Opinions on disavowing run from always to never. A decision tree built on counts is less exciting and more useful, and it takes three lookups to fill in.
LinkBuildingStat research team ·
What Google has said, and what it has not
Google has long said most sites do not need a disavow file. Its systems ignore much link spam automatically, which was part of what Penguin 4 changed in September 2016 and what the December 2022 link spam update reinforced.
Still, the tool exists for manual-action cases and for a pattern you created yourself. Competitors dropping thousands of junk links on your site are mostly neutralised, though nobody outside Google can prove it for your own domain.
So the useful question is not whether disavowing is allowed, but whether you have a pattern big enough to justify the effort.
The decision tree
01
Did Google send a manual action about links?
Check Search Console. If yes, follow its instructions, and a disavow file may be part of that. If no, go on.
02
Did you or a past vendor buy or swap links at scale?
If you can name a date and a vendor, you own the pattern. Count the affected domains, as in step three.
03
Count the share
Pull refdomains (2 credits) and anchors (2 credits). If a small share, such as a few percent, looks bad, leave them. If the bulk of anchors is one money phrase, read on.
04
If nothing above applies, do nothing and log it
Write the date and the counts. Revisit after the next monthly refresh.
Hypothetical counts and what we would do
Invented profiles, not real sites.
Profile
Counts
Reading
A
300 referring domains, 6 odd ones
Leave alone, log it
B
300 referring domains, 90 from one expired-blog network, 70 percent exact-match anchors
Pattern you can name: consider a disavow file
C
80 domains, spike of 40 junk links after a competitor feud
Watch two refreshes before acting
What to write down either way
The date, the counts, the decision and the reason. Four fields.
If a later update such as the August 2025 spam update coincides with a dip, you will want to know what you decided in advance and why. Memory edits the story; a dated sheet does not.
A disavow file is easy to write and hard to undo in your head, because you will never learn which links it removed from the calculation. That asymmetry is the best argument for restraint.
A word on negative SEO
Rare, and mostly noise. Google's systems, including SpamBrain, discard a great deal of link spam, so a competitor's junk links rarely do what the sender hopes.
Still, a sudden rise worth looking at will appear in our overview as a jump in referring domains, and in anchors as a pile of odd phrases. Log the date. Run the benchmark again a month later to see whether it persisted.
A short reading of the three profiles
Profile A is boring, and boring is good. Six odd domains out of 300 is two percent, a level that nearly every real site shows at some point, and a disavow file there would take effort and carry a small risk.
Profile B is different. Ninety of 300 domains from one expired-blog network is 30 percent, and the anchors lean one way. That combination looks like a past decision of yours, and it is the kind of case in which a file is worth considering.
Profile C splits opinion. A sudden pile after a quarrel feels urgent, but a monthly index cannot show you the first week of it, so wait for two refreshes before acting. Odd, but true. Then compare the counts with the post-mortem method.
A limit
Our index will not tell you which links Google counts. We can show how many, from where and with which anchor. The call on a file is yours, and a mistaken disavow can drop links that were helping.