Is AI Killing SEO? What Changed and What Didn’t
No — SEO is not dead, but the searches it used to convert into clicks have changed shape. AI summaries now answer a growing share of queries before anyone reaches a website, which is real and measurable, not a signal that ranking or good content stopped mattering.
What changed is the presence of a second goal alongside ranking: being the source an AI engine cites. What did not change is the technical and content foundation underneath both goals, which Google itself says has not grown any new requirements.
Is SEO dead, or just different?
SEO is not dead. What is dead is the assumption that a good rank position automatically produces a click. In the first four months of 2026, 68.01% of Google searches ended without a click, up from 60.45% in 2024 — a real, measured shift in how people use search, not a sign that ranking has stopped mattering.
The question “is AI killing SEO” usually means something narrower than it sounds: is it still worth the effort if fewer of those searches turn into a visit? The honest answer is that the effort changed shape rather than disappeared. Pages still need to rank. They also now need to be written so an AI summary can lift the answer and name the source, because a growing share of searches resolve inside that summary before a click ever happens.
Treating those as the same problem is the mistake. A page optimized only for rank position is chasing a shrinking share of the query. A page built for both outcomes is playing the game as it actually exists in 2026.
What actually changed in search
Three things changed, and none of them are mysterious. First, AI summaries now sit above the organic results for a meaningful share of queries, answering the question before the searcher scrolls. Second, when one of those summaries appears, people click through far less often — Pew Research Center found users clicked a traditional result 8% of the time when a Google AI summary appeared, versus 15% when it did not, based on browsing data from more than 900 US adults across 68,879 Google searches in March 2025.
Third, being cited inside that summary became its own goal, separate from ranking. A page can be the source an AI answer quotes without the searcher ever seeing a blue link, which is a genuinely new outcome to optimize for — not a replacement for ranking, an addition to it.
What has not changed is that Google still has to find, crawl and understand a page before anything else on this list can happen. AI features sit on top of the same index; they do not replace the need to be in it.
What didn’t change about ranking
Google has said directly that SEO best practices remain relevant for AI Overviews and AI Mode. There are no additional requirements or special optimizations required to appear in them, and no special schema.org structured data is needed. That is a documented statement from Google Search Central, not a guess.
The eligibility bar for being used as a supporting link inside an AI feature is the same bar that has always mattered: the page has to be indexed and eligible to be shown in Search with a snippet. A page that is noindexed, blocked, or too thin to earn a normal snippet was never going to be cited either.
In other words, the fundamentals — crawlability, clear writing, a page that actually answers the query it targets — are not legacy tactics being phased out. They are the same load-bearing wall the newer AI-visibility work is being built on top of.
Does Google still reward SEO best practices?
Yes, by Google’s own account. Its “AI features and your website” guidance states plainly that following long-standing SEO best practices is what gives a site the best chance of appearing across Search generally, including in AI features, and that there is nothing extra to bolt on.
That guidance also explains why AI Overviews do not show up on every query: Google’s systems only generate one when they judge it genuinely additive to the results, which means many searches never trigger one at all. A page that is not being summarized is not necessarily being penalised — the feature simply did not fire for that query.
Google has also documented that AI Overviews and AI Mode can use “query fan-out,” issuing multiple related searches across subtopics to assemble an answer, and that different underlying models can produce different sets of linked sources. A page that only targets one exact phrase is competing for a narrower slice of that fan-out than a page built to answer the topic more completely.
Why zero-click searches don’t mean zero value
A search that ends without a click is not automatically a wasted impression. Brand recall, being named as the source inside a summary, and simply being present at the moment someone is evaluating options all carry value that a click-through rate report does not capture.
That said, a business cannot run on citations alone forever — conversations, forms and calls still have to happen somewhere, and that somewhere is usually still the website. The practical read on the SparkToro and Pew numbers is not “stop building the site,” it is “stop measuring the site’s value only in clicks.”
Google now reports this directly: Search Console’s Generative AI performance report shows how often links to a site appeared in AI Overviews and AI Mode. A page can earn impressions there without a matching click, which is exactly why impressions, citations and enquiries now need to be read together rather than clicks alone.
| Then (rank-only SEO) | Now (rank plus citation) |
|---|---|
| Success measured by click-through rate alone. | Success measured by rank position and whether an AI summary cites the page. |
| Long throat-clearing intros before the answer. | Answer-first paragraphs an AI engine can lift cleanly. |
| One broad keyword per page. | A page written to survive query fan-out across related subtopics. |
| FAQ sections treated as an afterthought. | A real FAQ, written in the question-and-answer shape AI engines can lift. |
Is “AI SEO” a real discipline or a rebrand?
Some of it is a rebrand. A lot of what gets marketed as “AI SEO” is standard technical and content SEO with a new label, sold to businesses worried they are falling behind. Google’s own position — no special requirements, no special schema — supports that read.
