The State of AI Search in 2026

For marketers, this rewrites the job: ranking first for a keyword still has value, but the prize in 2026 is being the source an answer engine trusts enough to quote, which is often not the page sitting at the top of the organic results. The commercial question for 2026 is whether a brand gets cited when Google, ChatGPT, Gemini or Perplexity answers on its behalf, and what that visibility is worth once the click does arrive.

This guide sets out the state of AI search in 2026 for marketers: how the major answer engines work and how far they have scaled, what they are doing to clicks and to commercial performance, how these systems decide which sources to cite, where regulation is heading, and what a practical response looks like.

Key takeaways

• AI answers are now the default search surface. AI Overviews now appear on roughly 48% of tracked Google searches (BrightEdge, February 2026), and Google reported AI Mode at 1 billion monthly users and AI Overviews at 2.5 billion by May 2026.
• The click is getting scarcer. When an AI Overview appears, users click an organic result 8% of the time, against 15% when one does not (Pew Research Center, July 2025), and only 1% click a link inside the summary.
• The prize has moved from ranking to being cited. Around 85% of brand mentions in AI answers come from third-party pages rather than a brand’s own site (Superlines, 2026), and only about 12% of AI-cited URLs rank in Google’s top 10.
• AI referral traffic is small in volume and high in intent. AI-referred retail visitors converted 42% better and generated 37% higher revenue per visit than non-AI traffic in March 2026 (Adobe Digital Insights, via eMarketer).
• The engines have fragmented. ChatGPT’s share of AI web traffic fell from about 76% to about 53% in a year (Similarweb), so a single-assistant strategy now misses most of the audience.
• Regulation has arrived. In June 2026 the UK Competition and Markets Authority became the first regulator to make Google let publishers opt out of AI Overviews without losing normal search ranking.

What is AI search in 2026?

AI search is the practice of finding information through generative engines that read the web and write a single synthesised answer, rather than returning a ranked list of links for the user to click. It covers Google AI Overviews and AI Mode, and the standalone assistants ChatGPT, Gemini, Perplexity and Claude, all of which now answer questions directly and cite a handful of sources beneath the response.

ChatGPT reached 900 million weekly active users by February 2026, up from 400 million a year earlier, according to OpenAI, and Google reported AI Mode at 1 billion monthly users at its I/O event in May 2026. For a large share of everyday questions, the answer engine is now the first thing a searcher reads.

The shift matters commercially because it changes where attention lands. A user who reads a synthesised answer has still encountered information, and often a brand, without ever loading a website, which means visibility now includes being named or quoted inside an answer as well as earning a click. Our guide on how AI systems read and search the web covers the retrieval side in more depth.

How AI search actually works

When someone enters a query, an AI engine interprets the intent behind it, retrieves candidate pages from its index or a live search layer, scores them for relevance and credibility, and writes an answer that quotes a selection of them. This runs as a separate selection pass from traditional ranking, weighted towards source authority, entity clarity and how cleanly a passage can be extracted.

The pages an engine cites are not always the ones ranking first. Roughly 12% of URLs cited by ChatGPT, Perplexity and Copilot rank in Google’s top 10 (2026), which tells you the two systems are making different decisions about what deserves to be shown. A page can be invisible in AI answers while ranking well organically, and the reverse.

Underneath this sits natural language understanding, entity recognition and the relationships between topics, so engines increasingly respond to what a user means rather than the exact words typed. You can read how Google AI Overviews work in more detail in our dedicated guide.

How big has AI search actually got?

AI Overviews appeared on roughly 48% of tracked Google searches by February 2026, a 58% rise year on year, according to BrightEdge, with coverage estimates across different trackers ranging from about 20% to 50% depending on query mix. The feature has expanded well beyond simple informational queries into commercial, transactional and shopping searches.

AI Mode has scaled faster still, growing from 75 million daily users in late November 2025 to 1 billion monthly users by May 2026, after launching in the US in May 2025 and the UK in July 2025. These are self-declared figures rather than independently verified, so they are best read as direction and magnitude rather than precise counts.

Google’s Search revenue grew 19% year on year to $60.4 billion in the first quarter of 2026, so the change in surface has not dented the underlying business. Google is monetising the shift rather than losing to it.

What AI search is doing to clicks

The most rigorous read on click impact comes from Pew Research Center, which tracked 68,879 real Google searches from 900 US adults in March 2025. When an AI Overview was present, users clicked an organic result 8% of the time, against 15% without one, they clicked a link inside the summary just 1% of the time, and they ended the session without any further click in 26% of cases, compared with 16% otherwise.

Independent studies using different methods reach the same conclusion. Ahrefs measured a 58% drop in click-through rate on the top-ranking page for queries showing an AI Overview in February 2026, up from a 34.5% drop the previous April, and Sistrix found position-one click-through in Germany falling from 27% to 11% when a summary appeared.

