No. 029
The release says 60%. The study it links says 68.01%.
A paid release headlines 60% of searches ending without a click. The SparkToro study its own report page links reports 68.01%, counted a different way.
60%vs68.01%
What each page says
| Document | Date | The number | The denominator |
|---|---|---|---|
| The press release, distributed by Business Wire | 2026-10-07 | ”60% of Searches Now End Without a Click” (headline) | “Searches” |
| The same release, in the body | 2026-10-07 | ”60% of informational queries are now resolved within the search interface itself and without the user ever going to the brand’s domain" | "informational queries” |
| Incubeta’s report page | no date shown | ”over 60% of informational queries are now resolved within the search interface itself without a user ever navigating to a brand’s domain" | "informational queries” |
| SparkToro’s study, the page the report links | 2026-06-08 | ”In the first four months of 2026, a whopping 68.01% of Google searches ended without a click." | "Google searches” |
The release carries the label “This is a paid press release” on the page it was distributed to. It describes its publisher as “a marketing outcomes agency powered by AI”. Neither of those facts makes the 60% false. They do tell you which document to open next.
The release itself contains no instance of the words SparkToro, Similarweb, Gartner, survey, or the phrase “according to”. The sourcing lives one click away, on the report page, inside two pieces of anchor text. “Recent research” points at the SparkToro study. “Gartner projects” points at a Gartner press release from February 2024.
The word that is not in the source
The release and the report both say “informational queries”. That qualifier is doing a lot of work: it narrows the claim to the subset of searches where someone wants a fact rather than a product or a website, which is exactly the subset you would expect to be answered on the page.
The word “informational” appears zero times on the SparkToro page. SparkToro counted all Google searches in its panel and split them by what happened next, not by what the searcher wanted. There is no informational cut in the study, so there is no 60% of informational queries in it either.
So the narrower-sounding claim is not the cautious version of the broader one. It is a different claim, and the cited source does not contain it.
What a zero-click search includes
The second gap is the verb. The release says those searches are “resolved within the search interface itself”. SparkToro says they “ended without a click”. Those are not the same event, and SparkToro is explicit about the difference.
From the study’s own description of the metric it calls Clicks 1X+:
It includes clicks of any kind (save for another search in the default search box): Google’s other properties (like a search in Google Maps or a visit to YouTube), clicks on organic links to the open web
Read that backwards and you get the shape of the zero-click bucket. A search counts as zero-click when the only thing the searcher did next was type another search. SparkToro names that group directly and sizes its movement:
the percent of searchers who perform another search. It’s up 7.2 points from 2024 to 2026, showing the search giant’s success at persuading searchers to perform ever more queries
A searcher who types a query, gets nothing useful, and immediately retypes it has produced a zero-click search. Nothing was resolved. The study also records that “Only 0.34% of searches made their way to AI Mode” in the same period, which is a useful corrective to any reading of this number as an AI-answers number.
Two further scope limits, both stated on the page and both dropped in the retelling. The panel covers browser searches, and the study says its data “is not inclusive of the Google mobile search application, but rather mobile searches within the browser”. And the number is United States only.
The trend is real, and smaller than the gap
None of this makes the direction wrong. SparkToro’s own two-year comparison, from the page text:
| Period | Zero-click share of US Google searches | Source panel |
|---|---|---|
| Ten years ago | ”~45%“ | stated on the page as approximate |
| 2024 | 60.45% | Datos |
| January to April 2026 | 68.01% | Similarweb |
The study puts the two-year move at “12.5% growth (7.5 percentage points)” and the decade move at “a 33.8% increase (23 points)”. It also flags that the panels differ by year, which is why those are its numbers and not ours.
A reader handed “60%” and a reader handed “68.01%” will make the same strategic decision. The cost of the gap is not the decision. It is that the 60% arrives with no method, no period, no country and no definition, so there is nothing to check it against next quarter.
The Gartner sentence picked up three changes
The same paragraph of the report restates a second number, and this one can be diffed word for word, because the original is a press release with a fixed text.
| Gartner, 19 February 2024 | Incubeta’s report, 2026 | |
|---|---|---|
| Verb | ”Predicts” in the title; “will drop” in the body | ”projects” |
| Size | ”25%" | "at least 25%“ |
| Deadline | ”By 2026" | "by the end of 2026” |
Gartner’s sentence, from the archived release:
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.
The phrase “at least” does not occur anywhere in that release. A point forecast became a floor, which is the one edit that cannot be walked back: a floor can only be beaten, so the restated version is unfalsifiable in a way the original was not. The basis is also not public. Gartner’s release points its own clients at a subscriber document, so no reader outside Gartner can check how 25% was derived.
We cite the Internet Archive copy because Gartner’s live URL returns HTTP 403 to automated requests, and a citation that cannot be opened is not a citation.
A correction to three of our own posts
This is the part we owe readers before any of the above counts for anything.
Three articles on this site, all published on 18 July 2026, attribute the 68.01% figure to a 2024 SparkToro analysis and link to SparkToro’s 2019 post. The figure is from the January to April 2026 study, published on 8 June 2026. SparkToro’s 2024 figure is 60.45% on the page we should have been citing.
The three are “The position-1 CTR collapse: from 7.3% to 1.6% in two years”, “AI search visibility statistics 2026”, and “AI referral traffic converts”. The year and the link have been corrected in all three as part of publishing this post. The figure itself was right; the date and the destination were not.
We went looking for this because the discrepancy above is the kind that gets found by reading the linked page, and reading our own linked pages is the same job.
What to do with a 60% you are handed
Four checks, in the order that catches the most with the least effort.
- Open the link, not the number. If the claim is sourced behind anchor text, follow it. The gap in this case was visible in one click.
- Read the denominator out loud. “Searches”, “informational queries” and “sessions” are three different populations. A number that changes population between a headline and a body has been edited, not summarised.
- Find the verb. “Ended without a click” is a measurement. “Resolved” is an interpretation of that measurement, and it is the interpretation that sells.
- Check whether the figure can ever be wrong. “At least 25%” cannot. “25% by 2026” can, which is why it was worth publishing in the first place.
None of this requires access to a panel, a crawler or a tracking tool. It requires opening four pages and reading them in the order they cite each other.