Machine learning is a much-discussed topic in the customer loyalty space. On the one hand, it seems to promise a level of marketing individualization that has been dreamed about for years, ever since the idea first began to circulate. On the other hand, due to its being “hijacked” by so many in the MarTech space, each company defining it differently, the term doesn’t seem to be well understood, either at the technology level or at the concept level. To create value and make their offerings more sanguine, too many marketers assigned their own connotative definitions to the term. These conflicting definitions did not create clarity. Rather, they created obfuscation.
To clear things up, to help us better understand the true power of machine leaning, Loyalty360 recently sat down with Christian Selchau-Hansen, CEO and Co-Founder of Formation, a marketing technology provider that offers a platform that combines AI decisioning with flexible experience constructs. Selchau-Hansen presented an insightful background of knowledge on what machine learning is and how it can help improve a range of brand KPIs.
“One thing to remember,” Selchau-Hansen said, “is that machine learning is a technique or tool to solve many kinds of problems. It’s a relatively new topic, and there are challenges in terms of understanding how to think about it. That’s why it’s very important to know what type of problem you’re trying to solve with it.”
The problem, as Loyalty360 pointed out, is that marketers talk about machine learning in so many different ways that it can be difficult to know what they mean when they use the term. Some use it interchangeably with artificial intelligence, meaning a computer that can learn and function on its own after some initial inputs. Others we’ve heard from use it in a more limited fashion to describe algorithms that work semi-autonomously, requiring oversight by experts. Within the customer loyalty industry, this is something we continue to see as a challenge: multiple misunderstood connotative definitions around a topic.
Selchau-Hansen spoke with the second definition in mind, stressing that marketers must define the problems they want to solve before they can do anything with machine learning. “It’s like saying you’re using math as a technique for solving a problem,” he said. It isn’t that you can’t use math to figure things out. It’s that saying you’re doing it, without being more specific as to how, doesn’t communicate anything meaningful.
“The particular class of problems where we’ve been applying it,” Selchau-Hansen said, “is customer engagement, loyalty, and retention. Marketers think of optimizing mass messages, but with digital channels, you can reach individual consumers. That creates a challenge concerning how to tailor messages specifically to individuals. How do I reach a customer with a message that is interesting and unique to them?” The point he is making here is that, typically, marketers are trying to use machine learning to reach true personalization, a goal he definitely believes is possible.
Additional marketing goals identified by Selchau-Hansen include growth and increased lifetime value. He noted that two things can help marketers reach these goals. “High-level frameworks are helpful,” he said. But, perhaps even more importantly, brands must deepen customer relationships.
Still, brands should make sure to focus on machine learning as a tool to reach these deeper relationships without sweating and toiling over the technology itself. When you want your car to handle better, you can put in a supercharger, but the focus should be on the car driving faster, not on the nuanced understanding of the supercharger itself. The supercharger is the tool that gets you there, and machine learning should be viewed in the same manner.
Selchau-Hansen admits, however, that the marketing landscape is very confusing. “MarTech has exploded,” he said. This makes defining problems, solving for personalization, and figuring out what he calls “solution patterns” very difficult. Focusing on new technology is a challenge, and brands need to know what they want to accomplish before they drift off into the uncharted territory toward the sparkly object that will eventually lose its luster.
Selchau-Hansen believes there is hope, though. “We are in a time where there are some patterns that are beginning to emerge that can give marketers ideas about how to solve their problems,” he said. “MarTech architecture will need to become more established, but best-of-breed marketers are assembling pieces to enable them to present unified solutions to the customer.”
One issue brands will need to deal with to offer these unified solutions is internal communication. Sometimes, it can be difficult for separate departments within an organization to align on a common goal, either because they have different methods or directives or both. To tackle this obstacle, departments will have to come together. This should be the organizational focus.
“If you can share data layers and build apps on top of it,” Selchau-Hansen said, “that will assist people in having a common source of truth. At the same time, there is a strong desire to have a single, full view of the customer. That is a worthwhile goal, but it can often mean that marketers think they need to have that before they improve customer loyalty.” This is to say, steps can be made to deepen loyalty even before an organization constructs a full profile on a customer.
Of course, to eventually reach that profile, brands will have to devote time and energy to measurement. That’s a big challenge, Selchau-Hansen said. “But the bigger hurdle is a sense of required transformation. By focusing on what you’re trying to accomplish, what value you’re delivering, you can build on it to create a much clearer pathway to how you add value now and over time. This approach builds momentum in an organization, builds additional capabilities over time. What’s helpful is to have data on what’s working.”
One thing Selchau-Hansen believes is definitely not working is classic demographic segmentation. “That is no longer sufficient,” he said. “What’s required is a much higher degree of relevance. How do you make it easier for your customer to achieve goals? How do you market products to put your customer at the center? In order to do that, you need a motivational understanding of the customer.”
Loyalty360 pointed out that this means making the customer the hero of the story. Selchau-Hansen agreed. “It is a different way of thinking,” he said, “but I’m a firm believer if you want to establish loyalty, you have to put the customer first. Tell the customer how you can solve their problems or accomplish their goals.”
We then wondered if this would require small businesses to rearchitect themselves entirely. Selchau-Hansen replied, “In most cases, you don’t have to completely rearchitect. You’ve got to engage customers with a customer-centric lens. That’s the start, not a product-centric lens. It changes how you engage customers, what kinds of messages you send, etc. If you look at what the customer is trying to accomplish, you identify areas of friction and value, and look for how you solve customer points of friction. That could be in communication or how you’re packaging product, but the point is to reduce friction in the customer journey.”
Selchau-Hansen suggested that reducing friction for the customer can be done in a few different ways. First, brands should be leveraging the information and knowledge they already have, which enables prioritization of what’s most relevant and meaningful. Next, brands should be determining the capabilities they have to improve opportunities, make them better. From there, brands can create momentum and build up their strategies.
Selchau-Hansen concluded by returning to the subject of machine learning. “One aspect that should be understood is that machine learning is about augmenting and scaling decision making. If you’re trying to make more relevant content for your loyal customers, you begin at segmentations. Let’s say you’ve got half a dozen, and you try to increase relevance by going to 12. You’re doubling the complexity, but when you apply machine learning, you immediately have an additional set of capabilities you can bring to bear. You don’t have to deal with the challenge of linear rescaling. Given a specific objective, you can have machine learning begin to solve problems within a boundary, creating more relevant content with far less effort.”