February 17, 2017
We are very excited today to announce the release of Optimail — the newest and best way to automatically send, manage, and optimize your drip email marketing campaigns, built by our small team at Strong Analytics.
You might be thinking, Lots of companies offer email marketing automation products… what makes Optimail different? In a nutshell, while other products give you tools to build campaigns, run experiments, and tweak your campaigns, Optimail uses cutting-edge AI to optimize and manage your campaigns for you — for the same price that our competitors charge you to do the work yourself.
If the first wave of marketing automation was automating message sending and customer tracking, Optimail is leading the next wave by automating the continuous optimization of those campaigns to make your campaigns more effective every day.
But Optimail’s benefit isn’t only that it optimizes your campaigns for you (not that this isn’t a big deal!). It’s also what it optimizes that makes it unique. Most email marketing platforms — if they offer any optimization tools at all — try to optimize your email open and click rates using A/B tests. The problem is that the goal of an email campaign isn’t to have people open and click the emails. It’s to to influence customers’ behaviors, for example, you may want them to convert to a paid subscription, share your app with their network, or login and use your product more. This is what Optimail learns to optimize. It learns which sequences of emails work for which kinds of customers to drive them towards specific goals you define while building your campaign.
And, most importantly, it actually works! For an initial test of our algorithm, we worked with a company that offers a mobile app and a web app in the person productivity space. We randomly assigned new users either to the existing drip campaign (designed by a marketing expert) or an Optimail drip campaign, without any constraints on when/to whom emails could be sent. After just a few weeks, users in the Optimail campaign were engaging 25% more with the application, while Optimail was sending 20% fewer emails! This was huge validation for the concept of Optimail. Furthermore, it brought to light a previously unconsidered benefit: Optimail doesn’t just optimize campaigns for the companies who use it, it makes the campaigns less annoying for the customers who receive the emails (i.e., if spamming lots of emails leads to disengaged, annoyed users, Optimail will learn to avoid that, as it did in our test).
Although the technology behind Optimail is rather complex, we wanted the process of building and monitoring your campaigns to be as simple as possible.
All you need to do to get your first campaign off the ground is to:
As you dive into Optimail, you’ll also find some additional functionality that can help-fine tune your campaign. For example, you can constrain the strategies Optimail will try using email rules, customize email delivery windows, and provide a rate-limit on the amount of emails Optimail can send in a given time period.
Finally, as it learns, Optimail provides you with unique insights into your email campaign and the algorithm underlying its strategy. You’ll be able to dive deep into the strategies it thinks are best for different customer segments and see visualizations (e.g., cohort analyses) to quantify Optimail’s impact on the success of your campaign.
Optimail is built on a suite of algorithms that have all been designed from the ground up to address the problem of drip email campaign optimization.
At the core of this suite is a hierarchical deep reinforcement learning (RL) algorithm. Considering both immediate and long-term (sequence-general) time scales, we learn policy and value functions based on the predicted utility of messages and sequences that drive customers towards goals based on their behaviors and individual traits for many parallel customer instances. As email campaigns are inherently concurrent sequences of dependent actions, our deep RL algorithm is perfectly-suited to this problem. (It is also better suited than, for example, multi-armed bandit models which assume a customer responds immediately to the actions you take, and that future interactions with this customer are not influenced by your previous interactions).
Moreover, our specific implementation of deep RL is enhanced by the other components of our algorithms suite that feed into it. For example, we have a customer model which feeds in data about customers’ and their ongoing engagement with your website, app, and emails. We also have an independent ’micro’ timing model that optimizes the timing of email delivery to maximize customer engagement. And, as we begin working with more customers delivering a larger diversity of campaigns, we are excited to build out our email attributes algorithm which learns about the content of the messages being sent to better optimize their delivery and tailor them to individual customers.
Today is just the first step towards our goal of helping companies of all sizes leverage the power of AI to grow their businesses.
We’ve chosen to begin this journey by focusing on email marketing automation because, despite what you may hear about chatbots, email is and will continue to be the bedrock of online communication for many years to come. But our email strategies are getting stale, and we think that Optimail can help companies and customers by improving them.
Optimail is free to get started with, so please sign up to try it out and let us know what you think. We hope you enjoy using Optimail as much as we’ve enjoyed building it.
— Jacob & Brock, Strong Analytics