A lot of sites are seeing the same thing right now.
They use AI to publish faster, traffic rises, and it looks like the system is working.
Then the decline starts.
The term being used for that pattern is Mount AI (yes another marketing term to remember). One recent review of customers from an AI content vendor found that 57% had tanked and 22% showed the classic spike-then-crash shape.
Lily Ray has been making a similar point: plenty of tactics meant to win visibility in AI search can end up hurting SEO, which is still the foundation for getting cited in AI answers in the first place.
Why This Keeps Happening
The problem isn’t AI itself. The problem is what most teams do with it.
They confuse publishing speed with content quality. They take a tool that’s useful for research, drafting, and scale, then turn it into an autopilot publishing machine.
Pages go live with no editorial, no original insight, no strong point of view, and no reason to exist beyond catching long-tail traffic.
That can work for a while. Especially when a site suddenly covers more queries than it did before.
But short-term coverage is not the same as long-term authority.
Google has been very clear about this. It doesn’t say AI-generated content is bad.
It says using automation to produce large amounts of low-value, unoriginal content is a problem.
The Drop Hurts More Than It Used To
A few years ago, losing rankings mostly meant losing clicks. Now it means more than that.
Lily Ray’s February analysis of sites hit by Google’s January 2026 volatility found that losing organic visibility could also mean losing AI search citations.
That explains that SEO is still part of the discovery AI systems use for grounding. If your organic presence weakens, your visibility in AI answers can weaken with it.
That is the part many marketers are missing.
When rankings slip, brands often lose ground in both traditional search and AI search at the same time.
Even When You Keep Visibility, Clicks Are Suffering
There’s another issue here.
Even if you stay visible, the click value is changing.
Pew found that when Google users saw an AI summary, they clicked a traditional search result in 8% of visits, compared with 15% when no AI summary appeared.
They also clicked a link inside the AI summary in only 1% of visits. Ahrefs later updated its own research and said AI Overviews now correlate with a 58% lower CTR for the top-ranking page.
So if your AI-heavy content strategy is fragile enough to lose rankings, and the remaining rankings are worth fewer clicks anyway, the cost of getting this wrong is much higher than it was before.
What Marketers Should Do
The answer isn’t to stop using AI. It’s to stop using AI without control.
Put an editorial blocker in front of every piece of AI content generated.
If the page doesn’t pass, it shouldn’t go live. That could be first-hand experience, internal data, expert review, screenshots, examples, or stronger positioning.
AI can help you produce content faster. It can’t be the thing that makes the content worth reading.
Audit for durability, not output.
Some pages look successful because they spike early. That’s not the same as being resilient.
Look for content with thin differentiation, weak links, overlapping intent, poor engagement, or no conversion value. Those are usually the pages most exposed when algorithms tighten up.
Measure more than rankings alone.
You need to know which pages still perform in search, which prompts and topics you are showing up for in AI, and whether technical issues are blocking visibility.
You can do that by combining data from multiple tools in one place, but if you want it in one place, tools like Semrush One combine SEO, AI visibility, and auditing in the same workflow.
Beneficial if the goal is to catch problems before the crash becomes obvious.
The Bigger Takeaway
AI can help you climb faster. It can also help you fall faster if all you are doing is producing more pages with less thought behind them.
That’s really what Mount AI is. Not proof that AI content doesn’t work.
Proof that volume without differentiation is easier to produce, to copy, and for Google to devalue.
The marketers who avoid that crash will be the ones using AI to support strategy, not replace it.
If you find value in these emails, please share or subscribe if you haven’t.