An SEO disavowed every link to his site, Moz and Ahrefs included, and traffic barely moved. Google's Mueller has said the same: SEOs overestimate links.
Cyrus Shepard, an SEO who runs experiments for a living at Zyppy, disavowed every link listed to his site in both Google Search Console and Ahrefs: over 10,000 links from 1,473 unique domains, including links from Moz, Ahrefs, Semrush, and Backlinko. Not his weakest links. All of them, including the best ones he had. He left the disavow in place for seven weeks and watched his traffic. It didn't drop. If anything, it ticked up slightly, which he attributes to a new post he published around the same time rather than the disavow itself.
That's the closest thing to a controlled experiment for the exact fear a "backlink concentration risk" framing is built to sell: audit your link profile, and if too much of it traces back to a handful of domains, you're one algorithm hiccup or deindexed partner away from a collapse. It's a real number some link-building vendors like to cite (one agency's guide puts the danger zone at more than 30 to 40 percent of your links coming from fewer than 10 domains), but that threshold has no published methodology behind it anywhere it's cited. Shepard's experiment is a real, documented data point in the other direction: not losing a slice of a link profile, losing the entire thing, from a group of sites that would top anyone's "best backlinks" list.
Google's own answer is almost dismissive
In January 2025, an SEO asked John Mueller directly on Bluesky what happens when a large number of high-authority backlinks disappear all at once, using the hypothetical of a major social platform's links vanishing overnight. After a few rounds of back and forth, Mueller's answer was: "SEOs often overestimate links." He also pointed the asker toward Shepard's disavow experiment directly, as the kind of real-world data that backs up his answer.
That's not Google saying links are worthless. It's Google's own team saying the specific fear behind "concentration risk," that losing your best sources fast could tank you, gets overestimated by the people most likely to be selling an audit for it.
Where a single link does matter
None of this means individual links are interchangeable. Mueller has also said, in a separate conversation about whether total link counts matter, that "one really good link from one website out there" can be a genuinely strong relevance signal on its own, the same point covered in this blog's piece on why raw backlink counts don't track with rankings. A single link from a topically relevant, trusted source is doing real work. But "doing real work while it's there" and "leaves a hole that damages your rankings when it's gone" are two different claims, and the evidence lines up behind the first one, not the second.
The distinction matters because it changes what's worth worrying about. A link's value comes from what it signals about your page's relevance and trustworthiness to the sites and topics around it. That signal doesn't evaporate the moment one specific URL stops pointing at you, especially if the underlying reason you earned that link (real coverage, a real product, a real relationship) is still true. What damages rankings is usually something bigger than one link disappearing: a core update reweighting an entire topic area, a manual action, or a genuine decline in the page's own relevance to what people are searching for.
Why the fear still sells
A dashboard number is easy to build a warning around. "38% of your links come from 6 domains" sounds precise and actionable in a way that "your rankings are generally fine" doesn't. It's also worth remembering that link counts reported by tools like Search Console have their own history of undercounting and bugs, so a scary concentration percentage might be measuring a reporting gap as much as a real profile. None of that means diversifying your link sources is bad advice. A profile spread across many independent, relevant domains is a healthier signal of organic interest than one dominated by a handful of sources, for the same reason a single glowing review looks different from a hundred. It's just that the reason to diversify is what a concentrated profile implies about how the links were acquired, not a fragility that snaps the moment one of them goes away.
Where this differs from what's already on this blog
The safe link velocity post covers how fast you should acquire new links without looking manipulated. This is the mirror image: what happens on the way out, when links disappear, and whether that deserves the anxiety a "concentration risk" framing assigns it. The short answer, backed by an actual disavow-everything experiment and Google's own team, is less than the framing implies.
A directory listing doesn't carry this particular anxiety either way: it's one entry among many others in the same directory, never positioned as the one link a page's rankings secretly depend on.
