The email metrics worth watching (and the ones that lie)
Open rates have been unreliable since 2021. Here is what to measure instead, and the two numbers that decide whether you can send at all.
Most email dashboards lead with open rate. It is the least trustworthy number on the page.
Why open rates lie
An open is recorded when a 1×1 tracking pixel loads. Since Apple's Mail Privacy Protection in 2021, Apple Mail prefetches that pixel for every message, whether or not anyone reads it.
Apple Mail is roughly half the consumer market. So a meaningful share of your "opens" are machines.
Open rate is not useless — it is useful relatively. Comparing two subject lines on the same audience in the same week is fine, because the inflation applies to both. Treating 45% as "45% of people read this" is not.
The two that actually constrain you
Complaint rate. Keep below 0.1%. This is the number that gets you blocked, and there is no gradual warning — placement degrades for everything you send.
Bounce rate. Keep below 2%; 5% puts an Amazon SES account under review. High bounces mean your list is stale or was not collected honestly.
Neither is a growth metric. They are the constraints inside which everything else operates. Watch them daily.
The ones worth optimising
Click rate is harder to fake than opens. It needs a deliberate action, and while some security scanners follow links, the noise is far lower.
Click-to-open tells you whether your content delivered on your subject line. High opens with low clicks usually means the subject promised more than the body gave.
Delivery rate — accepted divided by sent — should sit above 98%. Anything lower points at list quality before it points at content.
Conversion. The only metric that pays for the email. An email with a 12% open rate driving real revenue beats a 60% open rate driving none.
The one nobody tracks
Reply rate, even for automated mail. Replies are the strongest positive signal a mailbox provider can see — it is unambiguous evidence a human wanted this.
Which is a good argument against noreply@. A reply-to address someone monitors is worth more to deliverability than most subject-line tuning.
Segment before you conclude
Aggregate numbers hide the thing you need to know. A 22% open rate might be 40% among people who signed up this quarter and 4% among those who signed up two years ago.
That second group is not underperforming. It is actively damaging you, and the fix is to stop emailing them.