Every business owner has heard the numbers by now. The conclusion most of them are drawing is wrong.
The problem is that “clicks” gets treated as one pool of value. It isn’t. A click from someone asking *what causes a slab leak* and a click from someone searching *emergency plumber near me open now* are not the same asset – and they are not disappearing at the same rate.
Here’s what the data actually says.
How much have clicks dropped since AI Overviews launched?
Pew Research tracked the real browsing behavior of users across nearly 69,000 actual Google searches. When an AI summary appeared, people clicked a traditional organic result 8% of the time. Without one, 15%.
Ahrefs analysed 300,000 keywords of Search Console data and measured a 58% drop in click-through rate for the top-ranking page when an AI Overview is present. That figure had worsened from 34.5% earlier in the same year.
SparkToro’s clickstream study puts zero-click searches at roughly 68% of US Google queries in early 2026, up from around 60% two years prior.
All of that is real. None of it tells you whether your revenue is at risk. After your Content is indexed and ranked in Google, you need to track how those keywords rank. All of that is real. None of it tells you whether your revenue is at risk. As SEO Agency SG often points out to clients, a page can climb the rankings and pull more traffic while the terms actually driving sales quietly slip – which is why performance should be measured against revenue and conversions, not just position and clicks.
AI Overviews mostly appear on informational searches
This is the number that matters most, and almost nobody quotes it.
Ahrefs analysed 146 million SERPs and found that 99.9% of keywords triggering an AI Overview are informational in intent.
The rest of the breakdown:
- Commercial intent: 5.5%
- Transactional intent: 1.2%
- Navigational intent: 0.1%
(Keywords can carry multiple intent classifications, which is why those overlap.)
Semrush reached the same conclusion from a different angle. Across a six-month study of more than 600,000 keywords in ten industries, AI Overviews on commercial-intent SERPs grew 71%. On transactional-intent SERPs, they declined 5%.
The pattern is consistent. Google is comfortable answering the research question. It is noticeably reluctant to intercept the moment of purchase.
Do AI Overviews appear on local search results?
Rarely.
The same Ahrefs dataset found AI Overviews appear on just 7.9% of queries classified as local, against a baseline of 20.5% across all keywords.
If your revenue comes from a service area – trades, legal, dental, home services – that is the single most important statistic in this entire conversation.
Are transactional queries safe from AI Overviews long term?
No, and it’s worth being honest about that.
Semrush also tracked transactional AI Overview coverage rising from roughly 2% in October 2024 to nearly 14% a year later. That’s close to a 600% increase.
But two things are true at once. 14% is not 99.9%. And the trend over the most recent six months has been flat to slightly down.
Watch it. Don’t panic about it.
What does this mean if your organic traffic is down?
If sessions are down 30% and enquiries are flat, you have not lost your business.
You have lost the traffic that was reading a blog post and leaving.
That is a reporting problem, not a revenue problem – although it becomes a revenue problem the moment nobody explains the difference to the person approving your marketing budget.
Is being cited in an AI Overview worth anything?
Yes, but less than ranking used to be.
Seer Interactive studied 53 brands, 5.47 million queries and 2.43 billion impressions. The click-through rates:
- Page cited inside an AI Overview: ~2.1%
- Uncited page on the same results page: ~0.9%
- Page on a results page with no AI Overview at all: ~3.3%
There’s a second finding in that data worth noting. CTR on queries *without* AI Overviews rose from 2.8% to 3.8% over roughly a year.
Fewer people click. The ones who do are further along.
Why doesn’t ChatGPT traffic show up properly in GA4?
Because the referrer often doesn’t survive the click.
Mobile apps open links in isolated in-app webviews that pass no referrer and no campaign parameters. Estimates put the miss rate for GA4’s built-in AI channel at somewhere between 35% and 70% of actual AI-driven traffic.
Anything Google’s own AI surfaces sends arrives from google.com – so it gets filed as Organic Search and can’t be separated out at all.
Then there’s the indirect path, now widely called the dark funnel:
- A buyer asks an AI assistant for a shortlist
- They get three names
- They Google one of those brands directly
- They convert
In your analytics, that’s branded organic or direct. In reality it was an AI recommendation you never saw and can’t attribute.
That said, scale and noise are different things. Across 100+ accounts, Safari Digital SEO Agency’s own data shows a stark rise in sessions attributed to LLMs – yet AI referrals still account for an average of just 1.3% of total sessions, ranging from 0.2% to 3.2% depending on vertical.
How should you measure SEO performance now?
Stop using raw session counts as the headline metric. Four things that hold up better:
Segment keywords by intent. Informational and transactional are moving in opposite directions. Averaging them hides everything.
Judge informational content on citations, not sessions. It’s now a brand asset, not a traffic asset.
Defend transactional and local pages hard. That’s where AI Overview pressure is lightest and where the money already is.
Track branded search volume. It’s the best available leading indicator of AI-driven discovery you can’t otherwise see.
Search hasn’t broken – the measurement layer just hasn’t caught up
The businesses that struggle over the next two years will mostly be the ones that mistook a reporting gap for a demand collapse.
Recent Comments