Apple announced Mail Privacy Protection in June 2021, and it arrived with iOS 15 that September. It’s been part of Apple Mail ever since, across iPhone, iPad, Mac and iCloud Mail, and it changed how much you can trust your open data.

It only affects Apple’s own Mail app. People reading Gmail in the Gmail app, or Outlook in Outlook, aren’t affected. But Apple Mail is popular, so for most newsletters a large chunk of the list is.

What it actually does

When a reader has it switched on, Apple Mail downloads the images and other remote content in an email in the background, as soon as the email arrives, rather than when the reader opens it. It does this through relays that hide the reader’s IP address.

Your ESP records an open when a tiny tracking image loads. So when Apple Mail loads that image automatically, it looks like an open, whether or not anyone read the email. The timing is wrong too, and the IP address belongs to Apple’s relay, so any location you infer from opens is unreliable.

Links aren’t affected. Clicks still come from real people (apart from security scanners, which are a separate problem).

What it breaks

Pretty much anything that relies on opens:

  • Open rates are inflated, sometimes heavily, depending on how many of your readers use Apple Mail.
  • Subject line tests judged on opens are unreliable, because a big share of the “opens” aren’t real.
  • Engagement segments that define active readers as “opened recently” will keep Apple Mail users looking active indefinitely.
  • Resends to non-openers skip people who never saw the first email.
  • Open-triggered automations fire at the wrong time, or for people who never read anything.
  • Countdown timers and live content load when the email arrives, not when it’s read, so they can show the wrong thing.
  • Inactive subscriber cleanup gets harder, because inactive Apple Mail users don’t look inactive.

How to work with it now

A few years on, the practical answer is fairly settled.

Use clicks for anything important. Judge tests on clicks, define engagement by clicks and other actions you can trust, and build re-engagement and sunset rules on “hasn’t clicked in a while” rather than “hasn’t opened”.

Treat Apple opens as unknown, not engaged. Many ESPs can flag likely machine opens. Find out whether yours does and what it’s excluding. If it can’t, at least look at your open rate with that caveat in mind.

Use opens for trends, not truth. Comparing open rates between issues on the same list still tells you something. Comparing them with other newsletters, or with benchmarks from before 2021, doesn’t.

Be honest with sponsors. If you sell ads on open-based numbers, they’re inflated. Good sponsors know this, and reporting clicks alongside opens builds more trust than a big open rate.

It’s not all bad. Plenty of newsletters were leaning too hard on open rates anyway, and this pushed everyone towards numbers that reflect what readers actually do.