How Mail Privacy Protection Broke Email Open Rates

If your email open rate jumped in late 2021 and never came back down, you are looking at how Mail Privacy Protection affects open rates. Apple's feature, introduced with iOS 15, changed what an "open" means — and it changed it for a large, unknowable share of your list.

This guide is written for the sender. It explains why opens inflated, what the number still tells you, and what to measure instead. The short version: stop treating open rate as a behavioural signal, and move your reporting and your automations onto clicks, replies, and conversions.

The short answer: why opens inflated#

Mail Privacy Protection (MPP) is a setting in the Apple Mail app. When it is on, Apple downloads the remote content in your email — images, and the invisible tracking pixel that measures opens — in the background, on its own infrastructure, before the recipient ever looks at the message.

Apple is explicit about the effect. In its 2021 developer session introducing the feature, Apple said: "You'll see your emails as being opened, regardless of if the user read it or not." Its privacy documentation says Mail downloads remote content in the background by default, regardless of whether you engage with the email.

So the pixel fires without a human. For every subscriber who reads your mail in the Apple Mail app with the setting on, the "open" is a machine event and the open timestamp is meaningless. Multiply that across a list and the aggregate open rate stops measuring attention.

The setting is opt-in, but it is prompted prominently the first time someone opens Mail on iOS 15 or later, and it covers the Apple Mail app on iPhone, iPad, and Mac. That makes it common enough — and invisible enough to the sender — that you have to assume a meaningful slice of your list is affected and design your reporting around it.

Before you start#

Before you rebuild your reporting, get three things straight. They stop you from over-correcting and throwing away the parts of the number that still work.

• It is the app, not the address. A subscriber on a Gmail or Outlook address who reads your mail in the Apple Mail app with MPP on is affected; the same person in the Gmail app is not. You are segmenting by mail client, not by email domain.
• You cannot see the adoption rate from outside. There is no reliable public figure for how many of your subscribers turned it on, and it varies by audience. Do not import a benchmark — pull your own ESP's client data.
• The signal is degraded, not gone. Open rate still works as a coarse list-level trend and as an on/off deliverability check. It just no longer works as a per-recipient behavioural signal or an automation trigger.

Steps: how to fix your measurement#

1. 1

   Demote open rate from KPI to health check

   Stop reporting raw open rate as a campaign success metric. Keep it on the dashboard as a trend line and a deliverability tripwire — a sudden collapse still means something — but do not judge a send by it.

2. 2

   Segment Apple Mail opens out

   Most ESPs now flag machine or MPP opens using the requesting client and the open timing. Build a segment that excludes them so your "engaged" audience is people, not proxies. If your ESP does not expose this, filter opens that land within seconds of send.

3. 3

   Make clicks your primary engagement metric

   Report click rate and click-to-open rate instead. A click needs a human to tap a link; MPP does not click for anyone. Track unique clicks rather than raw clicks to cut noise.

4. 4

   Add reply and conversion tracking

   For lifecycle and sales mail, a reply or a completed action — a purchase, a signup, a booking — is the signal that survives every privacy change. Wire these into reporting as the metrics that actually map to revenue.

5. 5

   Rebuild open-triggered automations

   Re-target "did not open" flows, sunset sequences, and re-send logic on clicks or on the absence of any human action, not on the pixel. An MPP open makes an unengaged subscriber look active and quietly poisons these journeys.

6. 6

   Judge the trend, not the single send

   Compare like-for-like sends over weeks. A list-level shift in clicks and conversions tells you more than any one campaign's opens, and it is immune to how many recipients toggled MPP this month.

Platform differences#

The Apple Mail app is the biggest source of inflated opens, but it is not the only client that touches your tracking pixel. Here is how the major ones behave and what still works in each. Behaviour changes over time — treat this as a starting point and verify against your own ESP data.

This table is accurate as of August 2026. Image-proxy behaviour at Google and Microsoft has changed before and will change again, so check your ESP's client breakdown rather than trusting any static summary indefinitely. The one durable rule underneath all of it: a click needs a person, and an image load no longer does.

What the numbers look like before and after#

The trap is treating a rising open rate as rising interest. After MPP, an increase can simply mean a larger share of your list reads in Apple Mail — more machine loads, not more readers. The same 40% open rate can hide two completely different audiences.

What to do when it doesn't work#

You switched to clicks and something still looks off. These are the usual culprits and the fixes.

• Clicks are also inflated. Security gateways and link-protection scanners — common on corporate Microsoft 365 and Google Workspace mail — pre-click links to vet them, firing click events with no human behind them. Filter known bot user-agents, count unique human clicks, and be suspicious of clicks that hit every link within a second of delivery.
• You cannot segment MPP opens. If your ESP does not label them, approximate: opens recorded within a few seconds of send, or from Apple Mail clients, are almost all machine loads. Some ESPs let you exclude Apple MPP opens directly — check the client or engagement settings.
• Automations still misfire. If a re-send-to-non-openers flow keeps skipping real readers or mailing dead addresses, it is still keyed to the pixel. Rebuild the entry condition on clicks or on any human action inside a window.
• Deliverability is slipping. This is the dangerous one, because the cause is the inflated opens themselves.

A faster way — and where AI Emaily fits#

There is no setting that makes open tracking honest again. The durable fix is the one above: measure clicks, replies, and conversions, and read open rate only as a rough list-level trend. That work is continuous — every send, forever.

It helps to see which way the ecosystem is moving. Opens broke because mail clients now load or block tracking pixels on the reader's behalf, and more clients are doing it, not fewer. AI Emaily is one of them: it is an AI email client that blocks remote tracking pixels by default and sandboxes links, so a recipient using it does not report an open to you at all. We build AI Emaily.

It is a mail client on the receiving side — not an ESP, an analytics platform, or a deliverability tester — so it will not fix your campaign reporting. For that, use your own ESP's MPP-aware metrics. What it tells you as a sender is simpler: betting your program back on the pixel is betting against the trend.