What is Send Time Optimisation?
Send Time Optimisation (STO) is a method of analysing historical campaign engagement data to determine the ideal send time of the next campaign, at an individual level. It’s powered by machine learning.
Why is it so important?
If you’ve ever had any involvement in an email marketing campaign then I would confidently wager a small fee that you’ve Googled this exact question;
“When’s the best time to send an email”?
I have too. And every article I read dangled the carrot of a “data-based” answer, aggregating open rates across the various reports that have come out in the last decade. None of those articles ever delivered on it.
So full disclosure, I’m not going to string you along with the promise of revealing email marketing’s golden hour. It doesn’t exist.
If you were hoping that the conclusion would be something as prescriptive as Wednesday. 12.47am. then you will be disappointed.
There is no easy, one-size-fits-all answer to the question. And I think you probably knew that.
In any case, the more interesting way of framing the question is “when does the customer want to read the email”?
Answering this for each contact in your email list is the goal of Send Time Optimisation.
How can Send Time Optimisation improve email campaigns?
What we’re focused on in this article is how you can develop email marketing campaigns that are personalised to the individual recipient.
You might already be doing that in terms of the message, using dynamic content or segmenting your lists to create relevant experiences. That’s brilliant, and will certainly yield positive results.
But what about personalised send times? That’s what the leading email marketers are aiming for with Send Time Optimisation.
To do this at scale, they’re utilising a type of artificial intelligence called machine learning.
What is machine learning in marketing?
For a long time, it felt like artificial intelligence was little more than a pipe dream for the average marketer. It was hard to see how it would ever play a day-to-day role.
Things have changed.
We are now seeing the first wave of practical, actionable A.I. that marketers can really lean on.
Machine learning is a method of predictive modelling, using historical data to forecast the best action a brand should take to increase conversions. The system learns from the success and failures of previous campaigns and makes recommendations.
The more data the system has to learn from, the more accurate the predictions become over time.
At Xtremepush, we use machine learning to help our clients optimise their campaigns in a couple of ways;
1) Identifying the best channel to connect with customers on (email, SMS, push notification etc)
2) Predicting the optimal send time for each campaign in order to maximise engagement rates.
Our platform analyzes all of the rich campaign data available, looking for reliable patterns of behaviour. It doesn’t rush into snap decisions, as this could actually be detrimental to your campaigns.
We’ve long been advocates of one to one marketing and know the value it can bring to any brand.
Send Time Optimisation is the next phase of personalisation.
How is Send Time Optimisation different from A/B testing?
I’m going to assume that right now, you aren’t using some level of Send Time Optimisation as we’ve just described it.
Maybe your service provider doesn’t offer it. If that’s the case, you might want to start thinking about switching email supplier.
Odds are, however, you’re A/B testing your send times, hoping to find the sweet spot. That’s all good and no doubt you’ve seen your engagement rates improve accordingly.
But there’s a very obvious limit to what A/B testing can do in this instance. No matter how many tests you run, the “winner” can only ever be picked from the limited list of options you’ve put together. Once the test has been run, you need to set up another to keep optimising.
And ultimately, it’s still boxing customers into groups rather than treating them as individuals. Even in a best-case scenario, the winning send time is only right for some (or hopefully most).
Send Time Optimisation is a better way of tackling the issue because it’s not searching for evidence to support a hunch, it’s objectively looking for the best possible solution.