No, not yet, and probably not in the way the headlines suggest. The best data we have says almost everyone using ChatGPT still uses Google. What’s changing is how often they search, which questions they take elsewhere, and how often they click through to a website once they’ve got an answer. For a business, that last change matters more than who wins the platform war.
I get asked some version of “is Google finished?” at nearly every business event I speak at. So here’s what the research actually shows, where it’s thin, and what I’d do about it.
Are people really using both Google and AI?
Yes. Similarweb data for May 2026 shows 95 per cent of ChatGPT users worldwide also use Google. People aren’t switching. They’re adding a tool and using whichever one suits the question in front of them.
That matches what we see in practice. A new client told us recently they’d found us through a mate’s recommendation, checked us on Google, then asked ChatGPT whether we were any good. Both platforms, used in sequence, to make one decision.
If people still use Google, what’s actually changing?
How much they use it, and for what.
A Bocconi University study tracked US desktop households from late 2024 to mid-2025 and compared those with access to ChatGPT Search against those without. Households with access ran 9.4 per cent fewer traditional searches, and the gap grew over time, reaching 17 per cent after 20 weeks. Google’s reach barely moved: about half of households used it either way. Same audience, fewer searches.
The drop wasn’t even across topics. It was steepest for informational questions. Referrals to academic sites fell by about a third, reference sites by a quarter, and developer and news sites by smaller but still real amounts. People took their “explain this to me” questions to AI first.
Not every study points the same way. Research by Gholami and colleagues, using data from over 4,000 desktop users, found people who started using a chatbot consistently visited more websites, while their traditional search activity rose at first and then declined. The authors describe their search estimates as noisy. Adoption seems to go through phases, which is a good reason to be wary of any single snapshot.
What happens to clicks when Google answers the question itself?
They fall, and by a fair bit. In a seven-day experiment published in August 2026, researchers recruited 1,100 US Chrome users and routed a randomly assigned group’s searches into Google’s AI Mode. Click-throughs to external websites dropped by 18.8 percentage points. People ran fewer search sessions each day, though each session ran a little longer. About one in seven participants said they found it harder to get to the specific website they wanted.
Google tells a different story from its side. It reports record query volumes, 2.5 billion monthly users of AI Overviews and a billion for AI Mode in its first year, and says people who use its AI features search more. That may well be true. But Google hasn’t released the baseline or user-level data that would let anyone check it, so it’s a claim rather than a finding.
Your customers still use Google. More of them are getting the answer without ever visiting your website.
What can’t the research tell us yet?
Quite a lot, and it’s worth being straight about that before anyone rebuilds a marketing plan around one statistic.
The studies measure different things. Audience overlap, query volume, click-through rates and referral traffic each capture a different slice of behaviour, so they can all be accurate and still seem to disagree. A stable audience doesn’t mean stable engagement, and more queries doesn’t mean more clicks.
The overlap figure is weaker evidence than it looks. That 95 per cent of ChatGPT users also use Google tells you the audiences overlap. It doesn’t prove AI is only a supplement. Habit keeps plenty of people on Google even as they move particular kinds of questions elsewhere.
Cause and effect is hard to pin down. Search behaviour shifts for lots of reasons, and separating AI’s effect from everything else is difficult even in well-designed studies.
Nobody is tracking individuals across platforms over time. We can’t yet see how one person moves between Google, ChatGPT and a website to finish a real task, which is the thing a business most wants to know.
Much of the data is American, and some of it is from the platforms themselves. Australian behaviour is likely to be similar, but it hasn’t been measured the same way here.
And the ground is moving. Google is testing ads inside AI Mode, which could change both how satisfied people are and how often they click. The next round of earnings reports and peer-reviewed studies should make the picture clearer.
What should a business do while the picture settles?
Plan for both, because your customers are using both. In practice that means four things.
- Keep the Google fundamentals strong. Google’s audience hasn’t shrunk, and the AI systems lean on the same signals: clear pages, accurate structured data and consistent business details.
- Make your pages worth quoting. If informational questions are moving to AI first, the businesses that get named are the ones whose content answers those questions plainly. That’s the core of the GEO side of the work.
- Stop judging content on clicks alone. A page that gets you named in an AI answer can do its job without a visit. Watch enquiries, phone calls and branded searches as well as sessions. We covered what you can and can’t measure in AI search in more detail.
- Check where you stand. Ask ChatGPT, Perplexity and Google’s AI Mode the questions your customers ask, and see who gets named. It takes about half an hour, and our step-by-step guide walks you through it.
None of that depends on predicting which platform wins. It works whichever way the numbers go.
Frequently asked questions
Are people leaving Google for ChatGPT?
Mostly not. Similarweb data shows 95 per cent of ChatGPT users also use Google. What’s changing is how people use each one: some informational questions are moving to AI tools first, and people who get an AI answer click through to websites less often.
Is AI reducing Google searches?
For some people and some topics, yes. A Bocconi University study found US households with access to ChatGPT Search ran 9.4 per cent fewer traditional searches, rising to 17 per cent after 20 weeks, with the biggest drops for academic, reference and news questions. Google’s overall audience stayed about the same size.
Does Google’s AI Mode reduce clicks to websites?
The evidence says it does. In one experiment with 1,100 US users, AI Mode cut click-throughs to external websites by 18.8 percentage points. Google says its AI features increase search use overall, but it hasn’t released data that would let independent researchers verify that.
What should my business do about AI search?
Plan for both Google and AI tools, because your customers use both. Keep your Google fundamentals strong, make your content clear enough to be quoted in AI answers, measure enquiries and calls as well as clicks, and check regularly whether AI tools name your business for the questions your customers ask.
Want to know where you stand in both?
Our AI readiness review checks where your business appears across Google, ChatGPT, Perplexity and Google’s AI Mode, compares you with the businesses being named, and sets out what it would take to close the gap.
Or get in touch with the team. For the wider picture, our AI search optimisation guide is a good place to start.
Figures in this article are drawn from Similarweb’s 2026 audience data, a Bocconi University study of ChatGPT Search adoption, research by Gholami et al. (2026), an AI Mode field experiment by Wang et al. (August 2026) and Google’s May 2026 statements, as reported by Search Engine Journal.
By Brendan Brooks, Founder and Managing Director of HyperWeb. Brendan has spent 25 years watching search evolve and helping Australian businesses stay visible through every shift.
