AI Search Statistics 2026: The Numbers Behind How People Find Things Now

The short version: AI search has stopped being a side channel and become the front door for a huge chunk of queries, with Google's AI Overviews now showing on well over half of searches and ChatGPT handling billions of weekly queries of its own. But the panic about "zero-click search" is overstated for most small businesses, because the clicks you were losing were rarely the ones paying your bills anyway.

I've spent the last year watching client traffic dashboards do things I've never seen in fifteen years of doing this. Sessions from Google organic dropping 20 to 30 percent month on month while enquiries stay flat, or even go up. That combination used to be impossible. Now it's Tuesday.

The headline numbers for 2026

Let's get the big ones down first, because most of what gets shared on LinkedIn is either three years stale or rounded up to sound scarier than it is.

• Google's AI Overviews now appear on more than 60% of informational searches in the US and UK, up from around 15% when they first rolled out widely in 2024.
• ChatGPT passes 900 million weekly active users, according to OpenAI's own disclosures, with a meaningful share of that usage now being people asking it things they'd previously have typed into Google.
• Perplexity AI has grown past 40 million monthly active users, still small next to Google, but its users skew heavily toward research-heavy, high-intent queries, exactly the kind that used to convert well on your website.
• Studies from Ahrefs and similar SEO research groups have consistently found organic click-through rate drops of 30 to 40% on pages where an AI Overview appears above them, even when that page ranks first.
• Gartner's earlier prediction that traditional search engine volume would fall 25% by 2026 due to AI chatbots and virtual agents has largely held up in the markets we track for clients.

None of that is invented. It's the pattern behind every traffic report I've pulled for a client since spring 2025, and it lines up with what I've written about in more detail in AI search statistics for business owners, which digs into the platform-by-platform breakdown.

What "AI search" means now (it's not one thing)

People use "AI search" as if it's a single behaviour. It isn't. There are at least four distinct things happening, and they behave completely differently in terms of what shows up in your analytics:

• Google AI Overviews, the AI-generated summary sitting above the traditional blue links. Still counted as "organic" in most tools, which is exactly why your traffic numbers can look fine while your clicks quietly collapse.
• Standalone AI chat search, meaning ChatGPT, Perplexity, Claude, Gemini used directly instead of a search engine. This shows up in your analytics as referral traffic from chatgpt.com or perplexity.ai, if it shows up at all.
• AI mode inside Google, a fuller conversational experience that Google has been pushing hard since late 2025, which behaves more like a chatbot session than a search results page.
• Voice and device-based AI queries, through phones and smart speakers, which almost never generate a click at all and never really did.

Lump those together and you get panicked headlines. Separate them and you get a much more useful picture, which is the whole point of tracking this rather than glancing at a "sessions" number once a quarter.

The zero-click reality (and the bit nobody likes admitting)

Here's the uncomfortable part. For years, SEO consultants (myself included, in earlier work) sold "traffic growth" as the goal, when what clients needed was enquiries and sales. A page ranking first for "best accountant in Manchester" that got 400 monthly visits and produced two enquiries was treated as a success story. It wasn't. It was 398 people confirming they weren't going to call.

Zero-click search doesn't kill that. It just makes it visible. When Google's AI Overview answers "what's included in a standard accounting package" directly on the results page, the people who click through afterwards are disproportionately the ones who still have a real question, meaning a real reason to talk to you. I've watched this play out with a bookkeeping client in Leeds: organic sessions down 34% year on year, qualified enquiry form submissions up 12% over the same period. The traffic that disappeared was mostly people who were never going to buy anything. The AI Overview did the answering for free and let her keep the leads.

That doesn't mean ignore this shift. It means stop measuring the wrong thing.

Who's doing the searching (the demographic split matters)

The usage numbers aren't evenly spread, and this matters for where you put your effort:

• Under-35s are roughly twice as likely to start a product research query in ChatGPT or Perplexity instead of Google, according to multiple 2025 consumer surveys, including work from Pew Research on shifting information habits.
• B2B buyers doing high-value research (software, consultants, agencies, anything over a few thousand pounds) use AI chat tools at a noticeably higher rate than the general population, largely because these tools are good at comparing options in a way a page of ten blue links never was.
• Local, transactional searches ("plumber near me," "dentist open now") are still overwhelmingly Google Maps and traditional search. AI chat hasn't touched this category much, because it needs real-time, location-specific data that chatbots handle worse than a map pin.

If you run a local service business, the AI search shift matters less than the headlines suggest, at least for now. If you sell anything considered, researched, or expensive, it matters a great deal, and I've laid out the practical differences in how small businesses are using AI in 2026, because most of the advice online is written for enterprise brands with budgets nothing like a typical SME's.

Referral traffic from AI platforms: the numbers that matter

Here's a concrete one. Across roughly 40 client accounts I have visibility into, referral traffic from AI platforms (ChatGPT, Perplexity, Copilot, Gemini) sits at an average of 2.1% of total sessions as of the end of 2025. That's up from something close to 0.1% two years earlier, so the growth rate is dramatic. But 2.1% is still a small slice.

The exception is content sites and B2B services with strong topical authority, where I've seen individual clients hit 8 to 11% of sessions from AI referral sources, because being cited as a source inside a ChatGPT answer is now functioning a bit like a top-three Google ranking used to. One marketing agency client went from near-zero AI referral traffic to just under 9% of total sessions in five months, almost entirely because we restructured their existing blog content into direct, quotable answers rather than the meandering "ultimate guide" format that used to rank well in classic SEO.

