In an ideal world, every interaction with a consumer would be certain. We’d not only instantly process everything about them — where they live, what they have purchased in the past, what images are most compelling — we would know exactly what action is going to convert them to a buyer. While the right data architecture gets us close to this ideal, we live mostly in a world of incomplete insight.
The only standard we can hold ourselves to is to make the best decision we could in the context of all of the information available in the moment, not the best decision that we could make given a crystal ball.
Making good decisions in the face of this uncertainty is the foundation for differentiating your brand. The good news is that the right personalization architecture can handle your constantly changing world without needing to drain uncertainty from interactions. But to capitalize on the opportunity, we have to let go of how we have traditionally approached audience segmentation.
Advertisement
If you start with the most basic optimization, you’re seeking ways to make a uniform experience — an experience that’s the same for everyone — better. To improve conversion, you try one or more alternatives. You see how well that experience works for that same universe of everyone, and pick the best option. The problem with this approach is that it assumes everybody looks like the average person in that universe. When you improve the average, you’re able to have a new collective experience for everyone that moves that average up — but you still did it in a way that made the experience better for some people and worse for others.
The natural thing to do from there is to say, “If I’m hurting Peter to improve Paul’s experience, then I should work on these different parts of the population separately.” We call that segmentation. It might start with just one or two, but could quickly move to tens or hundreds of segments, and you end up doing the exact same thing over again.
You say, “Here’s the midpoint of each of these populations. This becomes my universe.” Segmentation offers an improvement over treating everyone the same, but it too has limitations. Each segment is as resource-intensive as it was for the audience as a whole. In fact, it’s actually more complex because you have to discover the segments and prioritize them, in addition to having infrastructure that caters to the different segments. Your potential benefit is only the incremental improvement to the average of each segment.
That’s where segmentation falls apart. The work scales linearly. If you have three segments, you’re doing three times the work, but you can’t get three times the return because you are working with increasingly smaller groups of individuals.
Flawed Segmentation Stalls Personalization
Many approach personalization with this flawed segmentation framework. You may think of true personalization as creating a segment of one. Imagine you are offering ice cream to a stadium of fans. In the flawed view of the segment of one model, you would have to create a customized flavor of ice cream for each person. One person would have vanilla, the next chocolate chip, a third marshmallow and so on. This clearly is not scalable when we consider personalizing experience across every web site interaction, mobile inquiry or email communication.
True one-to-one personalization is something different. One-to-one is about individual decisions, not about building unique, one-off experiences for individuals. Instead of trying to create a new flavor of ice cream for every person you meet, you need to decide which of all the available options you have to offer is best for a specific person. In the ice cream analogy, that might mean having a dozen flavors to choose from and knowing which is going to appeal most to each person in the audience. The magic comes from building a flexible model whereby the way we come to the decision of what to offer may be completely unique because what we know about each individual varies.
This is a significant mind shift from thinking of personalization as a segment of one. When you think about experiences that respond to what someone is doing in the moment of a narrow buying window, it requires a solution that can first identify the customer as an individual, then know which channel they respond to best, push a message out as they are about to leave the site or pass by the storefront, and take into consideration their full purchase and behavioral history.
For example, imagine a customer visits an apparel retailer site and is presented with a navigation bar, imagery and promotions that are most relevant to him based on purchase history, age, location and search history. He is directed to a product page with a jacket that matches his personal style. He finds the color he loves, but it’s not available in his exact size. Two weeks later, a very similar jacket is re-stocked, and the customer gets an email alerting him that a similar style to the one he recently viewed is back in stock in his size. The email includes a link that takes him to a product page with not only the jacket, but matching belt and gloves that the algorithm suggested, along with brand imagery of guys in the same general demographic as him wearing the jacket, crowdsourced from social media.
The customer is impressed at the follow-up of the brand and ends up buying not only the jacket but the matching gloves. This was all without any kind of special promotion.
Operational Personalization Delivers Meaningful Interactions
The ability to reach this ideal is what we call operational personalization. In this new model, you load multiple creative assets and then optimize who receives them in order to maximize or minimize a given goal metric, such as bounce or conversion rate. No longer will you have to segment ahead of time or rely on guesswork.
Operational personalization is the ability to use everything known to deliver meaningful interactions with each individual at any moment across all touch points.
In order to reach the AI-powered future, a mind shift of what true one-to-one personalization is must occur. With the ability to create individualized experiences, the potential for profitability has expanded beyond what was possible with previous models, and it’s only a matter of time before we see how operational personalization will change retail as we know it.
Lucinda Duncalfe is CEO of Monetate. A seasoned entrepreneur and innovator, Duncalfe has proven experience building mature sales and operations cultures, and developing product strategies that accelerate company growth. She has served on Monetate’s Board of Directors for the past seven years, and helped Monetate create and establish a multi-billion-dollar market for digital personalization.
