You’ve probably seen the feature in your email platform: “Send Time Optimisation” or “Predictive Sending” or some variation that promises to deliver your newsletter at the exact moment each subscriber is most likely to engage.
Sounds brilliant, doesn’t it? Machine learning analysing individual subscriber behaviour, sending at their personal peak engagement window. Set it and forget it.
Except for most newsletters, it’s actively hurting performance.
The problem with optimising for individuals
Send time optimisation works by fragmenting your send over hours—sometimes over an entire day. Your newsletter trickles out subscriber by subscriber, based on when each person historically opened emails.
This creates three immediate problems. First, your content ages differently for different segments of your list. Someone receiving your newsletter at 6am gets fresh links and timely commentary. Someone getting it at 9pm sees content that’s potentially stale, with conversations already underway in replies and social channels they can’t join.
Second, you lose the momentum of a coordinated launch. When everyone receives your newsletter within the same hour, you get concentrated traffic, clustered replies, and genuine conversation. Spread that same audience across twelve hours and everything diffuses into silence.
Third—and this one’s subtle—you’re optimising for yesterday’s behaviour, not tomorrow’s. The algorithm looks at historical opens to predict future engagement. But subscriber habits change. The person who used to check email at 7am might have switched jobs, moved time zones, or simply changed their routine.
What the data actually shows
Multiple studies of email performance data reveal something most operators miss: the difference in open rates between “optimal” and “suboptimal” send times is typically 2–5%. That’s real, but it’s small.
What matters far more? Day of week consistency. Subscribers who know your newsletter arrives every Tuesday at 10am develop a habit. They anticipate it. Some even structure their morning around it.
When you optimise send times individually, you sacrifice this habitual behaviour for a marginal improvement in immediate opens. You’re trading long-term retention for short-term metrics.
The subscribers who genuinely want your newsletter will open it whether it arrives at their “predicted optimal time” or not. The subscribers who are marginal—the ones send time optimisation is designed to capture—probably weren’t going to engage meaningfully anyway.
When optimisation actually works
There are scenarios where send time optimisation makes sense. If you’re running a large promotional programme with multiple sends per week and your primary goal is transaction completion, the individual-level precision can move the needle.
If you have a genuinely global audience spread across eight or more time zones, some degree of send time variation is necessary. But even then, consider batching sends into two or three deliberate time windows rather than continuous optimisation.
For most operator-to-reader newsletters, though—the kind where you’re building a relationship, establishing a voice, creating a space for your subscribers—consistency beats optimisation every time.
What to do instead
Pick a specific day and time. Send every edition at that exact moment. Make it part of your brand: “In your inbox every Thursday at 9am GMT.”
Test different times if you want, but test them properly. Send at 9am for a month, then 2pm for a month. Look at opens, yes, but also look at replies, forwards, and unsubscribe rates. Look at the quality of conversation your newsletter generates.
You’ll probably find that consistency matters more than perfect timing. Your most engaged subscribers will adjust to your schedule. Your least engaged subscribers won’t be saved by an algorithm.
If you found this useful, reply and tell me what you’re currently optimising for—or what you’ve stopped optimising entirely. I read every response, and the best insights often come from what we’ve deliberately chosen not to do.