AI Search Just Became the Web’s Fastest-Growing Discovery Channel. Here’s What Brands Need to Know in 2026

Generative AI has gone from novelty to core internet plumbing. Similarweb’s new 2026 report puts hard numbers on it, and they change how any brand should think about getting found online. Here’s our read on what’s changed, what’s shifting right now, and where it’s going.

For a couple of years, “AI search” was easy to file under interesting but early. That’s over. Similarweb says generative AI platforms now pull around 9.5 billion web visits a month worldwide. That’s up 70% in a year. App downloads hit 4.4 billion (up 58%), and monthly unique visitors reached 655 million (up 57%).

Put those numbers in context. AI chatbots grew their unique visitors 57% year on year. Search grew 2%. Social and eCommerce grew 7% each. News actually shrank 5%. AI is now one of the fastest-growing categories on the whole web.

The headline for you is simple. Discovery is no longer tied to one site or one ranking. It’s spreading across assistants, embedded AI, shopping tools and social apps. The brands that fix their measurement and content now will own the next decade of getting found.

Here’s the report broken down into what it means in practice.

What’s changed: the market grew up, and got crowded

AI is no longer a young person’s toy. Two years ago, 61% of Gen AI users were aged 18 to 34. By May 2026 that’s down to 50%. So half of all users are now 35 or older, and the 45 to 54 and 55+ groups are growing fastest. As Miro’s Michael Horrocks notes in the report, when a tool spreads to older users, that’s usually the sign it’s become a real habit rather than a fad.

The competition has shifted just as hard. ChatGPT is still the biggest standalone AI site. But its share of web visits has fallen from over 80% to about 52% by May 2026. Gemini, Claude, Perplexity and DeepSeek have all built real audiences.

The app numbers tell the same story. Over the year to May 2026, ChatGPT monthly active users rose 87%. Claude jumped 349%, Grok 117%, Perplexity 94%, and Meta AI a huge 435%. Not everyone won. Microsoft 365 Copilot fell 31% and DeepSeek slipped 23%.

One number matters more than any market-share chart: 95% of ChatGPT users still use Google too. People aren’t picking one assistant and dropping the rest. They’re testing several at once. AI is stacking onto how people search, not replacing it.

What’s changing now: AI is disappearing into everything

If 2024 and 2025 were about standalone chatbots, 2026 is about embedded AI. Similarweb shows AI getting baked into every major category at once.

• Search: Google is pushing people into AI-native results. AI Overviews now show up in a large and rising share of searches, peaking above 40% in the sample. AI Mode visits have climbed steadily since launch. Google is turning Search itself into a chat surface.
• Commerce: Amazon’s Alexa for Shopping (formerly Rufus) drove roughly a sevenfold rise in product-page views over the year. It’s already sending traffic to brands like CeraVe, Skechers, Samsung and Lego.
• Social: Meta AI’s reported monthly users more than tripled in 18 months, reaching an estimated 1.2 billion by March 2026, through Instagram, Facebook, WhatsApp and Messenger.
• Productivity and browsers: Microsoft 365 Copilot, Google Workspace AI, Perplexity Comet and Opera AI are dropping assistants into the tools people already use.

The point Similarweb draws, and we agree with, is that there’s no single platform to win any more. You have to earn visibility wherever these AI chats happen.

Behaviour inside the tools is splitting too. Claude and ChatGPT lead on proper back-and-forth: longer prompts, more turns per chat. Gemini and Copilot are built for speed: short prompts, quick answers, tasks done. Perplexity sits in the middle as the research tool. Each even pulls a different crowd. ChatGPT skews to everyday consumers. Claude over-indexes with professionals and students, who are 25 times more likely to visit university sites than the Google average. Gemini leans tech.

Here’s a knock-on effect that matters for old-fashioned SEO. The way people search Google is changing because of AI. Average query length has crept up, and it jumped after AI Mode launched. People are bringing chatbot habits back to the search bar. Graphite’s Ethan Smith puts it well in the report: users are finally learning to search in longer, natural language, because LLMs actually reward it.

One to watch: Gen Z is cooling on AI

Not every arrow points up. Similarweb flags an anti-AI mood building among Gen Z. Think graduation-speech backlash, and the young crowd flocking to AI-free tools like DuckDuckGo’s No-AI search. Some of AI’s earliest fans are getting picky about when they actually want it. It’s small today. But it’s worth watching, because it breaks the lazy assumption that younger always means more AI-friendly.

Winning in AI search: visibility is the new ranking

This is the bit that matters most if you run content or affiliate sites. A few findings stand out.

Brand visibility is category-specific, and led by clear winners. In beauty, CeraVe tops the list (indexed at 100), then Ulta (86), Sephora (80), La Roche-Posay (74) and The Ordinary (67). As Growth Memo’s Kevin Indig argues in the report, share of voice, how often your brand gets mentioned, now matters more than citations, because users focus on the brand names in an answer rather than clicking the links.