Datos found the US zero-click rate easing from 24.5% in December 2025 to 22.4% in March 2026, which suggests the sharpest disruption may have passed as coverage stabilises. Informational and quick-answer queries lose the most, while complex and commercial searches still send traffic, and a brand cited inside the answer can gain visibility even as raw clicks fall.

The rise of GEO and AEO

Generative engine optimisation (GEO) and answer engine optimisation (AEO) are the disciplines that have grown up around this shift, both concerned with being the source an AI answer quotes rather than only ranking in a list. GEO leans towards the wider authority signals that decide which source an engine trusts, while AEO leans towards structuring content so a clean, self-contained answer can be lifted from the page.

Conductor found that 97% of digital leaders reported a positive impact from AEO in 2025, and 94% planned to increase GEO investment in 2026. Gartner forecast that traditional search engine volume would fall 25% by 2026, a projection now matched by the direction the behavioural data is taking.

For most brands this is an extension of good SEO rather than a replacement for it, with added weight on clarity, structure and authority. Our AI search optimisation work is built around that overlap, treating organic and AI visibility as one problem rather than two.

How AI engines decide who to cite

Around 85% of brand mentions in AI answers originate on third-party pages rather than owned domains (Superlines, 2026), with a brand’s own site accounting for only 5% to 10% of what engines reference. Earned coverage across credible sites does more for AI visibility than any single change to a homepage.

Brand mentions across the web correlate roughly three times more strongly with AI visibility than backlinks do (Authority Tech, 2026), though that is a correlation rather than proof of cause. Engines cite entities they can recognise and corroborate across multiple independent sources, which makes consistent presence off-page the foundation for citation on-page.

On the page itself, answering the question directly in the first 40 to 80 words gives the engine a clean passage to extract, current content earns citations more readily than stale content, and structured formatting makes claims easier to lift. Schema is more mixed than the advice suggests, with Ahrefs finding that adding JSON-LD produced no statistically significant change in AI citations, so it is worth doing for clarity rather than as a citation lever. We cover the authority side in our guide to becoming a trusted AI source.

The platform split: why one strategy no longer reaches everyone

ChatGPT’s share of worldwide AI web traffic fell from about 76% in June 2025 to about 53% by May 2026, according to Similarweb, with Gemini reaching roughly a quarter and Claude approaching a tenth. A strategy built entirely around ChatGPT now reaches a shrinking fraction of the AI audience, and each engine draws on its own source preferences, with ChatGPT leaning heavily on Wikipedia and Perplexity leaning on Reddit (Profound, 2025).

Market-share figures for AI referrals are where the data gets noisy, with different providers putting ChatGPT’s share anywhere from the low sixties to the low nineties depending on the panel and method. Sources differ enough on the true split that the figure worth trusting is a brand’s own, measured in its own analytics, and ours have doubled and then doubled again over the past 18 months.

The takeaway for planning is to measure AI referrals directly and to spread visibility work across engines rather than betting on one. Brands weighing where to focus can read our view on whether to keep optimising for ChatGPT, and those wanting a consistent measure over time can benchmark their AI visibility with our Search Radar tool so the trend is tracked rather than guessed.

Is AI search traffic actually worth having?

AI referral volume is small next to organic search, and the quality is the reason to care. Adobe Digital Insights, reported by eMarketer and drawn from more than a trillion visits to US retail sites, found that visitors arriving from AI assistants converted 42% better than non-AI traffic in March 2026, spent 48% more time on site, viewed 13% more pages, and generated 37% higher revenue per visit.

Twelve months earlier the same channel converted worse than non-AI traffic, so the quality gap opened inside a single year, and AI-driven traffic to US retail sites rose 393% year on year in the first quarter of 2026 off that low base.

A visitor who arrives after an AI has already weighed the options and made a recommendation comes with intent that is closer to validated, which is why AI referrals tend to matter to revenue out of proportion to their share of sessions, and why measuring them by outcome rather than volume is the sensible frame.

Agentic search: the shift already arriving

At I/O 2026 Google introduced Information Agents inside AI Mode that monitor the web and return synthesised updates, alongside agentic booking and calling features that can arrange appointments or contact businesses on a user’s behalf, expanding from restaurants into home services, beauty and local experiences through 2026. Search is beginning to complete tasks as well as answer questions.

For marketers, this raises the stakes on structured, machine-readable information, because an agent completing a task needs accurate prices, availability and contact details it can act on, not prose it has to interpret. The brands that make their key facts easy for a machine to use will be the ones an agent can transact with, and the direction of travel is worth planning for now.

Regulation is catching up

On 3 June 2026 the Competition and Markets Authority required Google to let publishers opt out of AI Overviews and AI Mode without losing normal search ranking, the first ruling of its kind anywhere, prompted in part by publisher evidence of steep traffic losses.

