Apple’s announcement earlier this week that it was turfing analytics for email opens has made quite a few people mad. I think their concerns are misplaced. They should be less worried that changes to Mail across Apple’s platforms are going to mess up their statistics, and more concerned that their analytics were wrong all along.
Let’s start with Joshua Benton of Nieman Lab:
These images are the only way newsletter senders know if their emails are actually being opened. And that open rate is an important part of how newsletter publishers sell ads — as well as how they judge the relative success or failure of the email.
Email open rates are notoriously unreliable. Some sources will say that open rates are underreported; others will say that they are way too high. That is because open rates are determined by the number of times that a tracking pixel in an email is downloaded. If users have images turned off, it will not be triggered; if a user’s email client automatically goes to the next message when an email is deleted, it may register as the email being opened again and again.
That explains statistics like the ones quoted by Benton:
There have long been ways to block tracking pixels, but they were mostly only used by nerds like me; this is Apple Mail, the dominant platform for email in the U.S. and elsewhere. According to the most recent market-share numbers from Litmus, for May 2021, 93.5% of all email opens on mobile come in Apple Mail on iPhones or iPads. On desktop, Apple Mail on Mac in responsible for 58.4% of all email opens.
Those numbers are crazy high — much higher than Apple’s device market share because Apple users spend a lot more time receiving and reading email than users on Android, Windows, or Linux. Overall, 61.7% of all emails are opened in Apple Mail, on one device or another.2 So even a small change in how it handles email has a huge impact on the newsletter industry writ large.
I find these numbers derived from hundreds of millions of email opens literally impossible to believe. These are from a worldwide report, but iOS market share does not exceed the mid-60% range in any country, and only accounts for about one out of every four smartphones sold. Yet, according to Litmus, the default Mail app accounts for well over nine out of every ten email opens in a mobile email app. Do Android users — of which there are many more of around the world — simply never check their email on their phones? I doubt it. Litmus also reports that Microsoft Outlook, the ubiquitous email client of offices worldwide, has only 40% of the desktop email client market, far below Mail’s 58%. But Windows still has a 76% share of the desktop PC market compared to MacOS’ 17%, a ratio of about 4:1 that is mirrored in Wikimedia’s analytics — and Apple’s Mail app is not available on Windows.
Litmus acknowledges a mesasuring error in its recent stats, stating that Gmail may be underrepresented. It is unclear whether that is all Gmail sources, just Gmail on the web, or just Gmail mobile apps. But mobile email was apparently dominated by iPhones and iPads to a similar degree for all of 2020. These sky-high iPhone figures are not an anomaly in Litmus’ data, and they remain completely disproportionate to actual iPhone market share.
The footnote in the second paragraph is a disclaimer from Benton that email client market share statistics are unreliable. No kidding. But these numbers are so clearly off the mark that I do not think they should be used until Litmus can provide a clear explanation of why iPhones are so overrepresented. Benton’s explanation makes no sense; since when is there a dearth of email-using Windows and Android users? Alex Williams’ theory gets closer to the truth, I think: all of Google’s Gmail clients may simply report “Gmail”. The fact is that these numbers may never make sense because automated email analytics simply are not very good.
In the grand scheme of things, this may be a small point, but it bothers me to see these numbers being cited by Benton and approvingly quoted by Casey Newton. The signal they should send is not that something like 90% of mobile audiences will be unmeasurable, but that these analytics never should have been used by marketers and email administrators.