If your WordPress site has slipped down a popularity list, a top-million count will not tell you why.
On November 1, WordPress co-founder Matt Mullenweg posted a BuiltWith chart on X and wrote, "I wonder if the Google AI Mode effect @Reddit just mentioned dropped 15-20% of @WordPress sites from the top million." It is a question, and the post gives no count, date range or list of domains.
A top-million ranking cannot separate a site that left WordPress from a site that lost rank. My read is that the AI Mode link could be right, but this post does not contain the test that would show it.
The idea traces back to Reddit's earnings call on October 29. Reddit CEO Steve Huffman told investors that Google's AI Mode changes where search traffic goes. The Verge reported his words: "The reality is, AI Mode does not drive traffic. Not for us, and not to anybody, as far as I know."
Reddit's own numbers gave him something to point at. The company's shareholder letter, as quoted by The Verge, said October search traffic "dropped by more than 1 million referrals on average per week." That is one very large site describing its own referrals. It tells us nothing directly about WordPress.
Mullenweg picked it up at 1:26 PM ET on November 1. He said the "Entire Internet" trend on BuiltWith, a technology-lookup service, looks very different from W3Techs, "However, when you look at just the top 1M, the drop is similar." Then came the question about 15 to 20 percent of WordPress sites leaving the top million. On the chart as posted, the top-million line is too flat to read a change of that size.
W3Techs surveys the whole web. Its sample includes well over 20 million sites. eSEOspace, citing W3Techs yearly data, puts WordPress at 43.0 percent of all websites on January 1, 2026 and 40.2 percent in October. A Review42 analysis of W3Techs numbers found that sites with no detectable CMS rose from 28.6 percent to 30.4 percent over the same seven months that WordPress fell 2.0 points. That is the same W3Techs data, not an independent panel of domains.
It does raise the possibility that some sites dropped out of the WordPress count because the CMS stopped being detectable.
The top-million count has its own problem, and it sits in the ranking. W3Techs says it uses Google's Chrome User Experience Report plus a customized version of the Tranco list for its ranking reports. Tranco says it averages ranks across several providers' lists over the past 30 days. Neither description is a measure of Google search clicks. A rank change is at best an indirect signal of a traffic change.
A top-million list is also a fixed-size ranking. A WordPress site can drop out because it lost rank, because another site rose past it, or because its owner changed platforms. Those are three different stories, and a single percentage cannot tell them apart. Shane Rounce made the same point in a reply, noting that a changing sample can make a platform shift look bigger than it is: "Those are very different stories." His account is small, and I cite him as an illustration only.
Other people replied with two views on AI Mode. Hiten Shah took Mullenweg's side, writing "I think that's probably what's going on." He added no data. Nat Miletic wrote that he is "bullish on WP in the AI era" because "It's open source, and AI models are very comfortable with it." That is a view on WordPress's future, and it does not bear on whether past rank changes came from AI Mode.
The test that would help is easy to describe. Take the Tranco top million from January and the current list. Pull a fixed sample of domains detected as WordPress in January, and record whether each still runs WordPress and where it ranks now. Sort them into left WordPress, stayed but fell out, and stayed in the list. Then compare organic traffic for the stayed-but-fell-out group. If that group lost search traffic and the others did not, the AI Mode story gets real support.
I would not call the AI Mode explanation wrong. A content site that loses Google referrals plausibly slides down popularity lists, and Reddit just described what that looks like at scale. But popularity rank is a slow, blended signal, and it moves for reasons that have nothing to do with search. My guess is that the drop mixes all three stories, and the evidence in this post cannot say how much each one counts.
Free newsletter
The Sherpa Brief
One email a week on what changed in AI search, what it means, and what to do about it. Written for marketers, SEOs, and publishers.
Free. Unsubscribe in one click.
