If your email open rate jumped in late 2021 and has stayed suspiciously high ever since, your subject lines didn't suddenly get better. Apple changed how its Mail app loads images, and in doing so broke the mechanism that open tracking depends on. Years later, many teams still report open rate as a headline KPI, still run subject line tests on it, and still have automations that fire when someone "opens".
The scale of the problem depends on your audience, but for many lists it's large. Litmus's email client data for July 2026, based on over a billion opens, put Apple's share of opens at 62.26%, and its page estimates that Mail Privacy Protection affects roughly 55–60% of all email opens. When the majority of your "opens" may be generated by a privacy proxy rather than a person, open rate tells you very little about attention.
This guide explains what Apple Mail Privacy Protection (MPP) does, how to estimate the distortion in your own data, what your email platform can and can't detect, which metrics to use instead, and how to repair automations, A/B tests, list hygiene rules and reports that still lean on opens.
What Mail Privacy Protection actually does to a tracking pixel
Email platforms count an open by embedding a tiny, unique image (a tracking pixel) in each email. When the recipient's mail app downloads that image, the platform's server logs an open, along with the IP address, which reveals approximate location and sometimes device details.
Apple announced MPP at its developer conference in June 2021 and shipped it with iOS 15 and macOS Monterey that September. When a user turns on "Protect Mail Activity", the Mail app downloads remote content privately in the background, through Apple's own relays, regardless of whether the user ever reads the message. Apple's privacy page describes it this way:
“Protect Mail Activity helps protect your privacy by preventing email senders”
— Apple, Mail Privacy Protection & Privacy
Apple's page adds that Mail routes remote content through two separate relays, so no single party can link the user's identity to the content, and that the user's IP address stays hidden even if they turn Protect Mail Activity off. The consequences for senders:
- Opens are recorded for emails nobody read. The pixel loads when Apple fetches content, not when a human looks.
- Open timestamps are wrong. They reflect when Apple fetched the content, which breaks send-time optimisation based on opens.
- Location and device data from opens are unreliable, because the IP belongs to Apple's relay, not the reader.
- Clicks still work. A click requires a human action, so click tracking is unaffected.
It applies to Apple's Mail app, regardless of which email provider the address belongs to. A Gmail address read in Apple Mail with MPP on is affected; the same address read in the Gmail app isn't.
How large the distortion is on your own list
Industry-wide estimates are useful context, but what matters is your audience. A B2B list of Outlook users will be barely affected; a consumer list skewed towards iPhone users will be heavily affected. Most platforms can show the share of opens flagged as Apple privacy opens.
| Email client | Share of opens, July 2026 (Litmus) |
|---|---|
| Apple (includes MPP opens) | 62.26% |
| Gmail | 27.03% |
| Outlook (desktop) | 5.83% |
| Yahoo Mail | 2.59% |
| Google Android | 1.45% |
| Outlook.com | 0.42% |
Note the circularity: because MPP generates opens for every delivered message, it inflates Apple's share of opens too. That's one reason market-share figures based on opens should be read with care. Litmus's own explanation is blunt:
“Apple Mail users are able to turn on a feature that will disable open tracking.”
— Litmus, Email Client Market Share
It also makes old benchmarks hard to use. Semrush's 2024 guide to open rates, for example, cites an average open rate of 34.23% from Mailchimp data. Whether a figure like that includes MPP opens depends on how and when it was calculated, so comparing your number with any published average tells you little unless both use the same method.
Spotting machine opens: what your platform can and can't tell you
Platforms can identify opens that came through Apple's proxy, because the request comes from known Apple infrastructure with a recognisable pattern. What they can't know is whether a human also read the email afterwards. That distinction is important: an "Apple privacy open" is not a fake open, it's an unknown.
| Platform | What it offers (per its own documentation) | What it doesn't do |
|---|---|---|
| Mailchimp | A checkbox to exclude MPP-affected open metrics in the marketing dashboard and custom reports for emails sent on or after June 22, 2024; guidance to rely on clicks, conversions and revenue | Can't determine location, device or client for MPP opens |
| Klaviyo | An Apple privacy open flag on opened-email events (reliable from November 20, 2021), optional Apple privacy open metrics in custom reports, and an option to omit those opens from attribution | Doesn't change how standard open rates are calculated in pre-built views |
| Most other major platforms | Some form of MPP flag, filter or adjusted open rate | Check your platform's help centre; labels and methods differ |
Klaviyo's decision shows why you can't assume your dashboard has been adjusted for you:
“Klaviyo has opted to continue reporting open rates as-is in pre-built analytics views”
— Klaviyo, How Klaviyo email marketing users prep for Apple iOS 15
So find out exactly which number your headline "open rate" is: all opens, opens excluding Apple privacy opens, or something else. Write the definition into your reporting template.
