Is SEO Dead? What the 2026 Data Actually Shows

Every budget season, someone forwards you a post saying search is finished.

What died and what still works in 2026

What did AI search actually change?

The click rate. Pew has the cleanest dataset, because it measures people instead of rankings.

Pew Research Center analysed the browsing data of 900 U.S. adults for March 2025, covering 68,879 unique Google searches. Users who saw an AI summary clicked a traditional search result link in 8% of visits. Without a summary, they clicked "nearly twice as often (15% of visits)". The links inside the summary did worse: users clicked one in "just 1% of all visits to pages with such a summary."

Pew also found that only 8% of one- or two-word searches produced an AI summary, but 53% of searches with 10 words or more did. Long questions get answered on the page.

Ahrefs measured the same effect from the publisher side. Across 300,000 keywords, comparing December 2023 with December 2025, an AI Overview "now correlates with a 58% lower average clickthrough rate for the top-ranking page." Its April 2025 study had found a 34.5% drop over one year, December 2023 to December 2024. The new figure spans two years, so the two readings are not a straight trend line.

Seer Interactive shows that the trend does not move in one direction only. Its September 2025 update found organic CTR at 0.52% where an AI Overview appeared and the brand was not cited, down 65.2% year over year. Queries with no AI Overview fell too, to 1.45% (down 46.2%), and Seer noted "even queries WITHOUT AIOs are declining significantly in CTRs". Then its larger April 2026 study found organic CTR on AI Overview queries "climbed from a floor of 1.3% in December 2025 to 2.4% in February 2026". Seer calls that directional, based on two months.

Google disputes reports like these. In an August 2025 post, published before both Seer studies and the 2026 Ahrefs study, Google's Head of Search, Liz Reid, wrote that "total organic click volume from Google Search to websites has been relatively stable year-over-year" and that Google is "sending slightly more quality clicks to websites than a year ago". She said third-party reports of dramatic declines in aggregate traffic were "often based on flawed methodologies, isolated examples, or traffic changes that occurred prior to the roll out of AI features in Search." Google published no figures behind the claim.

Both sides can be right. Google describes total clicks across every site; Pew, Ahrefs and Seer measure clicks per query or per visit. The same post concedes the mechanism: "user trends are shifting traffic to different sites, resulting in decreased traffic to some sites and increased traffic to others." Your company is one site, so plan for the per-query numbers.

How often an AI Overview shows up depends on what you search. Pew found one on 18% of all searches in March 2025. Seer's 2026 study puts it at 36% of informational queries. In our own 2026 report, covering a panel of 180 common B2B software searches pulled on 24 October 2026, an AI Overview appeared on all 180, at the top of the page on 179. As that report puts it, "It says nothing about search as a whole."

Demand did not collapse while all this happened. Google says it sees "more than 5 trillion searches on Google annually", and that in the U.S. and India, "AI Overviews are driving over 10% increase in usage of Google for the types of queries that show AI Overviews". People search more. They click less per search.

What does our Search Console data show?

We pulled non-brand query data from Search Console for five B2B sites we write for, comparing Q3 2025 (1 July to 30 September) with Q3 2026. We labelled any query of 10 or more words as long.

Search Console misreported impressions for nearly a year, including our 2025 window. Google lists only impressions, CTR and average position as affected, so we lead with two measures that lean less on impression totals: the share of distinct queries and the share of clicks. The click share is the one Google says the error did not touch. The distinct-query share is less exposed but not immune, because a query only shows up in Search Console once an impression has been logged for it.

The share of distinct non-brand queries with 10 or more words rose from 1.7% in Q3 2025 to 13.3% in Q3 2026, pooled across the five sites. It rose on all five. Per site, the Q3 2026 share ranged from 1.9% to 35.6%, up from 0% to 3.5% a year earlier. Those long queries brought in 0% of non-brand clicks in Q3 2025 and 1.2% in Q3 2026.

Across five B2B sites, queries of 10 or more words grew from 1.7% to 13.3% of distinct non-brand queries in a year, yet brought in only 1.2% of non-brand clicks.

Impressions tell the same story, louder. Pooled, 10+ word queries took 0.6% of non-brand impressions in Q3 2025 and 10.1% in Q3 2026. The 10+ word share rose on all 5 of 5 sites, with Q3 2026 shares between 2.2% and 29.8%. In Q3 2026, their pooled CTR rounds to 0.0%, against 0.4% for human-length queries.

