Possible Inaccuracies in StatCounter’s Safari-Based Version Number Reporting ⇥ webkit.org

Yesterday, I linked to a report from Ed Hardy, Cult of Mac, pointing to a shockingly low iOS 26 adoption rate compared to previous years. Hardy relied on date from StatCounter, which uses web traffic at massive scale to measure all kinds of stuff, including operating system versions.

Given that StatCounter’s data has been similar to Apple’s own reporting of version adoption in previous years, I wrote:

[…] StatCounter’s figures might be off, but it would be shocking if they were out by 40-plus percent. That would point to a serious measurement error that, somehow, did not impact previous reporting.

Well, it turns out there is likely a measurement difference that would not have impacted iOS 18 or before.

Jen Simmons and others who work on WebKit, in September:

Also, now in Safari on iOS, iPadOS, and visionOS 26 the user agent string no longer lists the current version of the operating system. Safari 18.6 on iOS has a UA string of:

Mozilla/5.0 (iPhone; CPU iPhone OS 18_6 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/18.6 Mobile/15E148 Safari/604.1

And Safari 26.0 on iOS has a UA string of:

Mozilla/5.0 (iPhone; CPU iPhone OS 18_6 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/26.0 Mobile/15E148 Safari/604.1

In both, you will notice iPhone OS is set to “18_6” despite only one of them actually running iOS 18.6. If StatCounter was relying on this part of the user agent string for calculating operating system version number, it could be inaccurate. There is still a Safari version number that could be a proxy for the operating system version in the latter part of the user agent string, however. On my iPhone, running iOS 26.3, the relevant section reads Version/26.3 Mobile/15E148 Safari. The iPhone OS string also reads “18_7”, which is also true for users running iOS 26.2.

It is not like StatCounter has no data for iOS 26. It shows traffic from iOS 26.1 and 26.2, indicating it likely updated its tracking metrics. It is possible some of the 18.6 and 18.7 traffic is also iOS 26 — we just do not know how much.

Data from TelemetryDeck seems more robust, and suggests about 55% of iOS users have updated to iOS 26, compared to about 78% of users one year ago running iOS 18. Not as bad as StatCounter’s figures, but still a twenty-point gap between latest version uptake last year and this year.

Thanks to Sam Gross for pointing me in this direction.