Open rate is the number of emails recorded as opened divided by the number delivered. Opens are detected when an invisible tracking image in the email loads. Since Apple introduced Mail Privacy Protection (MPP) in 2021, Apple Mail can preload those images whether or not the person reads the email, which inflates opens and makes open rate a weak signal on its own.
How it is calculated
Platforms differ on whether they count unique or total opens; unique opens are the usual default.
Open rate = Unique opens / Delivered emails
Example: 1,800 / 6,000 = 30%What Mail Privacy Protection changed
With MPP enabled, Apple’s servers fetch remote content, including tracking pixels, through a proxy, often soon after delivery. The sender sees an open from an Apple IP address with no link to whether a human read it. Other clients and security filters also prefetch images. Many email platforms now flag or filter suspected machine opens, but none can do so perfectly.
Open-based automations suffer most: “resend to non-openers”, open-triggered follow-ups, and subject line tests judged by opens all become less trustworthy.
What to use instead
- Click rate and click-to-open rate, while watching for bot clicks from security scanners.
- Replies, conversions and downstream product or pipeline outcomes.
- Open rate trends on non-Apple segments, if your platform reports them.
- Unsubscribe and complaint rates as a check on list health.
Example
Re-basing a subject line test
Hypothetical company Hazelport used to pick newsletter subject lines by open rate. After MPP, both variants show similarly high opens. The team switches to judging tests by clicks per delivered email and replies, using a larger sample.
Common mistakes
Suppressing “non-openers” blindly
With machine opens in the data, some people marked as engaged never read anything, and some real readers block images. Combine signals before cleaning a list.
Questions
Not entirely. Sudden drops can still point to deliverability problems. It just should not be your main measure of success.
