Roughly half of the email opens recorded in 2026 did not happen. Apple Mail Privacy Protection loads the tracking pixel whether or not a human ever looked at the message, and Apple Mail is the single largest mail client in the market. If your outreach process treats an open as a buying signal, it is acting on a number that is about half noise.
What changed
Email open tracking works by embedding a one-pixel image. When the mail client requests that image, the sender records an open. It was never precise, since a preview pane could trigger it and a text-only client would not, but it was directionally useful for years.
Apple Mail Privacy Protection ended that. It routes image requests through Apple's proxy and preloads them ahead of the recipient, so the pixel fires on delivery rather than on reading. The open is recorded, the human may never have seen the subject line.
The numbers
| Measure | Figure |
|---|---|
| Share of all tracked opens attributable to Apple MPP in 2026 | 49.29 percent, across a study of 939 companies, 46 industries and roughly 15 billion emails |
| Share of emails affected | About 40 percent |
| Jump in reported open rates when MPP rolled out | 10 to 30 percent, overnight, with no change in behaviour |
| Reported average open rate, MPP-adjusted | About 19 percent |
| Reported average open rate, including machine opens | About 43 percent |
| Best estimate of genuine human opens | 20 to 25 percent |
The gap between 19 and 43 percent is the same population measured with and without machine opens counted. Neither number is wrong. They answer different questions, which is why open rate benchmarks quoted without a methodology note are close to meaningless.
What this breaks in a sales process
Open-triggered follow-up. A cadence that fires a second touch when someone opens is now firing on a proxy server about half the time. The prospect gets chased for something they did not do.
Open-based lead scoring. Any score with an open weighting is inflated for Apple Mail users and deflated for everyone else, which is a systematic bias, not random noise.
Send-time optimisation. Optimising send time against open timestamps optimises against when Apple's proxy fetched the image, which has no relationship to when your buyer was at their desk.
The read receipt as a confidence signal. The rep who sees five opens and concludes there is interest is reading a number that, roughly half the time, means the message arrived. See what to do when an email is opened but not answered.
What to use instead
Replies, meetings booked and site visits from a named account are all unaffected by MPP, because they require a human to do something. They are also much rarer, which is exactly why open rate was attractive in the first place: it produced a large number early.
The deeper problem is that none of these metrics tell you whether the recipient had any reason to trust the sender. That is the variable that moves reply rate, and it is why send-later and follow-up reminders do not lift it. A warm introduction changes the starting position rather than the timing, which is measurable in reply rate benchmarks in a way that open data never was.
If you still track opens
Two things make the data usable rather than misleading. Segment Apple Mail opens out, which most modern sending platforms can now do, and report the two populations separately rather than blending them. And never let an open alone trigger an action towards the prospect. Use it as weak corroboration of a signal you already have, or not at all.
Frequently asked questions
Are email open rates accurate in 2026?
No. Apple Mail Privacy Protection accounts for about 49 percent of all tracked opens, and it fires the tracking pixel on delivery rather than when a person reads the message. Reported open rates run about 19 percent when machine opens are stripped out and about 43 percent when they are not.
Why did my email open rate suddenly increase?
If it jumped 10 to 30 percent with no change in your sending, that is Apple Mail Privacy Protection. It preloads images through Apple's proxy, so opens are recorded for messages nobody has looked at. The behaviour did not change, the measurement did.
What is a good email open rate now?
The honest answer is that the question no longer has a clean answer. Genuine human opens are probably 20 to 25 percent, but any benchmark quoted without stating whether machine opens are included is comparing two different measurements.
Should I stop using open tracking entirely?
Not necessarily, but never let an open alone trigger an action towards a prospect. Segment Apple Mail opens out and report the two populations separately. Treat an open as weak corroboration of a signal you already have.
What should I measure instead of opens?
Replies, meetings booked and site visits from named accounts. All three require a human to act, so Apple MPP does not affect them. They are much rarer than opens, which is precisely why open rate was appealing in the first place.
