You can't move for hot takes about AI killing search at the moment. ChatGPT is the new Google. SEO is dead. Your organic traffic is about to be hoovered into a chatbot, never to be seen again.

It's a fun story, and at industry events it goes down a treat. The actual 2026 data is a bit less dramatic, and honestly more interesting.

The short version: AI hasn't replaced commercial search. It's parked itself next to it. And the slightly counter-intuitive result is that British buyers, both consumers and B2B, are now researching more than they ever have. More questions, more tools, more time spent before they decide. The funnel hasn't shrunk. It's stretched.

If you're running an SME and trying to work out what to actually do about all this, that's worth sitting with for a minute.

Google search volume is, awkwardly, still going up

The simplest test of the "AI is eating search" idea is whether anyone is using Google less.

They aren't. Google handled around 5.9 trillion searches in 2026, up 18% on last year. UK market share is 93.35%, more or less where it was twelve months ago. Volume hasn't fallen off a cliff. It's gone up.

ChatGPT, meanwhile, racked up 1.8 billion UK visits in the first eight months of 2025, against 368 million in the same period the year before. Roughly a fivefold jump. Globally it now does about 12% of Google's search volume.

Both numbers are climbing at the same time, which is a bit of a problem for the substitution story. People aren't shifting their queries from one place to another. They're doing more of both. Total information-seeking sessions worldwide grew 26% between the start of 2023 and the end of 2025. The pie got bigger.

AI Overviews are mostly an "explainer" thing

The next sensible question is where AI is actually showing up. If it had really gone after commercial search, you'd expect to see AI Overviews and ChatGPT answers all over the queries that drive revenue, the "buy", "price", brand-name, local-services type stuff.

It hasn't, really. The Semrush data on intent is quite striking once you look at it:

• Informational queries: AI Overview shows up 36% of the time.

• Commercial queries: 8%.

• Transactional queries: 5%.

Within shopping, "best whatever" style queries now get an AI Overview 83% of the time, up from 5% a year ago. But pure transactional searches like "buy", "price", and named products are still only at 13 to 14%.

That gap is the bit most people miss. AI is doing a lot of work at the top of the funnel, where someone is trying to understand a topic. The bottom of the funnel, the moment someone is actually about to spend money, still happens on a fairly traditional search results page. The closer a query gets to a decision, the less AI seems to be in the way.

People are using both, and not picking sides

Here's where it gets useful. The 2026 buyer doesn't really choose AI or Google. They use both, in sequence, and they're quite organised about it.

Recent consumer research found 77% of buyers use AI and traditional search together. Only 4% mostly rely on AI. About 38% use AI specifically for product research, and 30% to compare options before deciding. Just over half use it during early discovery.

Then they double-check. 98% of consumers verify AI recommendations before buying. 45% Google the brand straight after. 78% say reviews are what really moves the needle on trust. Ofcom's qualitative work in the UK landed in roughly the same place: people are fine trusting AI for the easy stuff, but for anything that actually matters, they switch back to a traditional search and have a proper look at the sources.

So the journey now goes a bit like this. Someone asks ChatGPT to compare three accounting tools, gets a tidy shortlist with the reasoning, then opens Google, types the brand names, reads reviews and case studies, has a poke around the website, and then converts. AI did the thinking-out-loud bit. Google did the proof bit. Both got used, and both got more of the buyer's time than the old "Google it and click" pattern ever did.

The funnel hasn't shrunk, it's stretched

What used to be one trip ("Google it, click, decide") has quietly become two. The research bit happens in AI. The decision bit still happens in traditional search. Because AI makes it almost free to compare and reason about more options, people are doing more of both.

There's no real evidence this is making anyone decide faster. If anything, it's the opposite. Agencies are quietly reporting longer consideration windows and more touchpoints per conversion in 2026 than in 2024. The click data backs that up. Sites cited inside AI Overviews see a 35% increase in clicks compared with non-cited top-10 results, and that traffic converts at 14.2% against 2.8% for traditional organic. Branded queries with AI Overviews see an 18% lift in click-through rate. The clicks that come through are warmer, better-researched and closer to a buy.

For UK SMEs, that points to a few practical things, none of them particularly dramatic.

You're now visible, or invisible, in two places rather than one. Your brand needs to show up in the AI shortlist when someone is researching, and on the traditional SERP when they're verifying. Being good at one and weak at the other leaves a fairly obvious gap for a competitor to walk into.

Content has to do two jobs at once. It needs to be quotable enough that an AI tool will cite it during the research phase, and useful enough that a buyer actually converts when they land via Google during the verification phase. Thin SEO copy that exists to fill a keyword bucket fails both. Properly written content with a point of view, named experts and a bit of original data passes both.

And reporting needs a refresh. Looking only at Google rankings and sessions in 2026 misses half the picture. The more useful question is whether your brand is showing up inside AI answers, in what context, and how that traffic behaves when it lands.

The honest takeaway

AI search hasn't replaced traditional search where it actually matters commercially. What it's done is make British buyers do their homework before they spend. More questions, more comparing, more checking, across more tools.

That's good news if you've got something genuinely worth buying, and slightly awkward news if you've been getting by on shallow content and a strong domain. The pie hasn't got smaller. The bar for being picked has just gone up a notch.