Most brands still rely on personas when considering who their customers are and the best ways to engage them. However, personas are inherently generic. It’s difficult for any human to consider thousands—or even tens of thousands—of attributes. As a result, we rely on a handful of psycho-demographic and behavioral attributes, along with a spend level, to frame how we think about our customers and plan ways to engage them.
But people are inherently complex. That complexity, if not addressed, means you may be leaving quite a bit of money on the table. Two consumers can be high spenders for a particular brand—but for two different reasons that are driven by vastly different motivations. As such, each deserves a method of engagement that speaks to him or her personally. Graph analytics and clustering allows brands to find relationships between attributes, enabling marketers to understand the motivations that drive behavior.
Additionally, programmatic and dynamic creative optimization technology now allows brands to target and assess a multitude of customer types with messages that appeal to them. Now that technology has eliminated all that manual work, marketing strategists can consider looking at the many sub-personas that exist. Again, clustering is very helpful here; let’s take a look.
Clustering your customers
So what’s clustering exactly, and why is it valuable? Within each customer base, there are many separate communities made up of people who share unique commonalities. Clustering looks at every possible data point and rank-orders them.
To illustrate why this is valuable, let’s assume that a direct-to-consumer brand sells a variety of accessories and has developed an avid fan base. Some people love their products because they’re eco-friendly, a tremendous amount of care has gone into developing sustainable manufacturing. Others love the products because they’re made of natural ingredients that feel good against one’s skin. Still others are fashionistas who glom on to the latest trends, and these products are hot (for now).
Here we have three different customers who fit into a single persona of “big spenders.” It may be possible to prompt them to spend more, but without the nuanced insight into their specific motivations, knowing how to do that is a challenge. There’s simply no way to decipher how best to optimize other critical marketing initiatives, such as a loyalty program promotion, messaging, or ongoing email campaigns.
Here’s a real-life example. MECCA is a premier beauty retailer in Australia that wanted to validate the customer personas it had been using for a long time. To do the clustering, MECCA leveraged its first-party data gathered from its loyalty programs and sales transactions to answer strategic business questions: what did people buy? Where did they buy it? Are there distinct behavioral patterns that reveal new insight into who the shoppers are?
The exercise allowed them to look at their big spenders with a whole new lens. Rather than lumping them into a single, high-value group, MECCA realized there were important distinctions. For instance, the data identified affinities within a cluster of loyal customers who shop across multiple categories, but are defined more specifically by the number of items purchased rather than by the price point at which they shop. This contrasts with other big spenders, who tip the scale on all fronts: spend, frequency, units purchased, and channel adoption.
Complicating matters further, not everyone in this newly identified cluster is a big spender. Retailers can define some shoppers better by using a combination of behaviors than by standing out in any one. Hence this group would easily be missed, or many of the customers would be missing out from it using traditional persona development methods, such as surveys.
Any marketer can see just how valuable the newly discovered distinctions in customers are to a brand. These kinds of insights can drive unique promotions, messaging, and strategies that cater to each customer cluster. In fact, clustering allowed MECCA executives to understand the types of customers who are most likely to respond to a new product or store launch. It also allowed them to see how each store or brand launch evolves. Do the initial customers stay with the store or brand, or do they move on? And by monitoring the distinct clusters, MECCA can adapt its marketing or services to ensure it’s meeting its precise customer needs.
It’s hard to get around the fact that people are complex creatures, yet the personas long favored by marketing strategists can only segment customers on just a handful of dimensions. Clustering, on the other hand, looks at the universe of possible dimensions, which yields more strategic insights.
Tim Burke is CEO of Affinio. The company provides what it calls an “augmented analytics platform” to uncover hidden insights in marketing data.Favorite