In its submission to the CMA, DMG Media, publisher of MailOnline and Metro, reported desktop click-through falling from 25.23% to 2.79% on certain queries when an AI Overview appeared, a drop that goes directly to advertising revenue for a business built on reach.

Set against the EU AI Act and the wider push for transparency in AI systems, the CMA ruling reads as the first of more to come. Regulation of how answer engines use and credit third-party content is likely to grow, and marketers relying on organic reach should treat publisher-facing rules as a space to watch, because the terms on which content is used and attributed are still being written.

What this means for marketers: the 2026 AI search playbook

The response that holds up across all of this is to apply durable fundamentals to a new surface, rather than chase features.

• Build for citation as well as ranking
  Answer questions directly in the opening lines, structure content so an engine can extract it, and treat being quoted as a distinct goal alongside earning a click.
• Earn third-party authority
  With most AI mentions coming from earned media, coverage, mentions and consistent entity signals across credible sites do more for visibility than any single on-page change.
• Keep content current
  Fresh, maintained pages are cited more readily than stale ones, so a regular update cycle on key pages is an active visibility tactic.
• Measure your own AI referrals
  Track AI traffic by source and by outcome in your own analytics, since published market-share figures conflict and your data is the only reliable read on your position.
• Go multi-engine
  Spread visibility work across Google’s AI surfaces, ChatGPT, Gemini and Perplexity, because each cites differently and no single one reaches the whole audience.
• Keep people in the loop
  AI can scale content and analysis, but judgement, accuracy and brand voice still need human oversight, a point we make in full in our piece on keeping people in the loop.

Frequently asked questions

What is generative engine optimisation?

Generative engine optimisation (GEO) is the practice of shaping content and off-site authority so AI answer engines cite a brand when they respond to a query. It focuses on being the quoted source inside answers from Google AI Overviews, ChatGPT, Gemini and Perplexity, rather than only ranking in a list of links. GEO overlaps with SEO on quality and structure, and adds emphasis on entity clarity, third-party mentions and content an engine can lift cleanly.

What is answer engine optimisation (AEO)?

Answer engine optimisation (AEO) is the practice of formatting content so a search or AI system can extract a direct answer from it. It prioritises clear, self-contained answers of roughly 40 to 80 words, structured headings, and data an engine can quote with confidence. AEO and GEO are often used interchangeably; where they differ, AEO leans towards on-page answer structure and GEO towards the wider authority signals that decide which source an engine trusts.

How does AI search work?

AI search works by interpreting a query, retrieving relevant sources, and generating a written answer that cites a selection of them, rather than returning a ranked list of links. The engine runs a separate selection pass from traditional ranking, weighing source credibility, entity clarity and how easily a passage can be extracted. As a result, the pages an AI cites are not always the ones ranking first in organic results.

How do I appear in Google AI Overviews?

Appearing in Google AI Overviews depends on being a clear, credible source that Google’s systems can extract and trust. Answer the question directly in the opening lines, structure content with clean headings and lists, keep it current, and build recognition for the brand across authoritative third-party sites. Pages cited in AI Overviews are often not the top organic result, so clarity and authority matter more than position alone.

How do I rank on ChatGPT?

Ranking on ChatGPT means being cited when it answers a query, which depends heavily on presence across trusted third-party sources rather than a brand’s own pages. ChatGPT leans on high-authority references and earned media, so coverage, mentions and consistent entity signals across the web carry more weight than on-page tweaks. Clear, answer-first content and up-to-date pages improve the odds of extraction.

Will AI Overviews reduce my website traffic?

AI Overviews reduce clicks for many informational queries, though the effect depends on the query and the intent behind it. Pew Research Center found users clicked an organic result 8% of the time when an AI Overview appeared, against 15% without one. Simple quick-answer searches lose the most, while complex and commercial queries still send clicks, and being cited inside the answer can lift both visibility and referral quality.

Does AI search replace SEO?

AI search does not replace SEO; it changes the goal from ranking first to being cited as a trusted source. The underlying signals still matter, including quality content, clear structure, technical health and authority, but they now feed answer engines as well as the traditional results page. Brands that publish clear, credible, well-structured content remain best placed, whether the destination is a blue link or an AI-generated answer.

The bottom line

AI search in 2026 rewards the same fundamentals that have always mattered, applied to a surface that now writes the answer for the user. Visibility comes from being clear enough to extract, credible enough to cite, and present enough across the wider web that an engine trusts the brand in the first place, while the click has become scarcer and more valuable, which pushes measurement and authority to the top of the list for the year ahead.

Roar helps brands build and measure that visibility across Google AI Overviews, ChatGPT, Gemini and Perplexity, treating organic and AI search as one commercial problem. Learn about AI SEO solutions and where your brand stands today.