Blog · 25 September 2026 · By HeardOf

Did search volume fall 25% by 2026? Gartner's prediction against the measured figures, each in its own unit

Did Gartner's 25% search volume prediction come true?

Not on any page we read on 24 September 2026, and none of those pages measures the thing Gartner named, for 2026. The prediction is about "traditional search engine volume", which the release's own text treats as user queries; the measurements we read count a share of referrals (Statcounter), searches per searcher on a panel (Datos with SparkToro, for 2024), the share of searches that end without a click (Similarweb with SparkToro, for early 2026), clicks per results-page visit (Pew, for March 2025), and paid ad clicks and revenue (Alphabet). For 2025 and 2026, Google is the one party on those pages that reports on queries, and each Google statement that describes a change says they rose. Asked plainly, did the Gartner search volume 25% prediction come true? No page we read says it did, and no page we read holds an independent 2026 query count that could settle it either way.

What exactly did Gartner predict on 19 February 2024?

"By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents, according to Gartner, Inc." That is the first sentence of the release, datelined Stamford, 19 February 2024 (historical), under the heading "Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due to AI Chatbots and Other Virtual Agents". The page prints no base year, no baseline count, no definition of "traditional search engine volume" and no method; it says Gartner clients "can read more in" a report titled "Predicts 2024: How GenAI Will Reshape Tech Marketing", which we could not open. The nearest thing to a definition is the quote from Alan Antin, Vice President Analyst: "Generative AI (GenAI) solutions are becoming substitute answer engines, replacing user queries that previously may have been executed in traditional search engines." So the nearest the release comes to a unit is user queries, the base is a level the release does not state, and the sentence mixes two claims, a volume drop and a market-share loss, that would need two measurements. Gartner's page puts a Cloudflare check in front of scripts and browsers alike, so we read the Internet Archive's capture of it from 20 February 2024, the day after publication, which carries the text quoted here.

What has Google said about query volume since?

That queries grew, in every statement we read that describes a change, with one rounded count and no series. On 7 May 2025 (historical), in a statement on press reports about Search traffic: "We continue to see overall query growth in Search." On 20 May 2025 (historical): AI Overviews "is driving over 10% increase in usage of Google for the types of queries that show AI Overviews", in markets like the US and India. On 21 May 2025 (historical): "We see over 5 trillion searches a year on Google", in the Google Marketing Live post, and "more than 5 trillion searches on Google annually" in a second post the same day. On 6 August 2025 (historical), Liz Reid, VP, Head of Google Search: people "are searching more than ever", and "total organic click volume from Google Search to websites has been relatively stable year-over-year". On 4 February 2026 (historical), Alphabet's fourth-quarter release: "Search saw more usage than ever before". On 29 April 2026 (historical), the first-quarter release: "queries at an all time high". On 19 May 2026 (historical), the I/O post on Search: "last quarter, we saw queries reach an all-time high", and a companion post the same day: "AI Mode queries have more than doubled every quarter since launch". On 22 July 2026, the second-quarter release: "Our popular AI features are driving Search query growth." None of these is an independent measurement, and none prints a count beyond the 5 trillion, which carries no method and no year-on-year figure.

What do the measurements show, and in which unit?

Shares, rates, clicks and dollars, and none of them is Gartner's unit unless the page says so. The table lists the pages we read that bear on search volume in 2024, 2025 or 2026, with the unit each page states for itself, or none where the page prints no figure.

Read across, the table says three things. Statcounter answers which engine sent the click, not how many searches were made: "It's the click to a website that we measure, not the actual search queries that were performed", its FAQ says, so Google's fall of 0.59 points between February 2024 (historical) and August 2026 is a share of clicks out of results pages on a network Statcounter puts at more than 1 million sites. The one page whose unit is searches, SparkToro's of March 2025, stops at 2024; its June 2026 post measures a different thing on a different panel and calls its own 2024-to-2026 comparison "a bit of apples and oranges". Alphabet's filings give revenue and paid clicks, both up, and the 10-Q names "fluctuations in search queries resulting from changes in user adoption and usage" as one of several drivers of paid clicks, which is as close as a filing comes to the query count and is not one. Clicks on results pages that carry an AI Overview are a separate question, and the studies that measured them are compiled, each with its unit, in do AI Overviews cut clicks.

What could we not trace or not read?

Five things. Gartner's baseline: the release prints no base year and no count, and the report it points to is client access only, so the 25% cannot be placed against any figure Gartner itself states. futurefactors.ai's "over 90% of the search market", which its FAQ uses to answer "Did Google search traffic actually drop by 25% in 2026?" with "No", cites "Statista and SimilarWeb (April 2026)" and names no page; its "883 million monthly active users" for ChatGPT has no source beside it. Datos's State of Search Q1 and Q2 2026 report pages print no figure; the reports are behind a form we did not fill in. Search Engine Land, the page SparkToro links for Google's 5 trillion, answered 403 to us; the figure is on two of Google's own pages of 21 May 2025 (historical) instead, and that is where we took it from. And an independent count of total Google queries for 2025 or 2026: on the pages we read there is none; the two panel studies give a per-searcher rate for 2024 and shares of searches for early 2026.

What does this mean for a B2B software company?

That the public figures cannot answer the question for any one company, and the question that can be answered is a different one. The first of the three questions in GEO agency vs SEO agency is whether traditional search traffic is still growing for you, and on the pages above, the aggregate cannot settle it: Google's queries rose by Google's account, the share of searches that send a click fell on the panels SparkToro reports, and what happened in your category is in your own analytics and nowhere on these pages. The second question, whether your buyers use assistants before Google, has survey answers with samples and dates in do B2B buyers research software with AI. What none of the pages in this post measures is what the assistants say when a buyer asks. That is the measurement HeardOf makes: which brands are named and which sites are cited for buyer prompts in your category, on the four engines we ask, as updated 23 September 2026, our own settings from that post and from our own systems, not from any page here.

How was this read?

Every page was fetched on 24 September 2026 between 23:43 and 23:57 UTC with curl and a Chrome user-agent string, and the Internet Archive capture and the SEC EDGAR filings between 00:38 and 00:40 UTC on 25 September with a user-agent string naming this research, one machine, one location, and read in full; each figure was found in the text of the page it is attributed to. Gartner's newsroom and Alphabet's earnings index both put a Cloudflare check in front of scripts and browsers, and nothing was read past it. Gartner's release was read from the Internet Archive's capture of 20 February 2024. Alphabet's three earnings releases and two 10-Q filings were read from SEC EDGAR, where Alphabet files them: the releases as exhibit 99.1 to its 8-Ks of 4 February, 29 April and 22 July 2026, and the 10-Qs as filed on 30 April and 23 July 2026. Statcounter's monthly figures come from its own CSV download for Worldwide, all platforms, January 2024 to September 2026, fetched at 23:47 UTC; the September 2026 row is a month in progress, and Statcounter's about page says its stats are subject to revision for 45 days. Search Engine Land answered 403 with a Cloudflare block page and was not retried. Every URL came from the sites' own sitemaps, from links on pages we read, from our own blog's index, from the Internet Archive, or from SEC EDGAR; no search engine was used. No page we read carried text addressed to AI assistants or agents, and none linked to an AI assistant with a prefilled prompt. Everything this post says about the sources carries a link or a date and you can open any of it; what it says about our own system comes from our own posts and cannot be checked from here.

Do AI Overviews cut clicks? The published figures, with what each one measured