Before you cite a backlink statistic, trace it to the original study. Five steps, with a worked example of a dofollow concentration claim we could not find.
One claim reads that the top 1% of all websites own 78% of all dofollow backlinks. It is a tidy line, and it appears on link-building statistics pages with a tag naming a big data provider in brackets. It is also the kind of line that ends up in a pitch deck, a client report or a blog post, usually without anyone opening the study behind it.
This post is a method for what to do when a number like that arrives with a famous name attached and no way back to the source. We use this claim as the worked example, and the post is not a verdict on the figure.
Step 1: find the first link in the chain
Open the page that carries the statistic and look at the bracket or footnote. Is it a link to the study, or a name? A name is a claim about where the number came from. A link is a way to check. If there is only a name, search the named company's own site for the exact figure and the exact words.
In our example, the bracket named a data provider's web index and had no link. We searched that provider's current statistics article, which lists its backlink figures one by one with a link under each. We read the full page, last updated June 2, 2026, and found nothing resembling "top 1%" or a concentration share of dofollow links. That is the correct sentence to write: we could not find it there. Not that it was never published, which no one can show.
Step 2: see what a traceable statistic looks like
Compare it with a figure from the same page that does resolve. The page says 66.5% of links to sites over the last nine years are dead, and links to the study. That study states a sample of 2,062,173 websites, a start date of January 2013, a collection method, an author and an update date of February 2, 2024.
A traceable statistic gives you five things: who counted, what they counted, how many, over what period, and when it was last updated. The concentration claim gives you none of them. The gap between those two examples is the point.
Step 3: check that the number means what the sentence says
Even a real figure can be bent in a retelling. "66.5% of links to sites in the last nine years are dead" is about links found since 2013 that no longer resolve. It is not a statement that two thirds of the web's links are broken today. Before you reuse a figure, restate it in the study's own scope, then compare that with how your sentence uses it.
Step 4: look for the pattern of drift
We have seen the same shape before: a number attributed to a well-known tool, repeated across roundups in the same words, never resolving to the source. Our earlier looks at a 427-day backlink wait figure and a content-length backlink claim followed it. Roundups tend to copy each other, so ten pages with the same wording can trace back to one page.
Step 5: decide, and write the honest version
You have three options once the check is done. If you found the study, cite it directly and use its scope. If you found a related real figure, use that and say what it measures. If you found nothing, either drop the number or write reported on several statistics pages, source not located and move on. The third option is the one people skip, and it is the one that keeps you from repeating a figure you cannot defend.
A short routine you can reuse
Keep a log with one line per statistic you plan to use: the claim, the first link you found, the page where you read the number, the sample size, the date and a yes or no on whether you opened the original. It takes a minute per figure. When a client or editor asks where a number came from, you have the answer on hand, and when a source updates or vanishes you know which posts to revisit.
Everything above applies to numbers about backlinks, link rot, rankings and AI visibility alike. The tools that sell those numbers are not the only ones reporting them, and a name in brackets is not a source.
