stripe.comgithub.iocss-tricks.comlobste.rsalgolia.comweb.dev

toxic backlinks and the disavow file - the unpopular, accurate version

toxic is a vendor word, not a google one. what the score actually is, the narrow case where disavow is correct, and why the default answer is to do nothing.

pete the seo wizard
September 10, 2026 · 7 min read · 1,550 words
sharexlinkedin

most people arriving at this question have opened a tool, seen a red number next to some inbound links, and felt a spike of alarm. so the useful thing to do is take the pressure off first: "toxic" is a vendor word. google does not publish a toxicity score, does not use one, and has said for years that it ignores the overwhelming majority of junk links automatically. the red number is one company's guess, and the correct response to it is usually nothing at all.

the decision, in one line

open Search Console and look under Security and Manual Actions. if it is empty, you almost certainly have nothing to do. if it shows a link-related manual action, that is the narrow case where a disavow file is genuinely the right tool.

what a toxicity score actually is

every tool selling one has built a model that guesses which links might be harmful, typically by looking for the signatures of link networks, scraped directories, pages with hundreds of unrelated outbound links, and sites in unrelated languages or subjects.

those signals are reasonable heuristics. the problem is what happens next: the guess is rendered as a confident number in a red cell, and a heuristic presented as a verdict invites action that the underlying confidence does not support. the tell is that the same link routinely scores dangerous in one tool and clean in another, because each vendor runs a different model over a different crawl. if toxicity were a property of the link rather than of the model, they would agree.

why search engines ignore most of it anyway

there is a structural reason, and it is worth understanding because it explains the whole situation rather than asking you to take a position on trust.

you do not control who links to you. anybody can point anything at your domain, at essentially no cost. if bad inbound links reliably damaged rankings, then damaging a competitor would be cheap, easy, and constant, and search results would be shaped by whoever spent the most on sabotage. search engines have an overwhelming incentive to make ignoring such links the default, and google has said repeatedly that this is what it does.

that is why the honest default is inaction. not optimism, arithmetic: the alternative would be an unworkable system.

the narrow case where it is real

inbound links genuinely cause trouble in two situations, and they are both specific enough to check:

  • a manual action. a human reviewer at google has looked at the site and applied a penalty. this appears in Search Console explicitly. it is not something you infer from a ranking drop or a tool's score.
  • links you built. a paid link campaign, a private blog network, an agency that bought links on your behalf, whether or not you knew. the distinguishing feature is that the links wereconstructed rather than merely received.

notice that neither is detected by scanning a backlink report for red cells. both are things you either see in Search Console or already know.

how to use disavow, if you must

if you are in one of those two situations, the disavow file is the right tool and it should be used narrowly:

  1. try to get the links removed first. a removed link is better than a disavowed one, and it is evidence of effort if you are filing a reconsideration request.
  2. disavow at domain level for a network. if a whole network is involved, listing individual URLs is busywork.
  3. disavow only what you can name a reason for. "a tool coloured it red" is not a reason. "we paid for this link" is.
  4. keep the file. it is permanent until you change it. entries do not expire, and a disavow file assembled in a panic three years ago is still suppressing those links today.
the most common real damage

in practice the harm we see from this topic is not caused by bad links. it is caused by people bulk-disavowing hundreds of perfectly good domains because a third-party score flagged them, and quietly throwing away real signals in the process. the tool is not dangerous because it is weak. it is dangerous because it works.

what to do with the time

auditing inbound links for toxicity is, for nearly every site, effort spent on a problem that does not exist. the same hours spent on the opposite question compound instead: which credible domains link to your competitors and not to you.

that list is finite, specific, and every entry is an action rather than an anxiety. it is a link gap, and unlike a toxicity score it does not depend on trusting anyone else's model.

faq

Toxic backlink is a vendor term, not a Google one. Each SEO tool applies its own model to guess which inbound links might harm a site, typically flagging things like link networks, scraped directories, and pages stuffed with unrelated outbound links. Because every vendor uses a different model over a different crawl, the same link is routinely scored dangerous by one tool and fine by another.

Usually not. Google has said for years that it ignores the overwhelming majority of low-quality inbound links automatically, precisely because a site has no control over who links to it and penalising that would make sabotage trivial. The situation where inbound links genuinely cause harm is narrow: a manual action in Search Console, or a history of links you or someone working for you actually built.

Almost certainly not, unless you have a manual action in Search Console or you know links were built for the site in violation of the guidelines. Google's own guidance is that most sites should never need the disavow tool. It is a power tool that removes signals permanently, and the common failure is disavowing perfectly good links because a third-party score coloured them red.

How do I know if I have a manual action?

Search Console tells you directly, under Security and Manual Actions. If that section is empty you do not have one, and inbound-link cleanup is almost certainly solving a problem you do not have. If it does show a link-related action, that is the one situation where a disavow file is clearly the right response, alongside removing whatever links you can get removed.

What about negative SEO?

Deliberate attacks by pointing bad links at a competitor are far rarer than the industry implies, and the reason is structural: if they worked reliably, they would be cheap and constant, so search engines have strong incentive to make ignoring such links the default. If you suspect it, the first step is still to check Search Console for a manual action rather than to start disavowing.

ahrefs · backlinkslocked
upgrade required · $129/mo
crawlgraph · live $99 once
G
github.io92
C
css-tricks.com88
L
lobste.rs86
A
algolia.com84
W
web.dev80
same data · one-time
$99$129/moonce
unlock the data →
stripe checkout · instant access
guides#disavow#toxic backlinks#penalties#guides
sharexlinkedin
pete the seo wizard
author

writes the queries we run internally. ships one tactical post a week.

the dispatch
one email a month.

plus one when a new common crawl release lands. that is all.