Metrics that carry the weight open rate used to
Open rate used to do three jobs: measure interest in the subject line, measure audience engagement over time, and flag deliverability problems. Each job now needs a different metric.
| Job | Old metric | Better metric now | How to calculate |
|---|---|---|---|
| Did the email interest people? | Open rate | Click rate (unique clicks ÷ delivered) | Exclude clicks from security scanners where your platform flags them |
| Did it drive the business outcome? | Open rate, click-to-open rate | Conversion rate and revenue per recipient | Conversions or revenue attributed to the email ÷ delivered |
| Is the audience engaged over time? | Opens in the last 90 days | Clicks, site visits, purchases or replies in a set window | Share of list with any of these in 90 or 180 days |
| Is the email reaching inboxes? | Sudden open-rate drop | Non-Apple open rate, inbox placement tests, spam complaint rate, bounce rate | Open rate for non-MPP recipients only; Postmaster Tools for Gmail |
| Are people unhappy? | (Rarely tracked) | Unsubscribe rate and complaint rate per send | Unsubscribes or complaints ÷ delivered |
Click-to-open rate deserves a special warning. It divides clicks by opens, so an inflated denominator makes it look worse than it is, and the size of the distortion varies with the Apple share of each send. Either drop it or calculate it on non-Apple opens only.
For revenue attribution, make sure every link in your emails carries consistent UTM parameters so GA4 and your CRM agree on what email drove. Our set guide on passing UTM data into your CRM covers the plumbing.
Repairing automations that trigger on opens
Automations are where MPP does real damage, quietly. Mailchimp's guidance sums up the effect: open-triggered flows reach a larger audience than intended, while "not opened" triggers reach a smaller one. A re-send to non-openers, for example, skips most Apple Mail users because they all appear to have opened.
- List every automation, branch and segment that uses opens. Look for "opened", "did not open", "opened in last X days", engagement scores and send-time optimisation.
- Replace "opened" triggers with click or site-activity triggers. "Clicked any link in email 1" or "visited the pricing page" are stronger intent signals anyway.
- Rethink "did not open" re-sends. Either drop them or base them on "did not click", and cap the number of re-sends to avoid annoying people who read but didn't click.
- Fix branch logic in nurture flows. A "highly engaged" branch based on opens will be full of Apple users who may never have read anything.
- Check send-time optimisation. If your platform optimises on open times, ask whether it excludes Apple privacy opens. If it doesn't, treat its recommendations sceptically.
- Re-baseline the flows. Mailchimp suggests pausing open-triggered automations to preserve historical metrics before you rebuild. Record the before and after so stakeholders understand why numbers change.
Our guide to email automation flows has trigger ideas that don't rely on opens, and abandoned cart emails are a good model: they trigger on behaviour, not on email engagement.
Testing subject lines when opens can't be trusted
Subject line A/B tests decided on open rate are now partly decided by Apple's proxy, which opens both variants equally. The result is noise that looks like a signal. Better approaches:
- Use click rate as the winning metric. It's the closest remaining measure of whether the subject line attracted the right reader.
- Or use non-Apple open rate, if your platform can segment it, accepting that it represents only part of your audience.
- Plan for larger samples. Click rates are far lower than open rates, so detecting a difference takes many more recipients. Our set guide on A/B test statistical significance shows how to estimate the sample you need.
- Test fewer, bigger ideas. Instead of testing two near-identical subject lines, test genuinely different angles (benefit versus curiosity versus urgency) where a difference is detectable.
- Accumulate learnings across sends. Tag each send by subject line style and compare click rates over a quarter rather than relying on one test.
Our guides to writing email subject lines and the set guide on email copywriting cover the creative side.
