Two things happened recently in the SEO world that look unrelated until you put them side by side.
Kevin Indig published a study of 5.32 million search results asking whether Google has demoted listicles. Separately, reports started coming in that the August spam update went after mass-generated AI content.
One study about a content format, one update about a writing tool. Except both of them point at the same thing, and it isn’t the thing most business owners have been warned about.
What Google’s spam updates have targeted this year
There was an unannounced update in mid-January that hit sites built on self-serving listicles. Some SaaS brands lost between 29 and 49% of their organic visibility within weeks of it. Then came the August spam update, and the reports around it suggest a chunk of it was aimed at content produced in bulk by AI. Google also published research on a system it calls the Scalable Cluster Termination System, designed to shut down networks of mass-generated spam.
Kevin’s read on the January update is that the aggressive use of AI to churn out poor listicles created a flood, and Google went after the flood. His words for what Google seems to be measuring are quality and velocity.
Velocity, meaning how fast you’re producing it, not what you produced it with.
Listicles didn’t die and neither did AI content
If Google had decided listicles were spam, they’d have all vanished into the ether in the press of a button. They haven’t gone anywhere.
At least one listicle still appears in the top 10 for 55.1% of the queries Kevin looked at, and in the top three for 32.3%. Comparing a fixed subset of queries between January and August, listicles at position one fell from around 19% to around 15%, and across the top three from roughly 37% to 32%. That’s a few percentage points. Let’s call it a trim, not a cull, shall we?
Also, if you need more convincing, since January 2026, 62% of listicles written by brands about their own products gained estimated organic traffic, and the median page gained 38%. So the format is fine. Plenty of the pages that were supposedly penalised are doing better than they were.
What the August spam update reports suggest
This part comes with a health warning, because the evidence is anecdotal. It’s tweets and forum posts rather than a dataset, and Search Engine Journal says as much in its own write-up (linked above).
The pattern people are describing is that sites running fully automated posting got filtered out, while sites that had been operated by hand for a while before switching to automation often survived. One person running a publication that uses AI-generated content reported no penalty, and mentioned that every article gets checked by a human before it goes live.
The most useful line in the whole piece came from someone pointing out that the question isn’t whether generative AI made the content. It’s what the content was made for.
Why publishing at volume is the thing that gets caught
Put the two stories together and they describe the same behaviour, caught four months apart.
Content produced faster than anybody could reasonably check it, aimed at ranking rather than at a person reading it. In January that looked like a pile of “10 best” pages. In August it looked like automated posting. Different shapes, same underlying practice, and Google’s response was much the same both times.
The B2B software numbers in Kevin’s data show what piling in at volume looks like from the outside. Brand or vendor listicles appear in the top 10 on 46.2% of B2B software search results pages. In food, it’s 3.3%. In travel, 6.2%. That gap has nothing to do with how people search for software and everything to do with an entire sector reaching for the same tactic at the same time, mostly in the hope of getting cited in AI answers.
Which is roughly what I was getting at when I wrote about the difference between using AI to help you write and using AI to publish at scale. The tool was never the problem. The rate (or velocity) at which you did it was the issue.
What this means for your B2B website
If you’re using AI to help you write, and a person reads it, edits it, cuts the bits that sound like nobody, and adds something only your business knows, you aren’t what any of this was built to catch. Your production rate is still human.
If you’ve got a plan to publish 40 pages this month, you’re closer to the target, and it makes very little difference how those pages got written. A team of freelancers producing keyword-shaped pages at that speed would have had the same problem in 2018.
The distinction Google seems to be drawing is about whether anybody was involved. Not which software was open when it happened.
Ranking isn’t the prize it used to be anyway
There’s a bigger issue sitting underneath all this, and Kevin’s data catches it nicely.
Even the listicles that came through January intact are competing for a click that’s getting harder to win. 92% of the search results pages containing a top-10 listicle also had Reddit or YouTube somewhere on the page. For listicle-related B2B queries, AI Overviews appeared on 93.4% of them.
So a page can hold its position and lose the traffic anyway, which is a point I’ve made before – winning a volume game gets you a spot on a page where the click has already gone somewhere else.
Two caveats worth having in your head before you quote any of these numbers at anyone. Kevin’s study is US English, desktop only, one crawl wave over 47 hours in August, and it’s observational, so it shows associations rather than causes. He also classified the results with his own classifier, which he’s open about. And the spam update reporting, as I said, is people comparing notes rather than anybody’s data.
I’ve been saying some version of this for a couple of years now, usually to people who’ve just been sold a content package with a monthly page count attached to it. It’s quite nice to have 5.32 million rows of somebody else’s research pointing the same way.
If you’ve got a content plan you’re not sure about, or you’re using AI and are now worrying you’ve done something wrong, that’s a sensible thing to get a second opinion on before Google gives you one. I’m quite opinionated (you might have noticed) so why not grab an hour with me and we’ll see what’s what?
