Your Open Rates Are Fiction Now. Measure Email Like It's 2022.

Two full quarters of post-launch data have settled the question, and we are calling it formally. The open rate is fiction now, and email programs still optimizing against it are optimizing against noise.

The mechanism, for readers who filed last fall’s announcement under ‘later.’ Apple’s Mail Privacy Protection shipped with iOS 15 in September and spread at Apple-typical rates since. It routes messages for Apple Mail users through Apple’s servers. Those servers load every tracking pixel, whether or not the human ever opens the message.

Apple Mail clients read for 40 to 60 percent of most consumer and institutional lists. Check your own list mix. The share is usually a surprise in both directions. The practical effect: a large fraction of every send now reports ‘opened’ unconditionally. Open rates are inflated across the industry, and the inflation varies by list makeup. The metric that email marketing optimized, tested, triggered, and reported on for twenty years no longer measures human behavior on much of any list.

The impact studies across the email platforms agree. Our own client data agrees emphatically. The two-quarters mark is the right moment for this post because the mourning period is over. Grieve briefly, rebuild quickly, and re-plumb the automations before they do any more silent damage, which is the section most programs actually need.

The rebuilt stack, from programs we’ve transitioned

Here is the replacement setup, as run across the client programs we have moved since winter. Clicks and downstream conversions become the engagement core. The click is human. No infrastructure pre-clicks links. The conversion is the point: the donation, the sign-up, the application. The outcome the send existed for, measured in your first-party analytics. The reporting order flips to match. Conversions first. Clicks second. Delivery health third. Opens demoted to a rough footnote with a permanent asterisk.

Clicks per delivered message replace open-based benchmarks. The historical open-rate benchmarks, internal and industry, are now apples-to-fiction. The click-through-per-delivered series is consistent across the MPP boundary. Rebasing the program’s benchmarks on it restores a trustworthy trend line, with the September 2021 break marked in every long-run chart.

Testing redesigns around click outcomes. Subject-line tests graded by opens are now graded by noise. The redesigned versions grade by click-through and conversion. Those test the whole promise-to-content chain rather than the envelope alone, and were arguably the better test all along. Send-time optimization rebuilds on the same outcome basis. The general principle underneath: measure what subscribers do, not what infrastructure reports. The metrics that survive every privacy iteration keep being the behavioral ones. Click, convert, reply, unsubscribe. The signals a human actually generates.

The silent menace: automation logic keyed to opens

• Re-engagement and sunset flows, keeping the wrong people: the flow that keeps subscribers who ‘opened recently’ now keeps every Apple Mail address, no matter what. That includes the long-dead ones. List quality rots invisibly and delivery risk builds. The flow needs re-keying to clicks and conversions this quarter.

• Winback and purge logic, wrongly purging and sparing: the inverse flows are now misfiring in both directions. Engaged non-Apple readers who click without loading pixels risk suppression, while phantom-openers sail through. Every list-hygiene workflow keyed to opens is making wrong decisions daily until re-plumbed.

• Open-triggered sequences and branching: the journey logic branching on opens (the follow-up-if-unopened resend, the opened-but-not-clicked nudge) is branching on fiction for the Apple share. Rebuild the branches on click and conversion events. Where no behavioral signal exists, branch on time, honestly.

• The audit method: export every automation, flag each open-conditioned node, re-key to behavioral signals or time, and annotate the change. An afternoon per program, and the highest-priority item in this post. The reporting fiction merely misleads. The automation fiction acts.

The privacy trendline reaches email

This is part of a story five years running. The privacy trendline now runs through email too, and its shape is identical to every prior front. ITP degraded the third-party cookie’s testimony in 2017. The consent regimes constrained collection from 2018 forward. The cookie endgame is retiring cross-site identity. And MPP has now retired the open pixel’s testimony. At every front, machine-reported behavior gets less trustworthy, and human-shown behavior keeps being the measurement that survives.

The programs that barely noticed MPP share one property. They were built on outcome metrics already. Conversions, clicks, revenue per send. List growth net of honest hygiene, with opens as decoration. Their measurement had internalized the trendline before this iteration arrived.

That is the tell, and the target. The durable email program of the privacy era is the one we have described since the deliverability series a decade ago. Permission honestly earned, content worth opening, outcomes honestly measured. Each privacy iteration since has functioned as a stress test that the approach passes and the workaround-dependent programs fail. Fold email fully into the two-front strategy as of now. The next round, the link-wrapping and click-privacy experiments already visible, is pre-announced by the pattern. The outcome-metric setup is, once again, the position that does not care which acronym ships next.

The quarter’s assignment

Compressed. Rebase the benchmarks on click-to-delivery with the break annotated. Flip the reports to conversions-first. Redesign the active tests around click outcomes. And, above all, run the automation audit before another quarter of wrong decisions accrues. The fiction in the reports costs credibility. The fiction in the flows costs subscribers, delivery health, and revenue. Daily, silently, until re-plumbed.

And the same advice, restated once because email keeps proving it at every iteration. The list remains the institution’s most durable owned channel. Its value was never in the open rate. The programs that measure what humans actually do (click, convert, reply, give) have always been measuring the only thing that was ever real.

Apple did not break email measurement. It broke a stand-in the industry had mistaken for measurement. The programs that knew the difference have dashboards this spring that look exactly as trustworthy as they did two Septembers ago. Measure like it’s 2022, because it is, and the next September is already scheduled.