In the simplest terms, the goal of any Send Time Optimization software is for the email to arrive as your contact is looking at their inbox. You want that email to be as close to the top of your contact's Unread list as possible.
Imagine the Wow factor for your customer: your emails consistently land when they are checking their email.
"How did they know?"
But as we all know, the "best" time to send an email will vary from person to person and day-to-day. Sometimes your contact will be checking their inbox at 9 am, and other times they may be checking at 11 pm just to be sure there are no emergencies that can't wait until morning.
Humans are predictably unpredictable.
Many STOs rely on the last open signal received to determine the best time. So, philosophically speaking, if you were to ask these models, "Will the future be like the past?" their answer would be "Yes."
Our answer is "Not Necessarily." And if you think about your own behavior regarding email, do you open it at the same time every day? Maybe, maybe not. Answers will naturally vary from person to person.
So if Motiva's Send Time AI doesn't take your contact's last open signal at face value, what does it use? How does our Send Time AI make predictions about when your contacts will be most receptive to your messaging?
When a new customer signs up for Motiva, we pull all the engagement data collected by Eloqua over the last year and start processing.
Each contact is analyzed individually and then separated into four potential buckets, depending on how much engagement data has been collected on each one.
The first bucket is when we know nothing about a contact. This is the default group for a new contact when they sign up to receive your marketing communications. Our predictions for this group are pretty broad out of necessity.
From there, we use a version of Multi-Arm Bandit modeling. We have a longer article on MAB modeling, but the short version starts with picturing a group of slot machines, aka One-Arm Bandits.
You've heard that one slot machine in a group pays out at a higher frequency than the others, but you don't know which. So you begin testing.
Over time, you notice that one of them is, in fact, paying out at a higher frequency than the others. So naturally you focus your resources on the one that pays out most often.
It's a similar idea with our Send Time AI. Each time a contact flows through a Send Time AI step, we experiment with different send times and monitor the engagement data we receive. Each of these experiments represents a pull of the bandit's arm.
We look for which times of the day and which days of the week result in the least amount of delay between when your contact receives the email and when they engage with it.
Once we start collecting data on a contact, we can narrow our scope. Out of a possible 168 hours in a week, we break it down into a group of 42 possible sending hours.
As the data collection and analysis continues, we can determine the best seven send times and eventually find the very best hour and day for each one of your contacts. Jackpot!
Even better, you don't need to use a Send Time Optimization step to collect this data on your contact. Every time a contact flows through a Motiva step, we continue to gather data, even if it's a Message Testing or Simple step.
Motiva's Send Time AI also works independently of time zones, so there's no need to worry about those when configuring an STO step -- as long as you don't restrict send times.
Instead, Send Time AI looks at when your contact engages, whenever that may be, and builds on that information.
As you may have realized already, restricting the send times for a Send Time Optimization step makes STO much less effective.
Remember, the idea is to have that email land at the best time for your contact. Not everyone checks their email during regular work hours, and we can attest that we've seen some pretty non-obvious behaviors when looking at the data.
Also, consider that you may have contacts within a segment spread across multiple time zones. So if you're trying to restrict your STO step to working hours in one part of the world, it will be in the middle of the night for contacts in another part.
It is best practice to have a full seven days of unrestricted sending windows to ensure that the Send Time AI can work as expected and deliver the best results for you, your team, and your contact.
If your campaign is short on time, it's more effective to use a Simple email step to ensure that your audience receives the email quickly. We'll still be collecting engagement data on those contacts so it's available for situations where STO can flourish.
Send Time Optimization works best with more data. Of course, knowing how much data has been collected can be a bit tricky, but suffice it to say that new contacts — contacts who have just signed up for marketing communications — won't benefit from STO early on.
In fact, it may end up confusing them.
When a new contact signs up for a mailing list or newsletter, they expect to receive a welcome email from that company. However, since our Send Time AI is trying to determine the best send time without any data to work with, it could delay your Welcome email by up to seven days. This is not what your customer expects.
Welcome emails typically have about an 80% open rate. You don't want to waste the opportunity!
Save Send Time Optimization steps for those communications that are not time-sensitive and can run for at least seven days without restrictions. You'll be leveraging all the data you've been collecting to ensure that your contact receives your message at the moment they are most receptive.
Want to see Motiva in action for yourself? Click here to Schedule a Demo!