The anticipated value unlocked by artificial intelligence (AI) across Marketing and Sales is $1.4-$2.6 trillion, according to the Harvard Business Review. Yet, when businesses look to incorporate machine learning into their MarTech stacks, they are often met with insufficient, incomplete, or inadequate data.
This might explain why Adobe Campaign has prioritized such intuitive, marketer-centric workflows, rendering it one of the most popular campaign development platforms around. Adobe Campaign enables you to send emails directly, without going through a third party, all while conveniently logging your email history. In other words, all of your historic email sends, opens, clicks, and unsubscribes are automatically catalogued for you when running Adobe Campaign. This results in a well-populated, complete relational dataset, providing a wealth of history that can be applied to training machine learning models.
Privacy is an ongoing concern in the digital marketing space, with more uncertainty due to increased regulatory scrutiny. To design the most forward-thinking ML platform for customer journey personalization, our experts at Munvo primarily work to build around customer behaviour rather than customer characteristics. As such, any two customers that behave in the same way are virtually indistinguishable to our models; we cannot reverse engineer their identities from the data. In fact, our clients never upload any Personally Identifiable Information (PII) to our public cloud—no names, no ages, no ZIP codes.
Our ML algorithms sift through email history data to see patterns, and then create a model to predict each customer’s future behaviour, at any granularity. So, for instance, if someone is sent an email at 9AM on New Year’s Day, our models can predict whether they will open it by 10AM—or later that day, or even within the next week. This is incredibly powerful, but of limited use to marketers who want to increase open, click-through, and conversion rates. For optimal results in driving heightened engagement, you need to determine the best time, or most opportune moment, to send an email.
To accomplish this, our team runs simulations using our machine learning models to predict when a customer is most likely to be receptive to our communications within a given window. For example, if you set a timeframe of 9-12AM, we simulate email deployments at 9AM, 10AM, and 11AM. For each customer, the send time that results in the highest probability of an open or click will be registered as the optimal send time for that customer in this campaign. And since this simulation is re-run before every campaign execution, we can ensure that the send-time optimization effortlessly stays up to date with recent history—without ever going stale.
This factor proves even more important when looking to reduce opt-outs. Through our simulations, we can also determine when customers become fatigued and are most likely to opt out. This enables us to take preventative action and suppress messages to those customers.
By combining these two capabilities, we’re able take a more holistic approach to the customer journey because we now know how to both minimize fatigue and perfect send time. We can thereby personalize the most impactful touchpoints for each customer, over a journey lasting several months, solely using the data already in Adobe Campaign. All this, without compromising customer privacy at any point.