The word toxic came from tool vendors, not from Google, and it is worth being careful with it. Most links people panic about are simply ignored. A small number genuinely cause damage, and telling the two apart decides whether you spend a weekend on a disavow file or leave it alone.
What actually makes a link a problem?
Two conditions together: the link exists to manipulate rankings rather than to inform a reader, and it is part of a pattern rather than an isolated case. One scraper copying your page is not a pattern. Four hundred forum profiles built in a fortnight is.
The categories that matter, in the order we find them causing real trouble:
- Paid do-follow links at volume Links that pass ranking signals in exchange for money, unmarked. This is the category Google's link spam policy is actually about.
- Private blog networks Sites that exist only to link out. Shared hosting, shared templates, shared authors, and a footprint that gets the whole network flagged at once.
- Site-wide footer and sidebar links One relationship producing ten thousand links. It is trivially recognised for what it is.
- Exact match anchors far outside a natural spread When a third of your inbound anchors are one commercial phrase, no classifier needs to work hard.
- Links from hacked pages Injected into a site whose owner does not know. They vanish when the owner finds them, and until then they sit on pages Google already distrusts.
- Automated placement Comment, profile and directory spam. It is cheap because it does nothing.
What can you safely ignore?
Almost everything else. Scrapers, low quality directories you never submitted to, foreign language spam sites, links from pages with a bad score in somebody's tool. Google has spent twenty years learning to discount links it does not trust, and discounting is the normal outcome, not a penalty.
The clearest way to think about it: a link you did not build, on a site nobody visits, that Google can recognise as spam, is a link Google throws away. It costs you nothing. Spending a week cataloguing those is a week not spent on the thing that would have moved the needle.
Should you use the disavow file?
Only in two situations. First, you have a manual action for unnatural links in Search Console. Second, you or an agency you paid actually built manipulative links and you want them discounted before anyone looks. Outside those two, leave it alone.
The disavow tool is a loaded weapon pointed at your own link profile. Every domain you list stops passing value, permanently, and there is no report telling you what you gave up. We have seen more traffic lost to enthusiastic disavowing than to the links being disavowed.
If you do need it, three rules. Disavow at the domain level rather than the URL level, because the URL level is a game of whack-a-mole. Keep one file and treat it as cumulative, since uploading a new file replaces the old one entirely. And write a short comment above each block explaining why it is there, because the person reading it in two years will be you.
How do you tell a manual action from an algorithmic drop?
A manual action appears in Search Console under Security and Manual Actions, with a description and a date. If nothing is listed there, you do not have one, whatever a consultant tells you. Algorithmic drops are silent and are usually about the content, not the links.
| Manual action | Algorithmic drop | |
|---|---|---|
| How you find out | A notice in Search Console, with a reason. | Traffic falls. Nothing is announced. |
| Shape of the fall | Sharp, often to almost nothing, on a specific date. | Gradual, or stepped across a core update window. |
| What fixes it | Clean up, disavow what you built, file a reconsideration request. | Improve the pages. There is nothing to file. |
| Recovery time | Days to weeks after the request is accepted. | Usually the next update, which may be months away. |
A reconsideration request that says the links were nothing to do with us, without a cleanup, gets rejected. The reviewer wants to see what you removed, what you disavowed, and what you changed so it will not happen again.
Is negative SEO real?
It is real and it is rare. Pointing spam links at a competitor mostly does nothing, because the same systems that discount links you built also discount links somebody else built at you. It is worth checking when a drop is sudden and unexplained, and it is worth not assuming.
If you find a genuine attack, the response is the disavow file at domain level and a note in your own records with dates. We clean these up for clients. We do not run them, and we decline the work when that is what is being asked for.