The Science of Adaptive Sending: Optimizing Engagement with Probability

I’m going to let you in on a secret: the "best time to send an email" is a total myth. We’ve all seen the articles claiming "Tuesday at 10 AM" is the golden hour, but when we looked at the data while building GetMailer, we realized that advice is actually hurting your engagement.

Why We Stopped "Blasting"

When you send a broadcast to 10,000 people at once, you’re optimizing for an average that doesn't actually exist. We found that by forcing our schedule on our recipients, we were burying our messages. If your subscriber checks their phone at 6 PM, our 10 AM email is already fifty messages deep in their inbox. We knew we could do better.

Our Approach: Adaptive Sending

We built our Adaptive Sending engine to solve this. Instead of asking when we want to send, we built a system that asks when they want to receive. It’s a subtle shift in philosophy that has led to some of our biggest wins in deliverability.

How We Use Probability

Every time someone opens or clicks one of your emails, we record that moment. We’re not just looking for the most recent open; we’re building a probabilistic profile. We look at the distribution of engagement over the week, normalized to their local time. We’ve set our system to require at least five events to start forming a pattern, but the real magic happens once we hit twenty data points.

We also know that life changes. People change jobs, their schedules shift, or they go on vacation. That’s why we apply a decay factor (0.95) to older data. We want our predictions to be fresh, favoring what your users are doing now over what they did six months ago.

The Results We’ve Seen

Since we implemented this probabilistic scheduling, we’ve seen unique open rates jump by as much as 22%. By smoothing out our sending volume, we also found that we were warming up our IPs more naturally, which the major ISPs love to see.

In short: we’ve stopped trying to guess. We’re letting the data tell us when to show up, and it’s making all the difference.