We are all familiar with the saying: “Timing is everything”. Each one of us is an individual human being with their own behaviors, habits, and routines; some of us are night people, some of us morning people, some of us work normal office hours, and some work night shifts. Therefore, as individuals we all have widely divergent times when we are most receptive to interactions with a brand.

When it comes to email marketing, what does ‘timing’ even mean? If you think about it, there are two types of inboxes: one personal that you use for your everyday activity, including receiving emails from your family, friends, and your favorite brands.

The other type is your dedicated ad email inbox. Maybe it’s your promotion folder in gmail, or a separate email address you keep just for these occasions. You likely don’t use this account often – maybe once a day. So the question is: what time should these promotions arrive that they are most likely to be read?

Send time optimization (aka STO) has already been here for several years, but the rising import of the above questions mean its relevance is constantly increasing.

Surrounded by Ads

Marketers nowadays are facing a big challenge: attracting the consumer’s attention and engagement.

According to the market research firm Yankelovich Partners, consumers are coming into contact with 5,000 brand messages a day! Whether or not this number is a myth, it’s clear that in today’s society consumers are bombarded with a never-ending amount of ads, as well described by The New York Times: “Anywhere the Eye Can See, It’s Likely to See an Ad”.

In such a reality, digital marketers are continuously ‘fighting’ over the consumer’s attention and engagement. This creates an opportunity in the marketing software industry (expected to be worth $32.4B by 2018) to provide marketers with a proper AI-based STO solution.

This solution will solve the question “when is the best time to interact individually with each consumer?”, and has the potential to massively increase every email campaign’s engagement performance and revenue generation.

Times Square advertisements

The Truth About Send Time Optimization Solutions

Now it’s time to review the type of STO solutions already out there.

Based on research I conducted, some solutions, such as our old send time analysis report, use aggregate account averages; these are insufficient because they are not personalized. Other solutions determine each contact’s average best send time based on historical precedent. Again, they are not adequate solutions as they are limited to analyzing only what has been sent previously, and can’t discover new time slots. They are also static; they might bring an initial boost but they can’t provide continual improvements.

In addition, following a multitude of interviews with marketers, I concluded that the majority send their campaigns at a time based on broad market recommendations, or some A/B tests they ran in the past.

But are those general approaches good enough? Must we still be dependent on averages or A/B tests? Don’t we have technologies that enable us to come up with a solution unique to each customer, rather than sending the campaign to all customers at the same time?

Clearly, analyzing each consumer’s behavior separately and finding the time to interact with them on a personal level would be a better solution than interacting with all consumers at the same time. Big data provides us with access to an increasing number of relevant metrics and parameters, allowing us to really ‘pick the brain’ and understand our customers’ behavior.

But at the same time, it often feels like finding a needle in a haystack. Big data creates massive data dumps, and sorting through these silos can often be time consuming and extremely challenging.

Therefore, in order to address the timing issue properly, we must involve a machine learning algorithm that decides the right time to interact with each customer based on their behavior, an algorithm that also adapts to any behavioral change. To provide a solution that scales effectively, it must also contain an execution layer that will send the campaign to each consumer at the right time. That’s where AI Marketing comes into the picture, by bridging the gap between data science and execution!

Here at Emarsys, a lot of brainstorming and research went into solving this challenge, and we designed an AI marketing-based solution that will make sure your content arrives when it has the best chance to maximize engagement and drive open rates, all at the flip of a single switch. Matching the evolution in each consumer’s behavior via smart algorithms, and delivering every email campaign at the precise moment consumers are most likely to engage (even as browsing habits change), is one of our technology’s greatest advantages.


The method we used to solve the challenge is a Bayesian solution for the multi-armed bandit problem, that can be described with the following:

Imagine you’re in Las Vegas, standing in front of three slot machines, with 20 tokens to use. Each machine pays out differently, but you initially have no knowledge of what kind of payout schedules the machines follow.

Your goal: explore all the different options and maximize your winnings, exploiting the best-performing slot machine. This is known as the exploration vs. exploitation trade-off. The gambler is the algorithm, the slot machines are sending time slots, the tokens are the emails, and the winnings are the open rates.

Slot machines in casino

Send Time Future Thoughts

The timing question is a diverse topic that includes much more than just the best time to send to each customer. Other questions must be answered, such as:

  • What is the best sending frequency (aka marketing pressure)?
  • When is the optimal time to interact across each marketing channel?
  • When is the best time to send each day?

AI solutions are expected to boost consumer engagement by finding the right answer for each of the above on an individualized level.

When it comes to the best send time solution, it must align with a marketer’s business strategy in order to have the greatest impact, maximizing a customer’s engagement and by extension the campaign’s performance as a whole.

As an example, think of a marketer who is planning to promote a sale. If the AI marketing send time solution sends the campaign to each customer at the best time within a 24-hour window. The sale offer must be valid for longer than just a day, allowing the flexibility to take the solution to its maximum potential.

Final Thoughts

Nowadays, marketers are faced with four main questions:

  • When? (Timing)
  • What? (Content)
  • Which? (Channel)
  • Whom? (Segment)

The timing question is just one example of how AI marketing is starting to become an integral part of marketers’ lives, helping them understand each contact as an individual customer, and execute highly personalized and compelling campaigns at scale.

There is no doubt that AI marketing is a promising candidate to revolutionize the marketer’s user experience, completely reshaping their approach to marketing. This will also help to upgrade personalization to beyond-human capabilities, giving marketers more time to define strategy and create content.

Stay tuned for the next chapter to hear more about how Emarsys is breaking new ground with our AI marketing solutions. In the meantime, request a demo to see how your marketing team can begin revolutionizing your send time strategy with our innovative send time optimization solution.

New Call-to-action

Related Articles: