TL;DR
- SEO is not dead in 2026.
- Google reported year over year query growth, and independent clickstream data shows Google search grew 21.64% in 2024, roughly 373 times ChatGPT's volume.
- Clicks are the casualty: 68% of US searches ended without one in early 2026, and AI Overviews correlate with a 58% lower CTR for the top result.
- The work now spans rankings, AI citations, and share of voice across engines.
No, SEO is not dead in 2026. People run more Google searches than ever, and organic search still sends far more visits to websites than every AI engine combined. The part that died is the guaranteed click, because AI answers now resolve many queries right on the results page.
Both sides of the argument hold real numbers
We hear this question in almost every intake call. Two camps argue past each other, and each one holds a true half of the picture. The half the doomsayers own is clicks. The half the optimists own is demand.
Demand first. In Google’s Q2 2025 earnings remarks, Sundar Pichai reported that overall queries and commercial queries on Search continued to grow year over year, that AI Overviews passed 2 billion monthly users, and that AI Overviews drive over 10% more queries globally for the types of queries that show them. Independent panels back this up. Datos clickstream data published by SparkToro shows Google searches grew 21.64% in 2024, and that Google still receives roughly 373 times more searches than ChatGPT.
Is SEO dead? No. Search demand keeps growing: Google reports rising query volume, and its AI features generate extra searches. Click supply is shrinking: more than two thirds of US searches now end without a click, and AI Overviews cut clickthrough on affected queries. SEO in 2026 means earning rankings, citations inside AI answers, and measurable visibility across engines.
So the pie got bigger while the slices got thinner. For the wider picture across engines, adoption curves, and what it means for budgets, see our current read on where AI search stands.
What the “SEO is dead” crowd gets right
The click decline is real and well measured. A July 2025 Pew Research Center analysis of real browsing behavior found that people who saw an AI summary clicked a traditional result in just 8% of visits, versus 15% when no summary appeared. Links inside the summary did even worse. Users clicked one in only 1% of visits.
Pew also measured what people do next. Visits that showed an AI summary ended the browsing session entirely 26% of the time, compared with 16% on pages with only traditional results. The summary satisfies the query and the session simply stops. For a business, that means the impression happened, the influence happened, and analytics recorded nothing.
Ahrefs saw the same pattern from the website side, and it has got worse. Its April 2025 study put the gap at 34.5%. Re-running the same 300,000-keyword method on December 2025 data, the presence of an AI Overview now correlates with a 58% lower clickthrough rate for the top-ranking page. The damage falls off further down the page but never disappears: 50.8% at position two, 19.4% at position ten. We break down the forecasting consequences in our analysis of what AI answers do to organic clicks.
Zero-click search was already the majority case before the current wave, and it keeps climbing. Using Similarweb’s cross-device panel, SparkToro found that 68.01% of US Google searches ended without a click across the first four months of 2026, against 60.45% in 2024 and roughly 45% a decade ago. Put differently, fewer than one search in three now sends anyone to the open web. And the AI layer keeps spreading. Semrush tracked AI Overviews through 2025: triggered on 6.49% of queries in January, peaking near 25% in July, then settling at 15.69% in November. Ahrefs, working from 146 million SERPs, puts the rate a little higher at 21% of all keywords.
So when a founder shows us a traffic chart that fell off a cliff, we believe the chart. The response is a strategy question, and we wrote a separate playbook on getting value from searches that never produce a click.
What the doomsayers get wrong
Dead things shrink. Search is growing. The same earnings remarks that spooked marketers reported rising query volume, with AI features adding searches on top. A channel with growing demand and 2 billion people on its newest feature is many things. Dead is a strange word for it.
Scale is the second miss. ChatGPT is a real channel, and we track it for clients every week. It is still small next to Google. The 373 to 1 search ratio above comes from clickstream panels, and every independent estimate we have read lands in the same order of magnitude. Dropping organic work because “everyone searches in ChatGPT now” means firing your biggest channel to please your smallest one.
The third miss is treating AI visibility as a separate universe with separate physics. AI engines lean heavily on content that already ranks, and the overlap between classic rankings and AI citations is large. That overlap is why we treat them as two halves of the same discipline.
The last miss is overstating coverage. The trackers above land AI Overviews somewhere between 16% and 21% of queries, so the large majority of searches still return a classic results page. They also cluster on low-value terms: nearly 60% of the keywords that trigger one draw 100 or fewer monthly searches. That is not a permanent state, though. Informational queries made up 91.3% of triggers in January 2025 and only 57.1% by October, while commercial, transactional and navigational triggers all climbed. The searches that pay your bills still look familiar today, and they are the direction of travel.
