Your competitor's backlink count is lying to you: how to spot PBN and aged-domain spam

You pull a competitor's backlink report and the headline number lands like a punch: 4,800 referring domains against your 310. If that number goes into a battlecard or a board slide as-is, it will be doing damage, because a large share of most link profiles — yours and theirs — is manufactured noise: private blog networks, expired domains re-animated for link equity, and drive-by comment spam that attaches itself to any site that ranks for anything.

This is the unglamorous problem underneath link-based competitive analysis. Until you can separate real editorial links from the manufactured layer, competitor link counts are not conservative estimates or rough proxies. They are meaningless, and a PMM who quotes them unfiltered is reporting the size of the spam economy, not the strength of a competitor.

The footprint: what a link network looks like from the outside

• Throwaway TLD clusters. Batches of linking domains on cheap extensions, registered within the same narrow window, often with near-identical naming patterns. One is coincidence; forty registered the same month are a network.
• Templated titles. Page titles stamped from a fill-in-the-blanks formula — the same tactic-plus-audience-plus-year construction repeated across dozens of "different" sites. Real editorial coverage does not arrive in matching uniforms.
• Identical filler pages. Click through and the articles are interchangeable: same structure, same approximate word count, no named author with any history, stock phrasing that reads generated. The page exists to host a link, and nothing about it pretends otherwise for more than a paragraph.
• Abnormal link velocity. Hundreds of new referring domains inside a few weeks, then silence. Genuine link growth follows launches, coverage, and content that spreads — it is lumpy but explainable. A vertical spike with no corresponding public event is a purchase order, not a news cycle.

A manual triage checklist you can run today

You do not need enterprise tooling to do a first-pass filter. Export the competitor's referring domains, sort by first-seen date, and work through four questions: do registration dates cluster suspiciously; do the site titles share a template; does a sample of ten linking pages read like a human wrote each one for a reason; and does the velocity curve match anything real, like a launch a marketplace or developer tools company would plausibly have announced? Anything failing two or more checks goes in the noise bucket.

What the filtered profile is actually for

Whatever survives triage is the real asset: the publications, communities, and partner sites that chose to link. That filtered list is worth studying closely, because it maps where a competitor's genuine distribution comes from — which newsletters cover them, which integration partners promote them, which practitioner communities cite their content. It is usually a tenth the size of the headline number, and the filtered comparison often flips the story entirely: a competitor with 4,800 raw domains and 200 real ones is in a weaker distribution position than one with 600 raw and 350 real. That reversal is exactly the kind of finding worth putting in front of your team.

Before the next battlecard update, run the triage on your top two competitors and see how much of their headline count evaporates. And if you would rather start from a structured read of their public competitive surface than from a raw export, top-founders.com/analyze will run a free analysis on any competitor set you give it.