Blog · 3 min read
68% of US Google searches end without a click. Over half go unnamed
By Simplee Editorial Team. Reviewed by Vlad Sherbatov · Published · Updated
Sources checked October 20, 2026 · Methodology
68% of US Google searches ended without a click from January to April 2026, according to SparkToro's analysis of Similarweb panel data. In the same year, more than half of the 826 businesses that ran our free AI visibility audit, at least 51%, were never named by any AI engine (Simplee Digital first-party data, doiexist.ai, n=826).
Key figures
- 68% of US Google searches ended without a click from January to April 2026, according to SparkToro's analysis of Similarweb panel data.
- According to SparkToro, US Google searches ending without a click rose from 60.45% in 2024 on a Datos panel to 68.01% in January to April 2026 on a Similarweb panel, a comparison SparkToro calls imperfect.
- Only 0.34% of US Google searches went to AI Mode from January to April 2026, according to SparkToro's analysis of Similarweb panel data.
- More than half of 826 businesses that ran a free AI visibility audit between April and October 2026, at least 51%, were never named by ChatGPT, Claude or Gemini (Simplee Digital first-party data, doiexist.ai, n=826).
These are two numbers from two different measurements, and read together they point the same way: the click no longer tells a business whether it was seen.
SparkToro's panel puts zero-click searches at 68%
SparkToro counted how many Google searches ended with no click of any kind, except another search. The data comes from Similarweb's desktop and mobile web panel, US only, January to April 2026 (SparkToro, June 9, 2026).
The share rose from 60.45% in 2024 to 68.01% in 2026, according to SparkToro, though its 2024 figure came from a different panel (Datos), so SparkToro calls the comparison imperfect. SparkToro does not publish the panel's size, and searches in the Google mobile app are not included.
The same study found AI Mode is not the main cause yet: only 0.34% of US searches reached AI Mode from January to April 2026, according to SparkToro.
Our audits counted businesses no engine names
Our audits ask ChatGPT, Claude and Gemini the questions a buyer would ask, written to leave the business's name out. We count a business as never named when no answer from an engine names it, as our post on the 826 audits sets out in full. The methodology page gives the full rules.
We counted only questions that left the business's name out: 1,421 of 31,421 counted questions named the business or its domain and were excluded, along with 6 audits in which every question did, which leaves 826 (Simplee Digital first-party data, doiexist.ai, April to October 2026).
| Item | Share never named, at least (see method) (%) |
|---|---|
| ChatGPT, at least 59% | 59% |
| Claude, at least 61% | 61% |
| Gemini, at least 59% | 59% |
| None of the three, at least 51% | 51% |
Source: Simplee Digital first-party data, doiexist.ai, n=826 · Methodology
A missing name costs more than a missing click
Traffic reports undercount the problem. A search that ends without a click can still show a business's name or leave it out, and analytics records neither, so the number worth tracking is being named rather than being clicked.
A business that is named in the answer has been seen, whether or not anyone visits. We think the never-named share should be the first measure of a business's search work, with rankings and clicks still in the picture but second.
The two numbers describe different people
SparkToro's figure is US searchers in early 2026. Ours is 826 businesses that chose to run an audit from April to October 2026, with no country recorded, so the two cannot be multiplied or combined into one figure.
We are unsure how far either holds in Canada.
Never named, the one term we coined here
- Never named means no answer from that engine named the audited business, across all the questions that left its name out.
Source: SparkToro, "In 2026, Less than One Third of Google Searches Still Send a Click", published 2026-06-09, opened 2026-10-19. Methodology
Source: Simplee Digital first-party data, from doiexist.ai audits (826 audits, April 7 to October 19, 2026). Brand-free recompute (view blog.verify_brandless) of the benchmark export, pulled 2026-10-20. The method is on our methodology page. Methodology