What is genuinely new is the separate discipline of writing and structuring content so it survives being extracted and quoted by a generative engine, sometimes called generative engine optimization or GEO. That term comes from an actual academic paper, not a marketing department, and it describes a real and distinct set of concerns from classic rank-focused SEO. We cover what that means and how it differs from SEO in detail at what generative engine optimization actually is, so we will not re-litigate the definition here.
The practical takeaway is to be skeptical of anyone selling “AI SEO” as an entirely new field with entirely new rules, and equally skeptical of anyone insisting nothing has changed. Both are wrong in the same way: they collapse two related but genuinely different jobs into one.
How do you tell if AI is actually hurting your traffic?
Start in Search Console rather than guessing from a general sense of unease. Compare impressions to clicks over time for the queries that matter most — a rising impression count paired with a falling click count on informational queries is the clearest signature of AI summaries absorbing clicks that used to land on the site.
Next, check whether the pages losing clicks are actually eligible to be cited at all: are they indexed, do they carry a clean snippet, are they blocked by nosnippet, data-nosnippet, a restrictive max-snippet setting, or an accidental noindex? A page cannot be quoted in a summary if it cannot be shown with a snippet in the first place, and this is a more common cause of lost visibility than any change in the algorithm itself.
If you run your own diagnostic and the queries losing clicks are the same ones where an AI Overview now appears, that is a structural shift in that query, not a site-specific penalty. Related reading on the mechanics of Google’s AI features specifically, including a fuller diagnostic checklist, is at why AI Overviews aren’t showing your site.
What to actually do about it
Fix the eligibility basics first: make sure the pages that matter are indexed, fast, and not accidentally restricted from showing a snippet. None of the AI-visibility work below matters if a page cannot be shown in a normal search result to begin with.
Then write for extraction, not just for rank: answer the question in the first two or three sentences, use question-shaped headings that match how people actually ask, and build a real FAQ section rather than treating it as decoration. This is the same structure Google confirms it does not require anything special beyond — it simply rewards content that is already clear enough to lift.
Finally, widen coverage of a topic rather than narrowing to one exact-match keyword, since query fan-out means an AI engine may be searching several related angles on your behalf before it assembles an answer. This is the same discipline we build into our own audits — see what an AI marketing audit actually checks — and it applies whether or not a business ever hires anyone to do it.
Verified 23 September 2026 against SparkToro/Similarweb, Pew Research Center and Google Search Central’s AI features documentation
Building on the web since 1999 · Miami and Mexico City · $1B+ in closed real estate sales
Frequently asked questions
Has Google confirmed SEO still matters for AI Overviews?
Yes. Google’s own “AI features and your website” documentation states that existing SEO best practices remain relevant for AI Overviews and AI Mode, and that no special schema markup or additional optimizations are required to appear in them. The eligibility rule is the same one that has always applied: a page has to be indexed and able to show a normal snippet.
Why do my impressions look fine but my clicks are dropping?
This pattern usually means an AI summary is now answering the query before searchers reach the results, which Pew Research Center’s data on click behavior supports directly. The page is still being shown; fewer people are clicking through to it. Check whether an AI Overview now appears for those specific queries before assuming a ranking problem.
Does GEO replace SEO?
No. Generative engine optimization is a related but distinct discipline focused on how content gets extracted and cited by AI engines, while SEO is about being crawled, indexed and ranked in the first place. A page needs both jobs done, not one instead of the other. See our full explainer on what GEO covers for the detailed breakdown.
Should I stop investing in SEO because of falling click-through rates?
No. Falling click-through rates on some queries reflect a shift in where the click happens, not a shift in whether ranking and clear content still produce enquiries. Sites with weak SEO see AI Overviews and organic results alike go to competitors instead; stopping investment does not recover the queries an AI summary already resolves.
What is query fan-out and why does it matter for content?
Google has documented that AI Overviews and AI Mode can issue multiple related searches across subtopics — “query fan-out” — to assemble a fuller answer. A page written narrowly around one exact phrase competes for less of that fan-out than a page that answers the topic’s related questions as well.
Can blocking AI crawlers protect my search rankings?
No. Google says AI features in Search are controlled by the same Googlebot rules as Search itself, so blocking them that way means blocking Search too. Google-Extended is a separate control that limits use in some of Google’s other AI systems; it does not change how your pages rank. To limit what is shown from a page, Google points to nosnippet, max-snippet and noindex instead.
Sources
- Google Search Central — Optimizing your website for generative AI features on Google Search Google’s AI optimization guide, last updated 10 July 2026. Includes the Search Console control and Generative AI performance report.
- SparkToro — In 2026, less than one third of Google searches still send a click Published 9 June 2026. Similarweb clickstream panel, US, January–April 2026.
- Pew Research Center — Google users are less likely to click on links when an AI summary appears in the results Published 22 July 2025. Browsing data from 900+ US adults; 68,879 Google searches in March 2025.
- Google Search Central — AI features and your website Last updated 10 December 2025.
- Aggarwal et al. — GEO: Generative Engine Optimization Submitted 16 November 2023; accepted to KDD 2024.