The open rate has been broken since 2021. Here is what to report instead.

Email opens have always been measured with a trick: a 1×1 pixel image that the mailbox has to fetch, which tells the sender the message was rendered. In September 2021 Apple shipped Mail Privacy Protection with iOS 15, and the trick stopped working. When a user has MPP enabled, Apple Mail pre-fetches remote content through a proxy — regardless of whether the person ever opened the message — and strips the IP address while doing it.

So a share of your “opens” are a data centre. Which share depends on your audience's device mix, and you cannot reliably tell them apart at the individual level. Gmail's image proxy has cached and rewritten images since 2013 for different reasons, with its own distortions.

What this actually breaks

• Open rate as a performance metric. It now moves with device mix, not with your subject line.
• A/B tests decided on opens. If the winner is picked by open rate, you are partly testing which variant reached more Apple Mail users.
• Open-based segments. “Opened in the last 90 days” quietly includes people who have not looked at an email in a year, which is exactly the group you were trying to exclude.
• Open-triggered automations. A follow-up sent because someone “opened” may be following up on a proxy.
• Send-time optimisation based on open timestamps. Proxy fetches happen when the proxy feels like it.

The last two are the dangerous ones, because they do not look wrong on a dashboard. They just perform worse than they should, forever.

What to report instead

1. Click rate, as a share of delivered. Imperfect — corporate security scanners click links too — but it requires an actual decision from a human far more often than an open does. Filter known scanner user agents and clicks that land in the first second after delivery.
2. Conversion rate and revenue per recipient. The only metrics your finance team recognises. Revenue per recipient is the honest comparison between a small engaged segment and a large tired one.
3. List health: growth minus churn. New confirmed subscribers, unsubscribes, hard bounces and complaints, plotted together. A list that grows while complaints climb is not growing.
4. Complaint rate per campaign. This is now a hard operating constraint, not a vanity metric: Google enforces a 0.3 % spam rate threshold and recommends staying under 0.1 %.
5. Delivered rate by mailbox provider. Split Gmail, Outlook, Yahoo and the rest. Problems are almost always provider-specific, and an aggregate average hides them.

What opens are still good for

Do not delete the metric. It still has two legitimate uses:

• Relative movement within one campaign over time. If opens collapse on a send to a segment that has been stable for a year, something is wrong — usually deliverability, not content.
• A negative signal. Someone with zero opens and zero clicks over twelve months is genuinely disengaged. The combination is more reliable than either half.

The metric that got more important

Google Postmaster Tools reports your domain reputation, spam rate, authentication results and delivery errors as Gmail sees them — not as your platform guesses. Microsoft's SNDS does the same for Outlook and Hotmail. These are the only sources where the provider tells you directly what it thinks of you, and they are free.

If you have to choose one dashboard to check weekly, it is not the one in your email platform.

A one-page report that holds up

Open rate can stay in an appendix, clearly labelled as indicative. Anyone who insists on it as a headline number is reporting on Apple's proxy infrastructure.