Apple's Mail Privacy Protection landed in iOS 15 in September 2021. It pre-fetches every tracking pixel in every email opened in Apple Mail, which means open rates from any Apple Mail user are functionally invented. That's roughly half your audience for most consumer brands, and a higher share for any brand with a professional, Apple-device-heavy audience.
Three years on, most agencies are still leading monthly reports with open rates. Sometimes they dress it up as "read rates" or "engagement rates". The number is still invented. The report is still misleading.
This isn't an edge case or a caveat. It is the primary reason your email programme has felt harder to justify since 2022, and it's entirely solvable.
Why agencies still report opens
The honest answer is incentive misalignment. Open rate is easy to improve. Send more emails. Improve your subject lines. Re-send to non-openers. Every one of those tactics increases open rate without necessarily increasing revenue. An agency paid on outputs, sends delivered, campaigns executed, has no structural reason to push clients toward harder, revenue-linked metrics.
There's also a selection problem. Agencies that pitch on revenue per recipient have to be comfortable reporting a bad month. Agencies that pitch on open rates can usually find something that went up this month, even if the programme is declining. The more dishonest an agency is willing to be, the more appealing vanity metrics become.
The fix isn't to shame your current agency. It's to change what you ask for in the monthly report. If you request RPR and attribute it to real order data, the conversation changes immediately.
The metric you should be running
Revenue per recipient. Take the gross revenue attributable to the send (tracked via UTM or first-party purchase data), divide by the number of recipients, not opens, not clicks, not "engaged" subscribers. The resulting number is small. Single-digit pence per email in most consumer programmes. It doesn't matter. Track it weekly and watch the trend.
Revenue per recipient is the closest proxy to "did this send earn its place?" that you have. It collapses deliverability, engagement, offer quality, copy and timing into one number. It can't be gamed by an ESP changing its tracking methodology. It doesn't care whether Apple pre-fetches your pixel. It either made money or it didn't.
Secondary metrics worth keeping: click-to-purchase conversion rate (removes volume noise), unsubscribe rate per send (flags list quality problems), and revenue per active recipient (defined as someone who's clicked in the last 90 days). These three plus RPR give you a dashboard your CFO can read in 30 seconds.
What to do with click rate
Click rate survives as a directional proxy, though not an absolute one. It tells you whether the email made someone curious enough to leave their inbox. It doesn't tell you what they did next, and it doesn't tell you whether that action was worth the slot in the send.
Keep click rate as a hygiene signal: if your click-to-revenue conversion tanks while click rate holds steady, you have a landing page problem, not an email problem. If both fall together, you have a copy or offer problem. The diagnosis requires both numbers.
The attribution problem and how to work around it
Revenue attribution in email is genuinely hard. Most ESPs claim credit for any purchase that happens within 5 days of an email open, which, post-MPP, means they claim credit for purchases made by people who may never have seen the email at all. This inflates attributed revenue and makes lifecycle programmes look better than they are.
The cleaner approach: use a short click-based attribution window (24 hours for promotional sends, 72 hours for lifecycle triggers), and cross-reference against control groups where possible. If you're sending to a list of 50,000 and attributing £15,000 of revenue to the campaign, you should be able to show that a holdout group of 5,000 non-recipients generated materially less revenue in the same window. If you can't, your attribution model is counting revenue that would have happened anyway.
This isn't a reason to distrust the RPR metric, it's a reason to be deliberate about how you define attribution. Set the window. Document it. Hold it constant across all sends so you can compare periods fairly. The number doesn't need to be perfect; it needs to be consistent.
How to baseline RPR
Pull your last 90 days of email sends. For each send, match orders placed within a 72-hour attribution window (adjust to 7-day for lifecycle triggers where purchase cycles are longer). Calculate: attributable revenue ÷ total recipients = RPR. Build a simple table. You'll see outliers immediately, likely a promotional send around a sale event, and a win-back sequence underperforming.
Set your baseline as the rolling median, not the mean, sale-event spikes distort the mean. Then set a 90-day target. We start clients at 1.15× baseline, tightened to 1.3× at six months. Any single send that trails baseline three times in a row without a structural reason (seasonal suppression, list hygiene send) gets killed or rebuilt.
What to say to your stakeholders
The honest conversation is short: open rates were always a proxy for something else. That proxy broke. We now measure the thing the proxy was always pointing at, revenue. Here's what that looks like across the last 90 days. Here's the trend. Here's what we'd need to see to double it in 12 months.
If your agency won't have that conversation, ask them why. The answer tells you a lot about how they measure their own performance.