You will see many different numbers for how many Google searches end without a click. Some say about 47%, some about 59%, some about 68%. They are not contradicting each other as much as they seem. They use different data, different periods and different definitions.
This page sets out three studies we opened, what each measured, and how to use the figures without overstating them. It replaces our January 2026 version, which promised “final numbers” we could not support.
From our own check, 10 November 2026. We opened three studies. SparkToro’s 2024 study found 58.5% of US and 59.7% of EU Google searches ended with zero clicks. SparkToro’s 2026 study, using Similarweb data, found 68.01%. iPullRank’s study of 10.2 billion sessions found 46.96%.
They use different data, periods and definitions, so we show them side by side and do not call any of them final.
Table of Contents
ToggleWhat does "zero-click" mean?
A zero-click search is a search where the person does not click a result. SparkToro’s 2024 study counts a search as zero-click when the person does not click through, and its own caveats list searches answered by Google Assistant or the Google app, which it does not count [1]. The iPullRank study defines a zero-click session as one where no click row exists for that session [3]. These are not the same test, which is one reason the numbers differ.
What did three studies find?
| Study | Data and period | Zero-click share |
|---|---|---|
| SparkToro, 2024 study (published 2 July 2024) [1] | Datos clickstream panel, US and EU, September 2022 to May 2024 | 58.5% of US and 59.7% of EU Google searches |
| SparkToro, 2026 study (published 9 June 2026) [2] | Similarweb data, US, January to April 2026 | 68.01% of Google searches |
| iPullRank (30 July 2026) [3] | Datos clickstream, 10.2 billion sessions, October 2024 to December 2025 | 46.96% of Google searches |
The 2024 SparkToro study also reports clicks, not just the share without one. For every 1,000 US Google searches, 360 clicks go to a property that Google does not own and that does not pay Google for ads. In the EU the figure is 374 [1].
Why do the numbers differ?
- Different data providers. One study uses Datos panel data. The 2026 SparkToro study uses Similarweb data. The 2026 SparkToro page gives a 2024 US figure of 60.45%, which is not the 58.5% from the Datos study, and it does not explain the gap [2].
- Different devices. iPullRank says its panel is 94.5% desktop [3]. SparkToro’s 2024 study says mobile data only starts in January 2024 [1].
- Different periods. The studies cover different months and years.
- Different definitions. SparkToro’s 2024 study counts searches. iPullRank counts sessions.
SparkToro lists its own caveats too: the panel may not be representative, iOS coverage is minimal, and ad blockers may undercount paid clicks and overstate organic ones [1].
What do the three look like side by side?
The picture sets the three studies next to each other. You can zoom and pan it, or expand it. The same points are written out under the graphic.
Read the workflow as text
- 58.5% (US, 2024) SparkToro, Datos panel, Sep 2022 to May 2024. EU: 59.7%.
- 68.01% (US, 2026) SparkToro using Similarweb data, January to April 2026.
- 46.96% (2024 to 2025) iPullRank, Datos clickstream, 10.2 billion sessions. Panel 94.5% desktop.
What would each figure mean for your searches?
Enter a number of Google searches and pick a study. The tool shows how many would end without a click and how many would end with one. The second figure is the other side of the percentage, so it can include clicks on Google’s own results. It runs in your browser and sends nothing anywhere.
Enter a number of Google searches and choose a published zero-click share. The tool shows how many searches end without a click and how many end with one.
How should you use these figures?
- Quote a range, not one number. From these three studies, anywhere between about 47% and 68% depending on the method.
- Name the study and its period. “58.5% in 2024 (SparkToro, Datos panel)” is checkable. “About 60%” is not.
- Do not compare across providers. A 2026 Similarweb figure and a 2024 Datos figure are not a trend line.
- Look at your own queries. In Search Console, compare impressions with clicks for the queries you care about.
A note on the earlier version of this page. It was published in January 2026 with a headline of “final numbers” and no sources. We could not support that, so we rebuilt it in November 2026 using only the pages listed below.
Which guides should you read next?
- Publisher Google traffic fell by a third in 2025 looks at another set of traffic figures and what they cover.
- Answer-first content structures with AI covers what Google says about AI features.
- January 2026 SEO news, checked shows how we date and source news items.







