Are Email Open Rates Accurate? Apple Mail Privacy and the Measurement Problem

No. Open rates have been a distorted metric since 2021, when Apple's Mail Privacy Protection (MPP) started pre-loading tracking pixels for Apple Mail users — registering an "open" whether or not the email was ever read. Gmail's image proxy adds further noise. Opens remain useful as a deliverability signal, but as an engagement metric they overcount; replies are the number that can't lie.

How open tracking works — and how it broke

An "open" is a 1×1 transparent image (the tracking pixel) loading from the sender's server. That mechanism assumes images load when, and only when, a human views the email. Two providers broke that assumption:

• Apple Mail Privacy Protection (iOS 15+, 2021): Apple Mail fetches remote content — including tracking pixels — through Apple's proxies in advance, regardless of whether the user reads the message. Every MPP-protected recipient can register as an "open." With Apple Mail's large share of B2C and meaningful share of B2B reading, this alone inflates open rates substantially.
• Gmail's image proxy: Google has long cached images through its own servers, which masks the recipient's device and location and can decouple pixel loads from actual reads in prefetch scenarios.

The result: a reported 60% open rate might be 40% humans and 20% robots — and you cannot tell which from the dashboard.

Open rates still work as a deliverability signal

Open rates show direction, not truth. Because the inflation sources are roughly constant within your audience, changes in open rate still carry signal: a sudden drop usually means a deliverability problem, not a measurement artifact — the diagnostic sequence in why your open rate is dropping still applies. Cross-campaign comparisons on the same list remain informative. Absolute numbers ("we get 70% opens!") are marketing fiction.

What to measure instead of opens: replies, bounces, meetings

• Reply rate — a reply requires a human. It's the cold email KPI that survived the privacy era intact; benchmarks in what is a good reply rate.
• Positive reply rate — replies minus the "remove me" responses; the number that actually predicts pipeline.
• Bounce rate and spam-complaint rate — the health metrics mailbox providers themselves judge you on.
• Meetings booked per 100 sends — the business metric the whole funnel exists for.

In Sales.co platform data (5M+ cold emails, 2025–2026), the campaigns that look best on opens and the campaigns that book the most meetings are frequently different campaigns — subject-line bait inflates the first and quietly suppresses the second. Optimizing for the wrong metric isn't neutral; it's negative.

There's also a delivery-side reason to rethink the pixel itself — covered in do tracking pixels hurt deliverability. Platforms like Sales.co report reply-centric analytics by default, so campaign decisions ride on numbers a privacy proxy can't fake.