Defining an engaged subscriber for list hygiene
List hygiene has higher stakes than reporting. Mailing people who never engage harms deliverability, and Gmail's sender guidelines tie inbox placement to user complaints. Google advises keeping the spam rate reported in Postmaster Tools below 0.10% and warns senders to "avoid ever reaching a spam rate of 0.30% or higher". Since February 1, 2024, bulk senders to Gmail (around 5,000 or more messages a day) have also had to authenticate with SPF, DKIM and DMARC and support one-click unsubscribe.
If your "engaged" segment is based on opens, it now includes many people who haven't read anything in months, and your sunset policy (removing or pausing long-term inactive subscribers) never triggers for them. A more reliable definition:
- Engaged: clicked, replied, purchased or visited the site from email in the last 90 days.
- Uncertain: no clicks or other activity, but Apple privacy opens recorded. Treat as unknown, not active.
- Inactive: no clicks, activity or non-Apple opens in 180 days (adjust to your sending frequency and buying cycle).
Send a re-permission campaign to the uncertain and inactive groups asking them to click to stay subscribed, then pause those who don't. You'll lose some names, but the people you keep are real readers, and your complaint rate will fall. Our guide to email deliverability covers authentication and inbox placement in more depth, and the set guide to RFM segmentation shows how to combine email activity with purchase data.
Worked example: reporting a newsletter honestly
Hypothetical example with illustrative numbers.
A retailer sends a weekly newsletter to 20,000 delivered addresses. The platform reports 13,370 unique opens: a 66.9% open rate, which the team has been quoting as proof of a highly engaged list.
Breaking it down: the platform flags 11,000 recipients as Apple privacy openers, and almost all of them register an open on every send. Of the other 9,000 recipients, 2,700 opened: a non-Apple open rate of 30%. Unique clicks are 520, a 2.6% click rate. The email drove 41 orders worth £2,870, or about £0.14 revenue per delivered email.
The honest report replaces "66.9% open rate" with four lines: click rate 2.6% (trend versus last month), revenue per recipient £0.14, non-Apple open rate 30% (as a deliverability check), and unsubscribe plus complaint rate. The team also discovers that its "engaged in 90 days" segment contains 15,800 people, but only 4,100 have clicked or bought in that time. Moving the remainder into a re-permission flow cuts sends by a fifth without touching revenue.
Questions email marketers ask about open rates after MPP
Is open rate completely useless now?
Not completely. It's unreliable as a measure of human attention for Apple Mail users, but non-Apple open rates and sudden drops in total opens can still flag deliverability problems. Treat it as a diagnostic, not a KPI.
Can I tell which opens are from Apple Mail Privacy Protection?
Many platforms flag Apple privacy opens. Klaviyo added a flag in 2021, and Mailchimp lets you exclude MPP-affected open metrics for emails sent from June 22, 2024. None can tell you whether a human also read the email.
What should replace open rate in subject line tests?
Click rate or a downstream conversion, with a sample large enough to detect a difference.
How should I define an inactive subscriber now?
Use clicks, purchases, site visits and replies over 90 or 180 days. Treat Apple privacy opens as unknown.
Does MPP affect click tracking?
No. Clicks still work. Location and device data tied to opens become unreliable because the IP address is hidden.
Rebuilding your email scorecard this month
Start by finding out exactly what your platform's open rate includes, and how much of your audience is flagged as Apple privacy openers. Then audit every automation and segment that uses opens, switch your A/B tests to clicks, and redefine "engaged" around actions. Finally, rewrite your monthly report around click rate, revenue per recipient, complaint rate and a non-Apple open rate as a deliverability check.
If you'd like help rebuilding flows and reporting, our email marketing team can audit your account and set up metrics that reflect real engagement. Get a quote to start, or read our newsletter strategy guide first.
Further reading on email open rates
- Google Sheets for Marketers: Formulas and Templates That Save Hours
- GA4 BigQuery Export: When You Need It and How to Start
- GA4 Data Thresholds and Sampling: Why Your Numbers Look Off
Sources for this email open rate guide
- Apple: Mail Privacy Protection & Privacy
- Mailchimp: About Apple Mail Privacy Protection
- Litmus: Email Client Market Share (July 2026 data)
- Klaviyo: How Klaviyo email marketing users prep for Apple iOS 15 (updated January 2022)
- Gmail Help: Email sender guidelines
- Semrush: Email Open Rates (April 2024)
