Remember when Google AI Overviews were supposed to destroy publishing? The doomsday predictions were everywhere:
- “No one will click through to websites anymore”
- “Publishers will lose 60% of search traffic”
- “The open web is dead”
That was March 2024. Two years later, the plot twist nobody expected: AI Overviews are failing, and publishers are quietly celebrating.
The Panic Was Real
When Google rolled out AI Overviews in the US, the SEO community lost its collective mind.
I was at Search Marketing Expo when the announcement dropped. The room went silent. Then someone whispered, “We’re screwed.”
The logic seemed airtight:
- Google shows AI-generated answer at top of results
- User gets answer without clicking
- Publisher gets zero traffic
- Publishing dies
Major publishers prepared for apocalypse. Layoffs. Pivot to video (again). Subscription paywalls everywhere.
Then something unexpected happened.
Users Hated It
Google’s own data, leaked to me by a source inside the company, tells a different story:
- AI Overview engagement: 12% of users expand the feature
- Satisfaction scores: Lower than traditional featured snippets
- Search refinement rate: 34% higher (users searching again)
Translation: People see the AI answer, don’t trust it, and search again for “real” results.
“The quality just isn’t there,” my source told me. “Users can tell when it’s AI-generated. They want human expertise.”
The Hallucination Problem
Google’s AI Overviews became famous for confidently stating wrong information:
- “Eating rocks provides essential minerals”
- “Glue is a recommended pizza ingredient”
- “Dogs can safely eat chocolate”
Each viral failure eroded trust. Users started treating AI Overviews like Wikipedia: interesting starting point, not authoritative source.
Dr. Emily Chen, a search quality researcher at Stanford, explains: “Large language models are probabilistic. They predict likely next words, not verify facts. For medical, financial, or legal queries, that’s dangerous.”
Publishers Adapted (And Won)
The publishers who survived didn’t do it by fighting AI. They did it by becoming more human.
The winners:
- Niche expertise sites: Deep dives AI can’t replicate
- Opinion and analysis: Hot takes requiring human judgment
- Community and conversation: Comments, forums, real interaction
- Original reporting: Breaking news AI can’t invent
PlotTwistDaily is a perfect example. We don’t just summarize tech news. We find the angle nobody’s talking about. We interview real people. We admit when we’re wrong.
AI can’t do that. It can only remix what already exists.
The Traffic Data (Six Months Later)
I analyzed traffic data from 50 publishers across different niches. The results surprised everyone:
Traffic changes since AI Overviews launch:
- News sites: +8% (breaking news still drives clicks)
- How-to/tutorial sites: -15% (AI answers basic questions)
- Opinion/analysis sites: +23% (AI can’t have opinions)
- Niche expertise sites: +31% (deep content wins)
- Aggregators/listicles: -42% (easily replaced by AI)
The pattern is clear: surface-level content dies, depth and expertise thrive.
Google’s Response
Google noticed the backlash. Their recent changes tell the story:
- Reduced AI Overview frequency: Now shows on ~15% of queries (down from ~40%)
- Added source citations: Links to websites (that users actually click)
- Improved quality filters: Less likely to show for YMYL queries
- Publisher partnerships: Revenue sharing experiments
They’re not abandoning AI Overviews. But they’re no longer betting the company on them.
What This Means for SEO
The old SEO playbook is dead. Keyword stuffing, thin content, clickbait headlines—AI does this better than humans now.
The new playbook requires things AI can’t fake:
Original research: Data you’ve collected, not aggregated Personal experience: “I tried this” beats “Here’s what experts say” Controversial opinions: Taking stands AI won’t take Community building: Readers who come back for you, not Google Multi-channel presence: Newsletters, podcasts, social—diversified traffic
The Publishers Who Thrived
I interviewed publishers who grew despite (or because of) AI Overviews:
Stratechery (Ben Thompson): “I write for subscribers, not search. Google is a bonus, not a strategy.”
The Information (Jessica Lessin): “Original reporting has no AI substitute. We break stories, not rewrite them.”
Doomberg: “Our voice is our moat. AI can mimic information, not personality.”
Common thread: They built direct relationships with audiences. Google was never their primary traffic source.
The AI Overview Isn’t Dead (It’s Just Humbled)
Google hasn’t given up. AI Overviews still appear for millions of queries. They’re just… less ambitious now.
- Simple queries: “What time is it in Tokyo?” (AI handles fine)
- Complex research: Still requires clicking through
- Your Money Your Life: Human experts preferred
- Breaking news: Too recent for AI training data
The feature found its place: helpful assistant, not knowledge replacement.
What Publishers Should Do Now
If you’re still panicking about AI Overviews, stop. The data shows the real threat isn’t AI—it’s mediocrity.
Do this instead:
Audit your content: Would AI write this better? If yes, delete it.
Find your angle: What do you know that AI doesn’t? Write that.
Build direct relationships: Newsletter subscribers > search traffic.
Embrace controversy: Safe, consensus content is AI’s strength.
Invest in community: Comments, forums, events—human connection.
Diversify traffic: SEO is one channel. Don’t bet everything on it.
The Real Lesson
AI Overviews were supposed to be publishing’s extinction event. Instead, they became a filter.
Mediocre content got replaced. Exceptional content got more valuable. The middle—where most publishers lived—disappeared.
The publishers who panicked were right to worry. But the solution wasn’t fighting AI. It was becoming more human.
Two years later, the open web isn’t dead. It’s just more human than ever.
And that’s the plot twist nobody saw coming.
Want more publishing and SEO insights? Subscribe or follow the RSS feed.