Three years ago the formula was simple: build a site → rank it → collect traffic and send it to offers. By autumn 2026 that formula is officially dead. Google AI Mode and AI Overviews have reached nearly every commercially interesting query, and for the first time in search history most sessions end before they begin — without a single click on any website. It is the biggest traffic redistribution since mobile search. Here is the full breakdown: what happened, who lost how much, who managed to grow, and exactly what to do.
Timeline: how we got here
- 2023–2024. Google launches SGE, then AI Overviews — answer blocks above the results. The industry worries, but traffic drops are localized.
- 2025. AI Overviews roll out to hundreds of millions of queries. AI Mode appears — a full chat instead of a SERP. Publishers record their first double-digit organic declines.
- 2026. AI Mode becomes a familiar search tab; AI answers cover informational and, increasingly, commercial queries. SEO agencies estimate zero-click share on informational queries at 80–90%+. Major media publicly report search traffic down 30–60% year over year.
What exactly broke
The mechanics are simple and merciless. A user asks a question — Google used to show 10 links, and someone got the click. Now the AI reads those same 10 sites for the user and retells the essence right in the SERP — and nobody gets the click. Your site is still "ranking", the position reports look great, and the traffic is gone: you have become a free content supplier for someone else''s neural network.
"Same rankings as a year ago. Impressions up. Clicks — minus 47%. We are working for Google for free" — a typical complaint from a content site owner in SEO chats this autumn.
Who got hit hardest
- Article farms and SEO doorways. Sites built on long-tail informational queries ("how to choose X", "what is Y") — AI retells their content wholesale. Drops of up to 70%.
- Review sites and "top-10" listicles. The classic of white-hat affiliate — casino, service and VPN rankings. AI assembles its own ranking from several sources; links go to two or three "sources" out of dozens.
- News sites without a brand. The story is retold in the AI answer — why click through?
- Coupon and price aggregators. The promo code shows right in the SERP.
Who survived and grows
The most interesting part — not everyone fell. Three categories feel better in 2026 than before the AI era:
- Forums and UGC platforms. Google deliberately mixes live discussions into results and into AI answers themselves: the algorithm needs real human experience that cannot be generated. Reddit is the main beneficiary of the last two years of updates — and the same mechanics work for niche forums.
- Brands. A "site_name + topic" query is not intercepted by an AI block — the user already knows where they are going. Brand traffic became the most protected asset in SEO.
- Sources the AI cites. AI answers contain links. There are few, and clicks are rarer than a classic top-3 — but it is the new "position zero", and a separate battle for it is underway (GEO — generative engine optimization).
The survival plan: 6 steps
- Run a vulnerability audit. Split your queries into informational (die first), commercial (dying now) and transactional/brand (protected). Measure the share of traffic at risk.
- Shift content toward transactional intent. "Buy", "download", "deposit promo code", "sign up" — here the user needs the site, not a summary.
- Build a community layer. A forum, comments, Q&A, live case studies — UGC is currently the only content type Google promotes rather than cuts. This is literally why forums (like ours) are the right bet for 2026+.
- Optimize for AI citation (GEO). A direct answer in the first paragraph, clear definitions, numbers and tables, FAQ blocks, Article/FAQPage schema. The goal is to be the source the AI answer links to.
- Convert one-time traffic into your own base. Every search visitor must leave a trace: a Telegram subscription, an email, a forum account. They may never come from Google again.
- Diversify before it hurts. Push, Telegram, YouTube/Shorts, paid traffic — in 2026 SEO cannot be the only leg your project stands on.
FAQ
Is starting SEO in 2026 worth it at all?
Yes, but not as an "article farm for informational queries". What works: brand projects, community platforms, transactional pages and GEO optimization for citations.
What about other search engines?
The trend is universal: every major engine has AI answers; only rollout speed differs. Assume zero-click becomes the norm everywhere within 1–2 years.
Can AI answers be "hacked" to get cited?
"Hacked" — no. Systematically improving your odds — yes: structure, facts, freshness, brand mentions on third-party platforms. That is GEO, a new discipline worth learning right now.