Read the impression figures with one caveat. Google's Search Console help page says: "A logging error prevented Search Console from accurately reporting impressions from May 13, 2025 until April 27, 2026." Our Q3 2025 window sits inside that error and our Q3 2026 window sits after the fix, so the impression comparison sets a broken period against a clean one.

Google does not say who types these long queries. It does say AI Overviews and AI Mode "are included in the overall search traffic in Search Console", and that in AI Mode, "If a user asks a follow-up question within AI Mode, they are essentially performing a new query". When John Mueller was asked about odd queries such as 'yes go on', he replied: "Search Console includes information on AI Overviews and AI Mode in the general performance report". Search Engine Roundtable's counterpoint is fair: "It is not clear if these are AI Mode queries, they can just be super long Google Search queries."

Our read: the rise is consistent with AI Mode, AI Overview and agent activity. It is not proof of any one of them.

The human side is mixed. Human-length non-brand clicks (queries under 10 words) grew on 3 of the 5 sites and fell on 2. The two declines were -78% and -35%, and on both of those sites human-length impressions fell too (-55% and -58%), though those impression changes cross Google's logging error, so treat them as rough. Before you blame AI for a traffic drop, check whether your short-query clicks fell, and whether your comparison crosses that error window.

What died, and what replaced it?

Shortcuts died. Google made it official on 5 March 2024, when it announced "three new spam policies against bad practices we've seen grow in popularity: expired domain abuse, scaled content abuse, and site reputation abuse."

Scaled content abuse "is when many pages are generated for the primary purpose of manipulating Search rankings and not helping users". Google can act "no matter whether content is produced through automation, human efforts, or some combination of human and automated processes." Volume made for rankings is the offence, whatever wrote it.

Google expected the March 2024 changes to cut low-quality, unoriginal content by 40%. Its April 2024 update reported: "You'll now see 45% less low-quality, unoriginal content in search results versus the 40% improvement we expected across this work."

Google's helpful content guidance gives the line in one sentence: "SEO can be a helpful activity when it is applied to people-first content, rather than search engine-first content." One of its listed warning signs is blunt: "Are you using extensive automation to produce content on many topics?"

What replaced the shortcuts is harder to fake. Google says people are increasingly clicking on "sites with forums, videos, podcasts, and posts where they can hear authentic voices and first-hand perspectives", and are more likely to click into content that helps them learn more, such as "an in-depth review, an original post, a unique perspective or a thoughtful first-person analysis". If your content calendar is rewrites of what already ranks, you are producing the unoriginal content Google set out to reduce. First-party data and honest comparisons still earn a place.

Do AI answers still come from pages that rank?

Often, though less than before. Google says the ordinary rules apply. Its AI features guide states: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." The eligibility rule is ordinary SEO: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet". Our AEO vs GEO vs SEO breakdown covers where the disciplines overlap.

Ahrefs found that "37.9% of URLs cited in AI Overviews also appeared within the first 10 blocks"; a further 31.2% ranked in positions 11 to 100, and 31.0% sat beyond the top 100. Ahrefs says Google now picks far fewer pages straight from the results page than in July 2025 and puts that down to fan-out, which Google describes as "issuing multiple related searches across subtopics and data sources". Each of those sub-searches is a ranking you can win or lose.

Our own panel found a weaker link on software searches. The AI Overview cited the exact page ranking first on 22.2% of searches, and 19.7% of all citations also appeared in the same search's page-1 results. Different method from Ahrefs: about four in five citations in our panel did not appear in that search's page-1 results. Ranking helps you get cited. It does not guarantee the quote.

The rest of the pool is distribution. Ahrefs reports "YouTube is the most cited domain in AI Overviews today, and has grown 34% over the last six months." In our panel YouTube was cited in 73.9% of overviews and made up 17.4% of citations, against 2.5% of organic results.

ChatGPT search has its own crawler. OpenAI's documentation says: "OAI-SearchBot is for search. OAI-SearchBot is used to surface websites in search results in ChatGPT's search features. Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers".

Each engine trusts a different corpus. In our own category tracking from 3 August to 1 September 2026, YouTube and LinkedIn were the most cited domains in Google AI Overviews and Reddit was among the leaders, though each held only a small share of citations. ChatGPT's sources included arXiv and Reddit. Listicles made up 81% of Perplexity's citations. A Google plan is a distribution plan. Our guide to getting recommended in ChatGPT covers the ChatGPT side.

How do B2B buyers search now?

Nearly all of them use AI somewhere along the way. Forrester, drawing on its Buyers' Journey Survey, 2025, reports that the share of business buyers using AI in their buying process rose five points to 94%. It also found "twice as many buyers" named generative AI or conversational search as a more meaningful source than any other, and 61% use private AI tools their organisation provides.

