The open rate died in June 2021

by Francis Rozange | Sep 8, 2026 | Email Marketing

Abstract 3D render: a wall of identical indicator lights all switched on at once

On 7 June 2021, Apple announced Mail Privacy Protection for iOS 15, iPadOS 15, macOS Monterey and watchOS 8. The wording of the press release is unambiguous:

“Mail Privacy Protection stops senders from using invisible pixels to collect information about the user. The new feature helps users prevent senders from knowing when they open an email, and masks their IP address.”

Five years later, dashboards still display an open rate, agencies still bill it as a performance indicator, and campaigns are still decided on it.

What Apple documents exactly

The mechanism is described on Apple’s privacy pages, and it is worth reading in detail, because each element breaks a different measurement.

Remote content passes through two relays. “Protect Mail Activity routes all remote content downloaded by Mail through two separate relays operated by different entities.” One knows the user’s IP address but not the content requested, the other knows the content but not the address. The separation is structural, not contractual.

The IP address is masked. “Senders can’t use your IP address as a unique identifier to connect your activity across websites or apps to build a profile about you.” End of recipient geolocation, end of device recognition.

Content is preloaded, whatever happens. This is the decisive sentence: Mail “downloads remote content in the background by default, regardless of whether you engage with the email”.

The consequence Apple does not spell out

Apple nowhere says the open rate is dead. Apple says the sender can no longer know when, or how many times, a message was opened.

The consequence is mechanical, and it is ours to draw, not Apple’s: a tracking pixel that loads systematically, on a delay, independently of any human gesture, no longer measures an open. It measures a delivery, decorated with an arbitrary timestamp.

That distinction between what a source says and what one infers from it deserves to be held, here as elsewhere. It is what separates a fact from a well-informed opinion.

How many recipients are affected

Here the primary source is missing, and that is a problem in itself.

No public body, no standards organisation, no academic work publishes a market share measurement for mail clients. The only figures available come from tool vendors, drawn from their own customers.

The most cited is from Litmus, a vendor of email testing and analytics tools, therefore an interested party: for July 2026 it gives Apple 62.26 percent, Gmail 27.03 percent, Outlook 5.83 percent, Yahoo Mail 2.59 percent. The method is stated: more than one billion opens observed, client identified by the user agent string sent when the pixel loads.

And Litmus itself acknowledges the flaw: “Apple Mail opens impacted by Mail Privacy Protection (MPP) cannot be fully tracked.”

In other words, the measurement of Apple’s market share rests on the very pixel Apple neutralised. The figure has the merit of existing; it does not have the merit of being verifiable, and that is true of almost every figure in this trade.

Take the order of magnitude, not the decimal: a large majority of the installed base is affected.

What is broken, and still in use

The open rate, obviously. But the contamination spreads to everything that depends on it.

Segmentation by “openers” and “non-openers”: it now classifies as active recipients who never saw the message.

Re-engagement campaigns triggered after six months without an open: they no longer go out, because preloading produced an open every time.

Subject line tests decided on opens: they compare two preloading rates.

Send-time optimisation calculated from historical open hours: it optimises the passing time of an Apple relay.

List cleaning based on inactivity measured by opens: it keeps dead addresses and deletes live ones. That is the most expensive of the four, and the remedy lies elsewhere: hygiene based on bounces and on addresses actually existing, not on a manufactured signal.

What remains measurable

The click, with one caveat: security gateways that visit links to analyse them produce clicks that are not clicks. We observe this in the workshop, without being able to point to a published source, and we flag it as such.

The conversion, downstream, on your own site: it is the only signal nobody else manufactures.

The unsubscribe, the bounce, and the complaint. Those three are acts, not inferences. The complaint in particular remains the number that decides your deliverability.

What to do tomorrow morning

Take your last campaign report and calculate the ratio between open rate and click rate. If it is very high, you do not have a content problem: you have an open measurement polluted by preloading.

Then remove the open rate from your dashboards, rather than correcting it. A false indicator adjusted by a coefficient remains a false indicator, and the coefficient will be contested in every meeting.

Finally, rebuild your automation rules by replacing every open condition with a click or conversion condition. The size of your segments will drop, sharply. That drop is the measure of the gap between what you thought you knew and what you knew.

Sources

LaFactory works email on the evidence: headers, DNS records, rejection logs. No open rate promises, ever. Get in touch for a deliverability audit.