Here is a pattern I keep seeing in Search Console on sites I run: rankings hold, impressions climb, and clicks barely move. Nothing is broken. Google is answering the question on the results page, often using your content to do it, and the searcher never needs to visit.
That is what people mean when they say AI is killing SEO. It is not quite right, but it is close enough to deserve a real answer.
The click is what is dying
Search volume is fine. People still search constantly. What changed is how often a search ends in a visit to your site.
- AI answers are now the default. Similarweb found Google's AI Overviews on 43% of US searches in May 2026, up from about 15% a year before. Google says its AI Mode has passed one billion monthly users.
- Most searches end on Google. SparkToro's analysis of Similarweb data found 68% of US Google searches ended without a click in early 2026, up from about 60% in 2024.
- The top spot pays less. Ahrefs found the #1 result loses about 58% of its clicks when an AI Overview appears.
- People stop sooner. In Pew's study of real browsing, users clicked a result on 8% of visits with an AI summary versus 15% without one, and ended their session on the results page more often.
On top of that, many buyers skip Google for the first pass. G2 found 29% of B2B buyers now start research with an AI tool more often than with Google.
Which pages get hit hardest
Not all SEO is hurting equally. The pages losing the most are the ones an AI answer can fully replace:
- Definitions and "what is" explainers
- Generic how-to articles with no original angle
- Listicles that summarize what everyone else already said
- Keyword pages written to rank rather than to help
If a two-sentence AI summary covers everything your page says, the page has no reason to be visited. That content was always fragile. AI just made it obvious.
What still works
Search still sends valuable traffic, and some of it is worth more than before.
- Being the cited source. Seer Interactive found brands cited inside AI Overviews kept roughly twice the organic click-through of brands that were not cited on the same kinds of searches.
- Decision-stage searches. Pricing, comparisons, "X vs Y," "best X for my situation." Buyers still click when they are close to a decision and want detail an AI summary leaves out.
- Things only you have. Original data, real pricing, case results, tools and calculators, opinions with your name on them. An AI answer can cite these, but cannot replace them.
- Brand searches. When people meet your name in AI answers and on LinkedIn, they search for you directly. Those visits convert far better than generic traffic.
How I would rebuild an SEO program now
- Audit for replaceability. Go through your top pages and ask: would a summary make this unnecessary? Merge, deepen or retire the ones where the answer is yes.
- Lead with the answer. Open every important page with a direct, quotable answer, then go deeper than any summary can.
- Invest in original material. One page of real data or real numbers beats ten rewritten explainers.
- Get cited off your site. Reviews, directories, podcasts, communities and trade press are where AI tools look for confirmation.
- Change the scoreboard. Report impressions, AI citations, branded search and self-reported attribution next to traffic. My guide to measuring zero-click marketing has the dashboard.
So is SEO dead?
No. The old version, ranking a page and waiting for clicks, is fading. The new version is about being the answer, wherever the answer is shown. The fundamentals (crawlable, fast, useful, trustworthy pages) matter more than ever, because they are what AI tools draw on.
For the practical checklist, read 10 tips to get found in AI search results. For the B2B view of Google's changes, see Google AI Overviews and B2B, and for the strategy behind it, what zero-click marketing is.
Stop treating traffic as the scoreboard. Rank to be cited, write to be quoted, and measure success by branded search and pipeline.