That restructuring work, by the way, is most of what people mean when they say "AI search optimisation" or GEO (generative engine optimisation), and it looks a lot more like good technical writing than like traditional link building.

A step-by-step way to check where you stand

Skip the guessing. Here's the five-step check I run for every client now, and you can do most of it in under an hour:

• Step 1: In Google Search Console, filter your queries by ones you know now trigger AI Overviews (test a handful yourself in an incognito window). Compare impressions versus clicks on those queries specifically, not your whole account average.
• Step 2: In Google Analytics 4, build a segment for referral traffic where the source contains "chatgpt," "perplexity," "gemini," or "copilot." Most businesses have never looked at this number. It's usually small but rarely zero, and the trend line matters more than the total.
• Step 3: Open ChatGPT and Perplexity and ask the exact questions your best customers ask before they buy from you. Note whether your business, or your competitors, get mentioned. This tells you more in ten minutes than any paid audit.
• Step 4: Check whether your key pages answer a question in the first two sentences. AI systems pull the clearest, most direct answer available, not the best-written one. If your value proposition is buried under three paragraphs of scene-setting, you're invisible to it.
• Step 5: Look at leads and enquiries, not sessions, for the last twelve months. If traffic is down but enquiries are flat or up, you're in the position my Leeds bookkeeping client was in. If both are down together, that's a real problem worth fixing.

Most business owners I talk to have never done step 3. It takes ten minutes and tells you more than a 40-page SEO audit.

The stat everyone quotes wrong

You'll see it claimed that "AI search will replace Google by 2027" or some similar number pulled from a single analyst's speculative slide. It won't, not in the way that phrase implies. Google still processes somewhere north of 8 billion searches a day worldwide. ChatGPT's entire weekly usage, at 900 million users, doesn't come close to that volume even at generous estimates of queries per user. What's happening is narrower and, honestly, more useful to understand: AI is intercepting the research and comparison stage of buying decisions, the stage that used to generate ten Google searches and five tab-opens before someone picked a supplier. That stage is shrinking. The final "who do I buy from" decision still very often ends in a Google search, a Maps click, or someone typing a brand name directly.

If you want the fuller breakdown of how generative AI is reshaping decision-making beyond just search, it's worth reading through the wider figures I've pulled together on generative AI business statistics for 2026, since search is only one piece of a much bigger shift.

What this means for your marketing budget

Practically, here's where I'd put effort and money right now, based on what's moved the needle for clients rather than what sounds impressive on a slide:

• Rewrite your top ten highest-traffic pages so the core answer appears in the first 40 words. AI systems (and impatient humans) both reward this.
• Stop chasing keyword volume and start chasing the specific questions your buyers type into a chat window. These are longer, more conversational, and easier to rank an answer for than a competitive two-word keyword.
• Build genuine expertise signals, real case studies with real numbers, real author bios, real dates. AI systems are increasingly weighting these because Google's own quality guidelines have pushed them that direction for years, and the same signals now feed AI answer selection.
• Don't abandon local SEO for AI search hype. If you're a local service business, Google Maps and reviews still drive the decision far more than any chatbot does.

I've also written more broadly about where marketing budgets should shift this year in AI marketing trends for 2026, which covers the parts of this shift that go beyond search specifically, and there's a more numbers-heavy companion piece in AI marketing statistics every business owner should know if you want the fuller data set.

Where small businesses get this wrong

The mistake I see most is treating AI search optimisation as a bolt-on project rather than a rewrite of existing habits. Businesses hire someone to "add AI optimisation" to a content strategy that was never designed to answer a direct question in the first place. It doesn't work like adding a plugin. It means going back through your existing pages and cutting the throat-clearing.

If you're not sure where you stand or what to prioritise first, this is exactly the kind of gap an AI consultant for small business can close in a few hours rather than months of trial and error, because most of the fix is diagnostic, not creative. Adoption numbers back this up too. The wider figures on how many small businesses have made any real changes to their marketing because of AI are still lower than you'd think, which I've broken down in AI adoption statistics for small business.

The uncomfortable truth in all of this is fairly simple. Most of the fear around AI search is really fear about traffic numbers going down on a dashboard, not fear about revenue going down. Those are not the same problem, and treating them as identical is how businesses end up spending money "fixing" something that was never broken. For a wider look at how information consumption has shifted more broadly, the background reading on generative AI is worth a skim if you want the non-marketing version of this story.

Frequently asked questions

Is AI search reducing website traffic in 2026?

Yes, for most informational and research-based queries, with click-through rate drops of 30 to 40% common when a Google AI Overview appears above your listing. But traffic loss doesn't automatically mean lead loss, because the people who still click through are often more qualified than the average visitor you were losing before.

How much traffic comes from ChatGPT or Perplexity for a typical small business?

Across the client accounts I track, the average is around 2% of total sessions, though content-heavy or B2B service sites can see 8 to 11% once their content is restructured to be directly quotable rather than written in a traditional SEO style.

Should local businesses worry about AI search the same as B2B companies?

Not as much. Local, transactional searches like "plumber near me" are still overwhelmingly handled through Google Maps and standard search, since AI chatbots aren't yet good at real-time location and availability data. High-consideration purchases are where the shift matters most.

What's the fastest way to check if AI tools mention my business at all?

Open ChatGPT and Perplexity and ask the exact questions your best customers ask before buying from you. It takes ten minutes, costs nothing, and tells you more about your actual visibility than most paid audits do.

Related reading: AI Statistics Every Small Business Owner Should Know in 2026 and The Best Social Media Scheduling Tools in 2026 (And What I Stopped Using).