What actually died
Specific tactics died, and some deserved it. Publishing a thin page for every keyword variant stopped working once AI answers began compressing ten variants into one response. Fixed clickthrough assumptions for position one died with the Ahrefs numbers above. Traffic as the single reported outcome died when most searches stopped producing clicks.
| Old assumption | 2026 reality | What to do now |
|---|---|---|
| Rank number one and the click follows | An AI Overview correlates with a 58% lower CTR for the top result on affected queries | Track impressions, citations, and branded demand next to clicks |
| A page for every keyword variant wins the long tail | AI answers compress many variants into a single response | Build fewer, deeper pages that resolve the full question |
| Organic traffic is the KPI | 68% of US searches end with no click at all | Report AI citations, share of voice, and assisted revenue |
| Google is the only engine that matters | ChatGPT, Perplexity, Copilot, and AI Mode all shape shortlists | Track visibility per engine and weight by where your buyers ask |
| A keyword list captures demand | Buyers ask long conversational questions in AI engines | Run prompt research next to keyword research |
Nothing in the left column was a law of nature. Those assumptions held while the results page stayed static. The page moved.
What replaced the old playbook
The new job has three parts. The first is visibility across engines. Showing up when ChatGPT, Perplexity, Gemini, or AI Mode assembles an answer is now a trackable outcome with its own methods, and optimizing for generative engines has grown into a discipline of its own. The engines differ more than most teams expect. A brand can dominate Google, sit invisible in ChatGPT, and land somewhere in between on Perplexity, because each engine picks its sources its own way.
The second is citations. AI engines quote sources, and a citation carries your brand into the answer even when no click happens. Earning one takes quotable direct answers, clean structure, and authority signals, and the mechanics differ by engine. That is why getting recommended inside ChatGPT answers deserves its own playbook. Tables, stat lines, and clearly attributed claims travel well. Vague brand prose does not get quoted anywhere.
The third is being the source. Engines need something worth quoting. Original data, hands-on testing, and first-party numbers earn citations, while recycled summaries get skipped. A single original benchmark can out-earn a pile of rewritten listicles, because engines look for numbers they can attribute. Whether that work lands shows up in how often AI engines mention your brand against competitors.
Who should worry, and who can calm down
Worry if your revenue depends on informational pageviews. Publishers, affiliate sites, and ad-supported blogs sit exactly where AI answers absorb clicks. The Pew numbers describe that future plainly.
The pain concentrates on queries where the answer is the product. Definitions, conversions, quick facts, and generic how-to steps get absorbed in full. If your content model is built on those, the correction is a business-model question, and no amount of on-page tweaking reverses it. Original reporting, direct reader relationships, and channels that do not depend on search matter more than another optimization pass.
Worry much less if you sell a product or service. Transactional, local, and branded searches still produce visits, and comparison shoppers still open websites before buying. Firms with genuine expertise arguably gained ground, because engines favor demonstrated authority. That is where author-level authority signals pay for themselves.
For most businesses it comes out as a shifted budget line. Classic technical and content work still earns its keep, and a slice of the budget now goes to AI visibility. We shared real numbers on how teams are splitting spend between the two.
How the work changes in practice
None of this throws away the old craft. Crawlable sites, fast pages, clean structure, and real expertise still decide who gets ranked and who gets cited. The additions sit on top of that base. Here is what changed in our own weekly work:
- Measurement first. Default analytics undercounts AI traffic, so set up proper tracking for AI referrals in GA4 before judging any channel.
- New reporting. Impressions, citations, and share of voice belong next to clicks. We listed the numbers worth putting in front of leadership.
- Research prompts as well as keywords. Buyers ask engines long conversational questions, so we run prompt research alongside the keyword workflow.
- Write for extraction. Lead with the answer, keep claims atomic, and give engines a clean 40-60 word block worth quoting.
- Consolidate. Merge thin variants into fewer, deeper pages that fully resolve a question.
Where to start this week
Skip the existential debate and run a diagnostic instead. An afternoon of honest measurement beats another month of opinions.
- Export 12 months of Search Console queries, then join them to a rank tracker’s AI Overview flags. Search Console’s generative AI report counts impressions by page, not by query, so the trigger data has to come from a SERP tool.
- Fix attribution so ChatGPT, Perplexity, and Copilot referrals stop hiding inside direct traffic.
- Test your top 20 buying prompts across the major engines; a free AI visibility check shows where you stand today.
- Rewrite your five most valuable pages so each answers its core question in the first two sentences.
- Move reporting from traffic alone to traffic plus citations and share of voice.
SEO in 2026 pays the same way it always has, through compounding visibility. The scoreboard added columns. Teams that keep reading only the old column will keep concluding the game is over, while the rest of the market plays the whole board.