Your open rate is lying to you

In today's Mailbites, we talk about the open rate, the number you've watched, celebrated, and reported to investors for years, and why it quietly stopped meaning what you think it means. The takeaway isn't "track nothing." It's knowing which few numbers still tell you the truth, and we'll get to those.

Now, onto today's story.

The Story

For most of email's history, one number ran the show: the open rate.

You opened your dashboard, saw that a healthy slice of people had opened your campaign, and you had your proof. Proof the subject line landed. Proof the list was alive. Proof, when someone asked, that people actually wanted to hear from you.

The whole thing rested on a tiny trick. Every email carried an invisible one-pixel image. When a recipient opened the message, that pixel loaded from your server, and you logged an open. For years it was a rough but honest headcount.

That quiet little headcount broke in September 2021, and most dashboards still haven't admitted it. That was when Apple switched on Mail Privacy Protection. Now, the moment an email lands in Apple Mail, Apple's servers preload all of its images in the background, the tracking pixel included, whether or not a single human ever looks. The pixel fires regardless. And this isn't only iCloud addresses: if someone reads their Gmail or Yahoo mail through the Apple Mail app, they get counted too.

The first thing that does is fill your opens with ghosts. Apple Mail now accounts for close to half of all email opens, so when your report proudly says 44%, a big share of those "opens" are Apple's machines fetching an image, not people reading your words. Back in 2021, creators watched newsletters leap from a 28% open rate to 55% overnight, with not one extra human involved.

Apple isn't even alone anymore. Gmail, Yahoo, and Microsoft all run their own versions of preloading and pre-scanning. And the metric most people reach for as a backup, the click, is getting muddier as well. Security tools and inbox providers now auto-click links to scan them for threats before the recipient sees anything, logging clicks no human ever made. The phantom traffic has crept from your opens into your clicks.

The real damage, though, sits underneath the dashboard, in everything you wired to those numbers. Re-engagement flows that sunset anyone who "hasn't opened in 90 days" can quietly punish loyal Apple Mail readers whose pixels behave oddly, while machine opens keep genuinely dead addresses looking alive. Send-time optimisation tuned to open timestamps is reading Apple's delivery schedule, not your customers' habits. And an A/B test settled on open rate is being decided by a coin the house controls.

Put plainly: the number you trust most has become mostly noise, and it has been quietly steering the decisions sitting beneath it.

None of this makes the open rate worthless. It still does two small jobs honestly. As a rough pulse, a sudden collapse in opens can flag a deliverability problem worth chasing down. And as a trend line over months, big directional swings still mean something. Just don't hand it the job of proving your email worked, and never let it decide who you keep.

Because the metrics that came through all this clean share one trait: they track something only a human, doing a deliberate thing, can produce. Three are worth moving to the centre.

Conversions and revenue first. Clicks are useful, but strip out the obvious bot traffic and follow them all the way to the thing you actually care about: a sale, a signup, a booking. That is the only number that pays the bills, and no proxy server fakes a purchase.

Replies second, and this one is badly underrated. A machine won't write back. Even a 1% reply rate is real human investment, and mailbox providers read replies as one of the strongest trust signals you can earn, which quietly helps you reach the inbox in the first place. Worth occasionally sending an email that actually invites one.

Then watch the unhappy signals together: unsubscribes, complaints, and bounces. Providers now weigh those negatives more heavily than any positive, so a slow climb there means your program is degrading even while your open rate beams back at you. It is the early warning your inflated opens are busy hiding.

One practical move ties it all together: rebuild your re-engagement and sunset flows on clicks and purchases instead of opens. Do that, and in a single stroke you stop trusting machines and stop punishing real readers.

The open rate didn't set out to lie. The world changed underneath it, and the number kept reporting as though nothing had. Count the things a machine can't do for your customer, and you start seeing your actual customers again.

See ya next week,

Ani