G2 surveyed 1,076 B2B software buyers in March 2026 and found: "Half of B2B software buyers (51%) now start their research with an AI chatbot more often than with Google". In G2's 2025 report, that figure was 29%. G2 also found 61% use AI search alongside Google, and 69% chose a different vendor than planned because of a chatbot recommendation. G2 sells review-site presence, so it has an interest in that answer. Forrester's data covers AI use in buying, not where research starts, so it supports the trend without confirming G2's 51%.

The chatbot market is splitting, too. Similarweb found that ChatGPT's share of generative AI website visits "fell from about 76% in June 2025 to roughly 53% by May 2026", while Gemini rose to around 27 to 28%.

G2's 61% describes buyers who use AI search alongside Google, so one buying decision can run through several surfaces. A program built for one of them misses the rest.

So is SEO still worth it for a B2B SaaS company?

Yes, if you change what you buy it for. Seer's 2026 data gives you the split. AI Overviews show on 36% of informational queries, 8% of commercial queries and 5% of transactional ones. Among informational queries, comparison queries triggered one 95.4% of the time. Software searches run hotter: our panel found an AI Overview on all 180 common B2B software searches, including every commercial one.

Keep investing in commercial and transactional pages, because they are the pages buyers use to choose a vendor, and that is where pipeline comes from. Do not expect them to dodge AI Overviews on software searches; build them to be cited as well as ranked. Ranking alone will not carry them. In our panel, 30.0% of AI Overview citations on pricing searches also ranked on page 1, but on "best X" and alternatives searches the share fell to 11.6% and 12.4%, so most of what the overview cites there comes from outside page 1.

Change the mix where the answer now lives on Google's page, which means top-of-funnel explainers written to collect traffic. The fix is to rebuild them as pages worth citing, with original data or a comparison table and a clear verdict. Seer found being cited "delivers +120% more organic clicks per impression versus when you are not cited", though it "still underperforms No AIO by -38%".

If you are still deciding sequence, our breakdown of SEO vs AEO covers which to fund first, and how long SEO takes covers when to expect movement.

Then fix the scoreboard. Measure these four:

1. Clicks. Search Console defines them as "The number of times a user clicked your site from Google Search results". They escaped the logging error, so they are your cleanest trend line.
2. Non-brand clicks on human-length queries. Filter out brand terms and queries of 10 or more words. In our data for Q3 2026, after Google fixed its impression-logging error, 10+ word queries took 10.1% of non-brand impressions and 1.2% of clicks, so leaving them in inflates impressions without adding clicks.
3. Pipeline. Organic search exists to fill the pipeline with qualified leads, and that is the number a board should judge it on. Google points the same way: "In addition to Search Console, you could also track conversions and time spent on your site in other tools, such as Google Analytics."
4. AI named rate, per engine. Track how often each engine names your brand on your buyers' questions. Each engine draws on a different corpus, so a blended number hides the one you are losing. Our explainer on AI visibility covers how it is measured.

Seer's advice belongs on every SEO dashboard: "Always look at clicks and impressions separately before calling something a problem."

SEO and AI search draw on the same pages and the same authority, so we run them as one SEO and AEO program.

How we measured this

The first-party figures come from the Search Console Search Analytics API, query dimension, non-brand queries only (brand terms removed per site), read on 25 October 2026. We compared Q3 2025 (1 July to 30 September 2025) with Q3 2026 (1 July to 30 September 2026).

• Sites. Five B2B sites we write for, ours and our clients'. They were selected, so treat click growth as true of these sites only. Six other sites had no 2025 data and were excluded. One site had a very small 2025 baseline.
• Long queries. Long means 10 or more words; human-length means under 10. The labels describe length and prove nothing about who typed the query. The 10-word cut-off follows an example regex in a Practical Ecommerce article, which suggests experimenting with any length.
• Anonymised queries. Search Console drops rare, anonymised queries from query-level rows.
• The logging error. Google reports an impression-logging error from 13 May 2025 to 27 April 2026. Our 2025 window falls inside it, our 2026 window after it, so we added two measures that lean less on impression totals: distinct-query share and click share. Google says clicks were unaffected; a query row still needs a logged impression, so the distinct-query share is less exposed, not immune.
• The September 2025 change. Google stopped supporting the 100-results-per-page parameter in September 2025, and many sites saw desktop impressions fall. If scraper tools had inflated 2025 impressions, the true change in either long-query share, by impressions or by distinct queries, could be larger or smaller than shown.