SEO Is Not Dead. We Checked the Prediction That Said It Would Be.
In February 2024, Gartner predicted that traditional search engine volume would drop 25% by 2026. The forecast was repeated everywhere, usually as settled fact. It is now the second half of 2026, so we went and checked. Search volume did not fall. A different number fell hard, and it is the one that decides whether a customer ever reaches you.
The Prediction Everyone Quotes
If you have read anything about AI and search in the past two years, you have met this statistic. It appears in agency pitch decks, vendor landing pages, conference keynotes, and a very large number of blog posts about answer engine optimization. The wording is consistent because everyone is quoting the same press release: "By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents."
Gartner published that on 19 February 2024. It was a prediction about a specific measurable thing, search volume, with a specific magnitude, 25%, and a specific deadline, 2026. That is an unusually testable forecast, which is to its credit. Most industry predictions are vague enough to never be wrong.
What happened next is the interesting part. The statistic detached from its source and began circulating as a fact about the present rather than a guess about the future. We have read dozens of pages in this category that cite the 25% figure without noting that it is a forecast, without naming the year it was made, and without ever going back to see whether it came true.
People Questioned It at the Time
The prediction was not accepted uncritically by everyone. It was challenged immediately by people who cover search professionally, and their objections are worth revisiting because several of them turned out to be correct.
Search Engine Land's Danny Goodwin published a piece the same day it was announced, and his central objection was epistemic rather than technical. "A prediction is just a guess," he wrote. "Gartner's predictions are typically informed, but they're still best guesses." He made a sharper point about how the industry treats analyst forecasts: "Respected analyst firms are rarely called out when they're wrong, because people rarely go back and check." He also noted that the underlying Gartner consumer survey had 299 respondents.
Search Engine Journal followed in March 2024 with a piece listing reasons to doubt the forecast. Its strongest argument was that the prediction assumed AI chatbots would substitute for search engines, while search engines were in fact busy absorbing AI into themselves. If the incumbent adds the new capability to its own product, users do not need to leave, and the volume never moves.
That objection turned out to be the decisive one.
What Actually Happened to Search Volume
Google did not lose a quarter of its search volume. It integrated generative AI directly into the results page through AI Overviews and AI Mode, and by its own account usage went up rather than down.
In Alphabet's Q1 2026 results, reported on 29 April 2026, Sundar Pichai said queries were at an all time high and that "People love our AI experiences like AI Mode and AI Overviews, and they're coming back to Search more." Search and other revenue for the quarter was $60.4 billion, up 19% year over year, an acceleration from the previous quarter rather than a decline.
A caveat belongs here, and we would rather state it than let a reader find it later. That is a company describing its own product to investors. It is not independent measurement, Google does not publish raw query counts, and "all time high" is a claim we cannot audit. But the direction is not ambiguous, and it is not the direction the forecast required. A business losing 25% of its volume does not report accelerating revenue growth from that same product.
The prediction was wrong in direction, not merely in magnitude. Search volume was supposed to fall by a quarter. It rose. The year is more than half over, and nothing in the public record suggests a collapse in the remaining months.
The Steelman, Because It Deserves One
There is a genuine defence of the forecast and it should be dealt with honestly rather than skipped.
Gartner's wording was "traditional search engine volume." When a user types a question and receives a generated answer at the top of the page, it is arguable that something other than traditional search just happened, even though it was counted as a Google query. On that reading the prediction was directionally right about behaviour and merely wrong about which metric would record it.
We think that defence is partly fair and mostly not. It is fair because user behaviour genuinely did change, and the people who felt something shifting were not imagining it. It is not fair because the prediction named a number that people then used to make decisions, and the mechanism it described, users leaving search engines for chatbots, is not what occurred. Users did not leave. The answer moved to meet them where they already were.
The distinction matters commercially. If your customers had left Google for ChatGPT, your response would be to go compete on ChatGPT. That is not what happened, and a business that reallocated on that basis was solving the wrong problem.
The Number That Actually Moved
Here is what the 25% conversation obscured for two years. While everyone argued about whether people would stop searching, the thing that changed was what happens after they search.
The Pew Research Center studied this directly and published in July 2025. The method matters, so here it is: 900 US adults from Pew's KnowledgePanel agreed to share their browsing activity, and Pew analysed 68,879 unique Google searches performed during March 2025. Of those, 12,593 produced an AI summary.
