The link risks worth worrying about

There are two conversations about backlink risk, and they get confused constantly.

The first is about penalties, meaning a manual action from a human reviewer at Google, visible in Search Console, that tanks your visibility until you fix it and file a reconsideration request. These are rare and usually the result of something egregious.

The second is about devaluation, meaning Google simply ignoring links it does not trust. This is not rare. It is the default outcome for most low-quality links, and it costs you money silently.

Most of what people call "toxic backlinks" is the second thing. Your links are not hurting you, they are doing nothing, which is a different problem with a different fix.

What gets devalued

Links that get ignored share recognisable traits:

• No topical relationship between the linking site and yours
• The host site has no organic traffic, so nobody reads it and nobody would naturally cite from it
• Every outbound link on the site is commercial, with no editorial links at all
• The content around the link is thin or auto-generated
• The site's entire link profile is reciprocal, linking to everyone who links to it

None of these will earn you a manual action. They will earn you a €10 charge for nothing.

The practical response is not to disavow them. It is to stop buying them, and to raise the floor on what you will accept. A traffic threshold alone removes most of this category before purchase.

The patterns that actually attract scrutiny

Manual actions come from patterns, not individual links. The ones that matter:

Anchor over-optimisation. A profile where 40% of anchors are the same commercial keyword is the clearest signal there is. Natural profiles are dominated by branded, generic and long-tail anchors, because that is how people actually write.

Velocity spikes. Forty new referring domains in a month for a site that averaged three is visible from orbit. Growth should look like growth, which is the whole argument for pacing a budget across the year.

Network concentration. Twenty links across twenty domains that share a hosting range, a WHOIS registrar, a WordPress theme, and an outbound-link pattern is one link's worth of diversity wearing twenty costumes.

Single-target concentration. Every link you have ever built pointing at one commercial URL, while the rest of the site has none. Real sites accumulate links across many pages.

Site-wide and footer links. A link in the footer of every page of a host site is both a large quantity signal and an obviously purchased placement.

Notice that four of the five are about distribution, not about any individual link. This is the core insight: risk lives in the shape of the profile, not in the quality of any one link.

Auditing your own profile

Once or twice a year, pull your referring domains and check the distribution rather than the individual entries.

• What percentage of anchors are exact-match commercial keywords? (Target: under 10%)
• How many distinct anchor strings do you have relative to link count? (Higher is better)
• What share of links point at your top single URL? (Under 30% is comfortable)
• Is monthly acquisition roughly steady, or spiky?
• How many of your referring domains have zero estimated organic traffic?
• Do any host domains appear more than twice?

If several of these look wrong, the fix is forward-looking. Change what you buy and how you distribute it, and let the profile normalise over the next two quarters. Retroactive cleanup is rarely the efficient move.

When to disavow

Disavowing is overused. Google's own guidance is that it is unnecessary for the vast majority of sites, and it is easy to do more harm than good by disavowing links that were actually helping.

Use it in two situations:

You have a manual action citing unnatural links. Here disavowing is part of the required remediation, alongside actually removing what you can.

You are the target of a negative SEO campaign, where thousands of links appeared from an obviously coordinated source that you did not build. This is much rarer than people believe, but it does happen.

Outside those two cases, leave it alone. Links you bought that turned out to be worthless are already being ignored. Adding them to a disavow file changes nothing except the chance that you accidentally include something valuable.

The honest summary

If you buy links from sites with real traffic, in themes adjacent to yours, with varied anchors, spread across many target pages, at a steady monthly pace, you are not in the risk zone that people worry about.

The realistic downside of a well-run paid campaign is not a penalty. It is that some fraction of your spend produces nothing, and you never find out which fraction because you did not record what you bought or what happened afterwards.