Early in my banking career at Wells Fargo, the leadership of retail banking was open to transformative ideas that could help the bank stand out in a crowded marketplace. At the time, the bank relied on traditional survey-based market research and knew very little about its customers’ goals and attitudes.
Ethnography in banking
A consultancy called the Doblin Group introduced the bank to corporate ethnography—research techniques made famous by Margaret Mead—applied to the financial lives of consumers. This ethnographic research involved extensive interviews with customers about their attitudes toward money and everyday practices around budgeting, saving, and investing.
The results were controversial. They revealed customers’ confusion and disappointment with traditional banking practices, and the sometimes-adversarial relationship they had with the bank. The bitter truth can be hard to swallow—I witnessed the head of retail banking get into a shouting match with the lead consultant over the impact of the bank’s overdraft fee policy on consumers.
Personas as a design principle
As digital became the dominant banking channel, research practices evolved into “design ethnography,” emphasizing human-centered design principles pioneered by the technology consulting firm Sapient.¹
Another innovation from user-centered digital design was the concept of a “persona”—a personified attitudinal customer segment. Personas are stereotypes of user profiles, given names, personalities, and even portraits as if they were real people. Their application at the bank led to a sharper focus on end-user needs, resulting in better design decisions and clearer definitions of a product’s feature set.²
The bank’s digital product team identified a dozen personas, each with a name and a rich profile of attitudes, interests, and psychological attributes. These personas heavily influenced the design of the bank’s website.
Personas assigned to customer data
My team was tasked with labeling the bank’s customer base using these personas. While demographics and product holdings were useful inputs, we also needed proxies for psychological attributes, drawing on third-party data from vendors like Equifax. The identification and assignment of personas turned out to be a blend of art and science.
This synthesis of qualitative and quantitative data led to an iterative process of refining the personas to better represent the customer base. A major challenge was determining the optimal number of personas—enough to capture distinct profiles, but few enough to remain actionable. For this, we used an unsupervised machine learning algorithm called k-means clustering (available in Teradata ClearScape Analytics®), which minimizes variation within clusters while maximizing variation across them.
Ultimately, we defined eight personas that were widely adopted across the bank. They played a central role in the evolution of the bank’s digital platform and were later adapted for marketing, efficiently capturing customers’ product needs and channel preferences.
1. Melissa Cefkin, “Ethnography and the Corporate Encounter: Reflection on Research in and of Corporations,” New York, Berghahn Books, 2009.
2. Lene Nielsen, “Personas,” Interaction Design Foundation, 2014.