The finding was stark. When an AI summary appeared, users clicked a traditional search result in 8% of visits. When no AI summary appeared, they clicked a result in 15% of visits, nearly twice as often. Clicks on links inside the AI summary itself were rarer still, at just 1% of visits to pages carrying one. Users were also more likely to end their browsing session entirely on pages with a summary, 26% against 16% without.
Those are the honest limits of that study: US adults only, Google only, one month, and a panel who consented to be tracked. It is a single well constructed snapshot rather than a permanent law. It is also the most rigorous public measurement we found of the thing that actually matters, and no one repeating the 25% figure was citing it.
The Query Survived. The Click Did Not.
Put the two findings next to each other and the real story assembles itself.
People did not stop searching. They search more than ever. What they stopped doing is leaving the results page. The query survived intact and the click through it used to generate roughly halved in the presence of an AI summary.
For a business, that is a far more serious development than the one that was predicted, and it is serious in a different way. A 25% fall in search volume would have been a proportional problem: everyone loses a quarter, relative position holds, the market shrinks a little. What actually happened is not proportional at all. When the engine composes one answer and names a few sources, the businesses inside that answer keep their visibility and the businesses outside it lose almost all of theirs. There is no page two to be on. There is the answer, and there is everything the answer left out.
The competition did not shrink. It narrowed.
Everyone spent two years arguing about whether people would stop searching. They did not. They stopped clicking. That is the number that should have been on the slide.
What This Means If You Run a Business
Three things follow, and none of them is "abandon SEO."
The fundamentals did not become optional, they became the entry fee. A page that a machine cannot reach, render, or parse cannot be retrieved into an answer. Crawlability, clean markup, fast stable pages, and accurate structured data are prerequisites now rather than achievements. The businesses declaring SEO dead are often the ones whose technical foundations are quietly failing.
Being ranked and being cited are now separate outcomes. You can hold position three for a query and still be absent from the answer that query produces. Those are different contests decided by different signals, and only one of them still reliably produces a visitor. This is the practical reason we score retrieval legibility rather than rank.
A citation with no click is now a real form of visibility. If a customer asks an engine who the good options are and hears your name with a sentence about why, that had value even though your analytics never recorded a visit. Measuring your presence purely in sessions will increasingly understate what is happening to you, in both directions.
What We Could Not Verify
Researching this piece meant reading a large amount of writing in this category, and a striking share of the statistics in it do not survive being followed back to a source. In the interest of not adding to the pile, here is what we deliberately left out.
We found widely repeated figures for zero click search rates, organic traffic decline percentages, click through rate collapse, and conversion rate multiples for AI referred traffic. In each case the trail led to a marketing blog citing another marketing blog, with no primary study at the end of it. We also found specific funding and valuation figures for several AI visibility vendors that appear only in competitor comparison posts and not in any primary announcement.
None of those numbers appear above. Some of them may well be true. We could not confirm them, and a statistic you cannot trace is worth less than no statistic at all, particularly on a page arguing that other people should check their sources.
The Short Version
SEO is not dead. The prediction that said it would be got the direction wrong, and it got two years of attention that should have gone to the measurement that mattered. Search volume held. Click through, in the presence of an AI answer, roughly halved.
Which means the question worth asking about your own business is no longer whether you rank. It is whether a machine that reads you can understand who you are, verify it against something other than your own marketing, and trust you enough to put your name in the answer.
That is a question with a concrete answer, and it is the one we built our evaluation to give you.
Related reading: AEO vs GEO vs SEO sorts out the vocabulary this field keeps arguing about. AI Visibility Is Not SEO 2.0 argues that this is a different contest rather than a new version of the old one. Why SEO Alone Is No Longer Enough covers what to add once the foundations are sound.
Sources
- Gartner press release, 19 February 2024 The original prediction: "By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents." gartner.com
- Search Engine Land, Danny Goodwin, 19 February 2024 Contemporaneous skepticism, and the note that the underlying Gartner consumer survey had 299 respondents. searchengineland.com
- Search Engine Journal, March 2024 Seven reasons to doubt the forecast, including the argument that search engines would absorb AI rather than lose users to it. searchenginejournal.com
- Pew Research Center, 22 July 2025 900 US adults, 68,879 Google searches during March 2025, 12,593 of which produced an AI summary. Click through of 8% with a summary against 15% without, 1% clicks inside summaries, 26% session abandonment against 16%. pewresearch.org
- Alphabet Q1 2026 results, reported 29 April 2026 Search and other revenue of $60.4 billion, up 19% year over year, with Sundar Pichai stating that queries were at an all time high. searchenginejournal.com