Connecting Personalization and Strategy
BY CHUCK DENSINGER – COO & CO-FOUNDER
The hot topic of the day in customer analytics is personalization. It seems every software tool, cloud data platform, and martech suite tout the ability to personalize the customer experience to drive conversion, loyalty, and customer value. But there’s a lot of imprecise thinking out there—we want to call every variable treatment personalization. However, is just varying the product placed in an email campaign personalization? If we send 12 versions of an email to subsegments of our 5M customer population, is that personalization? And how do we personalize in the context of customer strategies? We can’t do strategy at the individual customer level—we need to understand broader patterns. Is acting on those broader patterns personalization?
At Elicit, we’ve got a simple framework that we think helps to clear this up. It’s a structure that collectively makes up what we call a Customer Foundation.
The bottom layer is Raw Customer Data. All of their interactions, purchases, preferences, shopping trips, etc., are stored as a fact base. This data consists of web visits, mobile app usage, calls to the call center, purchases in all channels, contact information, loyalty program engagement, contact preferences, and much more. In a well-instrumented customer experience, it should encompass every single interaction between a customer and your brand.
The middle layer is a set of simple models we call Customer Attributes. A Customer Attribute explains or predicts some facet of the customer’s behavior relative to all of these touch points. Each Customer Attribute answers a business question related to customer behaviors or experiences. How often do they shop online? In store? What kinds of things have they called customer support for? What is their propensity to purchase in a given category? At a given price point? What location do we think they’ll visit next, and when? What do we think their NPS is (if we haven’t been able to directly ask them)? We literally build hundreds of Customer Attributes and they collectively comprise a body of insight about each individual customer. The attributes provide a structured layer of analytics that can be examined for patterns.
The top layer is Strategic Segmentation. There are innumerable ways to segment a large population of customers. Indeed, each of the Customer Attributes provides a simplistic basis for segmentation. Strategic Segmentation answers the question: “What are the most important patterns of customer engagement with my brand?” By sifting through and analyzing the hundreds of attributes in the middle layer, engagement patterns emerge. Customers aren’t as unique as they like to believe; there are dominant patterns of channel interaction, product category preferences, marketing and CRM response, price sensitivity, etc. These patterns differ for every company, even within an industry. But they become very clear when the data and analytics are organized in this way.
The top layer, Strategic Segmentation, answers the “who” question—what groups of customers emerge from the most important patterns of engagement with the brand? This is all based on facts: behavioral data. The customers have voted, we’re just counting the votes. You must change your product (including pricing), customer experience, or marketing to alter those patterns of behavior.
The segments also enable you to see your financial performance in customer terms—where does my revenue come from? If revenue is down in a quarter, which segments spent less? Since all revenue comes from customers, this is more logical than looking at which geographies, product categories, or channels underperformed—or at least as important.
But the segments alone don’t answer the “what” question. You need Customer Attributes to do that. In our “best customer” segment, for instance—those most engaged with the brand—retention is clearly a top priority. Some customers in that segment will be happily engaged, while others are at risk of leaving. Churn risk, and the behaviors that indicate it, will be visible through the Customer Attributes, and the details will vary from one customer to another. Deeper analysis is needed to develop tactics to reduce churn. Those tactics must be tested and refined. The Customer Attributes tell us who to target with those activities.
But even at this level, we must go deeper. Each individual customer is at a different point in their journey. If they just bought a given product, we shouldn’t recommend that product to them in the next email we send. But this happens with retailers and e-tailers today…All. The. Time. Personalization requires us to apply what we know at the individual customer level using their own individual data—preferences, recent purchases and experiences, support calls they’ve made, and so on. After deploying the strategies and tactics suggested by the Strategic Segmentation and Customer Attributes, we refine with the “how.”
So, let’s put it together. We identify that a given customer, we’ll call her Kyla, is a best customer (Strategic Segmentation), but at risk of churning because we’ve seen her web browsing drop off (Customer Attributes)—an early warning sign previous analysis has told us is 72% likely to indicate waning engagement. Our re-engagement strategy includes an email offer with a discount, or with a loyalty points offer (which we’ve learned from studying Customer Attributes). Customer Attributes tell us Kyla is more likely to respond to the points offer—she has been an active redeemer of points in the past. She currently has a points balance that puts her 132 points away from being able to redeem a free gift with purchase (individual customer level data). We decide to send her an email offer for 150 loyalty points (to get her over the threshold) with the purchase of any product over $30 (Strategic Segmentation and product strategy decision), making it clear in the offer messaging that this will put her over her redemption threshold. The email populates a handful of product inspiration items based on her propensity to purchase in various categories and brands (Customer Attribute), filtered by items she has actually purchased (Customer Level Data). If she clicks through to the website, additional suggested items are displayed using similar logic.
So, where did the personalization happen? That’s the wrong question! Retaining valued customers, growing mid-level customers, acquiring new customers, all require us first to have a strategy for who we’re targeting; next, to develop tactics for how to engage them; and finally, to customize messaging to make sure we’re as relevant as we can be. You might say personalization happens throughout—we realized Kyla was a really important customer who needed special attention, we had some proven methods for giving her that attention, and we tailored them to her specific interests and past purchases.
Micro-segmenting traffic on a website; singling out the top 10% of your customers for special treatment (while ignoring others); buying “personalization tools” that use AI or machine learning to drop dynamic content into emails; using DMPs and ad servers to flash banner ads at “acquisition targets;” these techniques all have some value, and can make individual touch-point interactions more profitable on the margins. But are we turning our customers into brand-lovers by doing those things? Are we growing the total value of our customer portfolio—the whole collection of our strategic segments? Likely not.
Kyla comes to us because of the products she loves, the value we provide, and the way we make her feel. The Who-What-How framework leveraging Strategic Segmentation, Customer Attributes, and Customer Level Data enables us to talk to her when it’s important, say something relevant, and let her know we know a little bit about her.
Now doesn’t that sound personal?