AI recommendations genuinely move people. Similarweb tracked users who got an AI recommendation, and recommended brands saw two to four times as many follow-up visits as rivals that weren’t named. AI won’t decide every purchase. But it’s shaping the shortlist. SparkToro’s Rand Fishkin compares it to how 20th-century advertisers measured billboard and TV lift through store visits. The influence is real. What has to change is how you measure and attribute it.

AI is citing the live web far more than it used to. The share of ChatGPT answers with web citations rose more than fivefold in under a year, to about 6.8%. And it’s wildly uneven by topic. Travel triggers cited answers 22.6% of the time, retail 13.5%, sports 10.7%. Those are the moments people compare products and want current info. That’s exactly where good content can shape the answer.

There’s no single GEO playbook. The type of source AI cites swings hard by topic. Beauty answers lean on retail and e-commerce sources. Travel leans on reviews and user-generated content. Finance pulls from specialist financial publishers. You have to know what content AI trusts in your category.

GEO is built on SEO, not instead of it. This is the point we’d underline hardest. Amsive’s Lily Ray explains in the report that every big AI search product runs on RAG (retrieval-augmented generation). So your AI visibility depends on your organic visibility. If your rankings slip, models are less likely to pull your content. Her sharper warning: brands that mass-produced AI content to chase AI visibility often wrecked their SEO, and now face a harder climb back in both. Cutting corners on quality fails twice over.

Links and citations do different jobs. Orainti’s Aleyda Solis flags a real gap in the data. About 65% of cited URLs sit two or three folders deep. Those are the evidence pages AI leans on. But 58.8% of AI referral traffic lands on homepages. So treat them separately. Strengthen the deep pages that feed AI answers. Make sure your homepage is a sharp, conversion-ready front door for the clicks that actually arrive.

That homepage point got sharper recently. After ChatGPT’s 7 May 2026 search update, referral traffic jumped, and the share of visits landing on homepages more than doubled, from about 25% to nearly 60%. ChatGPT is increasingly a gateway to brands, not to single pages.

One more pattern. AI referral traffic is growing fastest in smaller verticals. Beauty grew 312.5% and Fashion 278.2%. Marketplaces like Amazon and Temu still hold the biggest raw volume. For now, AI favours big multi-category platforms.

The new bit: AI advertising arrives

The clearest signal of what’s next is money. Paid acquisition is now part of the AI playbook. Gemini ramped its PPC spend the hardest. But Claude got the strongest audience growth on far less spend, which shows budget alone isn’t deciding the winners.

Bigger still: ads inside ChatGPT are becoming normal. By June 2026, around 26% of ChatGPT chats in the US carried ads, up sharply in a couple of months. And these ads behave nothing like search or display. 66.3% show up after the second prompt or later, so targeting is grounded in what you’ve actually said. 65.3% of sessions carry on after an ad appears. Click-through sits around 0.50%. Early US advertisers already span SaaS, productivity, ecommerce and recruitment, including Resume.io, Monday.com, Framer, SAP, Home Depot and VistaPrint.

Similarweb’s Harel Amir frames it as more than a new ad format. It’s a preview of where discovery is going, with ads folded into the user’s journey rather than interrupting it.

Where this is heading

Pull it together, and here’s where the data points.

1. Embedded AI wins on volume. Standalone chatbots will keep growing. But most AI moments will happen inside search, social, shopping and work tools people already use. Optimising only for chatgpt.com misses most of the action.
2. Share of voice becomes a headline metric. Expect “AI brand visibility” and “brand mention share” to sit next to rankings and traffic in your reports. Citations still count. Getting named counts more.
3. SEO and GEO merge. The winners will be sites with real authority, genuinely useful content and clean technical foundations. Same fundamentals, now feeding two discovery layers. The AI-spam shortcut is already backfiring.
4. Measurement has to catch up. As Similarweb’s Baruch Toledano puts it, visibility isn’t about ranking in one place any more. It’s about being findable wherever decisions get made, then tying that back to branded demand, conversions and revenue.
5. AI advertising becomes a real channel. Conversational ads are finding their feet fast. Within a year or two, expect a proper buying market and a fresh scramble for limited in-chat space.

What’s this mean for Jodana Partners

Short version: this is the shift Jodana Partners has been getting ready for, and the data backs the plan.

• Double down on authority, not volume. Lily Ray’s warning is the whole game. Sites that pumped out AI filler are losing both SEO and AI visibility, so quality content on solid technical foundations wins twice.
• Treat GEO and SEO as one job. AI pulls from search rankings via RAG, so the organic work we already do is what feeds the AI answers. We don’t need a separate AI strategy. We need a sharper version of the current one.
• Fix the homepage and the deep pages separately. Deep pages earn the citations that shape AI answers. Homepages catch nearly 60% of the AI clicks. Both need to pull their weight.
• Watch the high-citation verticals. Travel, retail and beauty trigger cited AI answers most often. If a partner sits in those spaces, there’s more room to influence the answer with strong content.
• Get measurement ready now. Brand mentions and share of voice are the new KPIs. The partners who start tracking AI visibility early can prove the value before rivals even notice the channel.