calculated a different way, and the two often disagree about the same site.

How Spam Score compares to the other risk metrics

Each vendor's risk score comes from its own crawl and its own model, so none of them is authoritative on its own.

Metric	Vendor	Scale	Measures
Spam Score	Moz	1-100	Modelled likelihood of penalty, site-level
Toxicity Score	Semrush	0-100	Weighted count of toxic markers on a linking page
Trust Flow / Citation Flow gap	Majestic	0-100 each	Trusted-neighbourhood links vs raw link volume

The sensible way to use them is together. When Spam Score is high, Toxicity is high, and Trust Flow sits far below Citation Flow, three separate models built on three separate crawls have reached the same conclusion. That agreement is worth far more than any single number.

Using it to vet a link: the part that matters

Spam Score earns its keep as a first-pass filter on a list of prospects, not as the answer.

A workable process when you're evaluating a site you might get a link from:

Sort by Spam Score, don't decide by it. Anything in the high band gets dropped. Everything else moves to step 2.
Open the site. Thirty seconds of looking answers more than the score does. Is there a real author? Is the writing about one subject or forty unrelated ones? Would you send a customer here?
Check its outbound links. Links to gambling, pharmaceuticals or crypto from a supposedly unrelated blog are the classic tell, and the pattern Spam Score picks up most reliably.
Check estimated organic traffic. A high authority score with near-zero traffic is one of the clearest signs of a manufactured profile. See what makes a backlink bad for the full list of warning signs.
Check whether the page your link would sit on is actually indexed. An unindexed page passes nothing at all.

Steps 2 through 5 are what actually decide it. Step 1 just saves you from doing them 400 times.

What to do about a high Spam Score pointing at your site

Usually nothing, because Google's systems are built to ignore most low-quality inbound links rather than punish the site they point to.

Google's own page on the disavow links tool calls it "an advanced feature" to use with caution, and says to disavow only if you have a considerable number of spammy or artificial links and they have caused, or likely will cause, a manual action. It also warns that misusing the tool can hurt your site's performance in Search.

So disavowing is worth considering only in a narrow case: you have a manual action in Search Console citing unnatural links, or you know you paid for links in the past and want to clean up. Outside that, mass-disavowing every domain over an arbitrary Spam Score threshold throws away useful links that happened to score high for structural reasons.

The better use of your time is the input side: getting new links from sites you've actually looked at. Exchanging directly with a real site owner in a related niche helps here, because you can see who you're dealing with before anything goes live. On Backlinkster, both links sit inside real articles and each swap is verified by API rather than taken on trust.

Frequently asked questions

What is a good Spam Score? Under 30% is Moz's low-risk band and covers most ordinary websites. But "good" is relative to the site's size and age. A small, new, legitimate blog can score in the 30s or 40s because it shares structural features with disposable sites, so read a mid-range score as "look closer," not "reject."

Does Google use Moz Spam Score? No. Spam Score is a Moz metric calculated from Moz's own crawl index. Google has no access to it and does not rank by it. The same is true of Semrush's Toxicity Score and Majestic's Trust Flow.

Does a high Spam Score mean my site is penalized? No. A penalty is something you can see: a manual action in Google Search Console, or an unmistakable drop in impressions across the whole site. If neither is present, a high Spam Score is a modelling artefact, not evidence of anything Google has done.

Should I disavow links from high Spam Score domains? Only if you have, or expect, a manual action for unnatural links, which is the condition Google's disavow documentation sets. Google ignores most low-quality links automatically, and bulk-disavowing by score routinely removes legitimate links.

Why did my Spam Score go up without anything changing? Because Moz periodically recalculates the model across its whole index. Scores can shift for every site at once with no change in the underlying websites, the same recalibration effect that moves Domain Authority without any link being gained or lost.

Can I lower my site's Spam Score? Somewhat, mostly by building what makes a site look established: more indexed pages of real content, a proper about and contact page, working internal links and genuine referring domains. There's no setting to change and no quick fix, because the score describes what your site looks like from the outside.

The bottom line

Spam Score is a probability estimate that looks like a verdict. Read it as "how many sites like this one got penalized," use it to cut a long prospect list down to a short one, and then decide by opening the site and looking at it. When it agrees with two other vendors' risk metrics, take it seriously. When it's the only red flag on an otherwise healthy site, it's usually telling you the site is small, not that it's spam.