Skip to content

How machine learning will optimize marketing results

By Lars Petersen , Tuesday, September 12, 2017

Mind in the Machine blog series

A Q&A with Lars Petersen, Vice President, Business Optimization, Sitecore

In a recent interview here, Paul Roetzer, founder of the Marketing Artificial Intelligence Institute, explained how machine learning will revolutionize marketing. Today, Lars Petersen, Sitecore’s VP of Business Optimization, zooms in on how ML will enhance the customer experience through an unprecedented level of personalization.  

Q: How will machine learning help marketers get a 360-degree view of their customers?

A: A lot of marketers start putting that picture together the first time a customer fills out a form, creates an account, or gives you a credit card number. Considering how customer journeys really work, that’s really rather late in the game. We start putting the profile together from the very first anonymous interaction. And as that picture comes together, we use behavioral profiling to give us an idea of what to do next in a specific customer interaction or what we should be doing next, in order to be relevant for the customer.  

Don’t forget, this picture is only your view of the customer—it’s not what you do with that view. What we call “owning the experience” means acting on what you know, not only about the customer as a person and their history with you, but about the context of the situation. Are they on a mobile device? Is there already an item in their shopping cart? Are there predictable next steps to which they’ll respond positively? 

Q: How will “owning the experience” change the way marketers personalize interactions? 

A: Say you conduct a test on your website. You have three variants of the home page, and each time a variant is shown to the visitor, you also collect information about the visitor’s background, location, device, persona type, whether they’re arriving from a marketing campaign, everything down to time of day. Normally, you’d determine the winning variant based on overall results: Which of the three generated the most-desired responses?  

But machine learning can cluster different visitors into different groups, across any number of factors, to really subdivide the audience segment and let you personalize. Variant two may be the overall winner, but why stop there? It could be that visitors from Norway prefer variant three, except in the morning, or on a smartphone, when they prefer number one. Machine learning will help inform the decision—and then make and act on that decision—about which variant best serves a specific visitor. You don’t have to have human intervention to manually set up one rule for Norwegian visitors on laptops at dusk and another for Swedes visiting from an iPhone in the morning. 

Q: Are we heading toward a world where marketers don’t have to set anything up, where marketing just happens?

A: Well, not quite. There will always be some foundational things you need to set up. You will have to set the business objective; the goal you’re optimizing for. And you’ll need to make sure that certain data sets or sources are in place. Once you have that data in place, then essentially you have the right elements for the machine learning to detect the segments and come up with ways to optimize, ways to bring you better analysis, etc. In short, “quality in” means “quality out” and better outcomes. 

Then you can add automation. So the machine learning identifies the best experience to deliver to a given segment and the automation acts on that.  

Q: How does that scale beyond responding to one visitor at a time, whether from Norway or Sweden? 

A: From your data, the machine learning can identify a segment you haven’t recognized previously. Male visitors from the U.S. West Coast, for instance, might have been seen to more often respond to a certain kind of promotional offer. Having identified a segment, automation could trigger a new campaign aimed at that audience. And of course, based on the response to that new campaign, the data about that segment is refined, and future actions or campaigns become even smarter.  

Q: It sounds like this capability will allow marketers to act in real time to optimize the customer experience. Is that right?

A: That’s what machine learning is bringing: the ability to evaluate an evolving set of data and take appropriate action automatically. Turning insight into optimization is a win-win because it provides a relevant experience for the customer and it helps increase our business results.  


Follow Lars on Twitter @LarsBirkholm


Sign up for a demo today!

Take a closer look at the Sitecore Experience Cloud