Adaptive Customer Profiling: Integrating Quantitative and Qualitative Customer Analytics

Most business leaders now recognize that organic growth is a direct result of their ability to deliver a differentiated, compelling, and increasingly personalized customer experience.  Effectively delivering such an experience is dependent on the organization’s ability to understand what attracts customers’ attention and what drives customers’ behavior. 

As you know, recent advances have lowered the investment threshold for consolidating and analyzing the massive amount of data that most organizations’ have about their customers.  Predictive modeling can then be used to make increasingly effective and individualized decisions about the treatment of customers.  For example, these approaches can be used to leverage customers’ past behavior to predict: the value of each customer, how likely that customer is to respond to specific offers, that customer’s price sensitivity, or how likely that customer is to attrite, as well as, what retention actions are likely to be effective.  (See:  Using Predictive Modeling to Optimize Customer Relationships)

Despite the enormous potential, purely quantitative approaches are insufficient.  In particular, quantitative customer analysis has natural limitations, including:

• Trying to predict the future based on information about the past
• Data gathered at a limited number of customer touch-points rather than an end-to-end understanding of the customers’ experience, including the more important non-touch-points
• Surface level behaviors rather than a deeper perspective on customers’ motives, goals, plans, as well as, how they think and feel about their experiences

Trying to understand the customer based purely on quantitative analysis can feel a little like trying to determine how the furniture upstairs is arranged…  by tapping on the ceiling!  Obviously, you’d get a much clearer picture if you just went and took a look… rather than trying to infer what’s going on through indirect and limited data sources.

In addition, inferences drawn from purely quantitative approaches are prone to interpretation errors.  Without an adequate qualitative context for understanding the data, we’ve seen too many organizations draw conclusions akin to “Our customers in South Florida are born Hispanic and die Jewish.“

The most powerful results come from the synergy between qualitative insight and quantitative analytics.

• Qualitative Insight: Leveraging knowledge from in-depth research, observation, elicitation, as well as, listening to the conversations that take place between customers in emerging social networks.  This qualitative insight is used to frame and guide quantitative analysis.
• Quantitative Analytics: Leveraging patterns in demographic and transactional customer data in order to predict, classify, and optimize elements of the customer experience. This quantitative analysis is to validate, refine, and populate the context created via qualitative insight.

In practice, organizations and the functional departments within them tend to have a strong bias for one of these modes.  More “left brained” organizations or functions emphasize the quantitative approach and feel uncomfortable with going out to actually observe what’s happening with customers.  More “right brained” organizations or functions emphasize the qualitative approach, are out living with their customers, but also tend to make decisions that aren’t supported by sufficient analytical rigor.  As a result, it’s difficult for organizations to put together the pieces in a way that generates a holistic perspective on the customer.

In our customer experience work with clients we are beginning to create Adaptive Customer Profiles that can be used to integrate quantitative and qualitative knowledge about the customer. 

An Adaptive Customer Profile is…  

… a formal knowledge representation structure used to capture the customer intelligence necessary to effectively customize communications, effectively assign service resources, optimize the presentation of high probability offers, and adapt pricing to customers’ price sensitivity.

Adaptive Customer Profiles for a given business situation generally include:

• Descriptive Information:  Identifiers, demographic characteristics, etc…
• Potential and Current Value:  The expected and current value of this customer.
• Customer Network Information:  The customers’ role and placement in an influence network of customer relationships.
• Personae Classification:  The degree to which the customer demonstrates an affinity for one or more personae classes that exist in the marketplace.  These personae classes are an extension of psychographic segments that define the predominant “mental models” in the marketplace.  These personae are characterized by shared customer goals and preferences, goal-directed behavioral patterns, cognitive schema, and temperamental characteristics.  These temperamental characteristics include the customers’ orientation towards novelty seeking, harm avoidance, reward dependence, and persistence.  (See Cognitive Ergonomics:  Designing Experiences that Fit the Customers’ Mental Model)
• Relationship State:  The level of attachment this customer feels towards our business as evidenced by their transactional and interactional behavior.
• Context Sensitive Behavioral History:  key behavioral indicators derived from inquiry and order history, service records, etc…

Adaptive Customer Profiles are derived through an integrated set of qualitative and quantitative activities.  Qualitative work includes customer observation and elicitation (See:  Observation and Elicitation:  We Like to Watch!) in order to uncover insight that is used to develop an effective personae classification scheme.  Quantitative work involves predictive modeling focused on the leading indicators of customer behavior and measuring the affinity that customers demonstrate for one or more personae.

For example, we are working with a leading healthcare organization to design an integrated patient-physician experience that can adapt to the fact that different patients have fundamentally different mental models associated with their health and the consumption of health related services.  Some customers will be high novelty seeking naturalists; some will be low persistence avoiders; some will be more high harm avoidant active consumers, etc…  The experience design integrates an Adaptive Customer Profiling module that identifies the extent to which each customer fits one or more of the common personae that exist in the marketplace.  Based on that Adaptive Customer Profile, we can then customize patient communications, instructions on courses of treatment, the presentation of wellness programs, etc…

We are also developing a similar personae classification scheme focused on Mass Affluent consumers of financial services.  Almost every financial services company is currently targeting this valuable “segment.”  The issue is that, by its’ very nature, the “Mass Affluent” segment is an exceptionally diverse group of individuals that only share the fact that they have assets and/or income above a certain level.  Companies that attack this market with a mass market mentality will almost certainly lose.  However, financial institutions that can target meaningful sub-segments of this market with a highly differentiated offer can create an experience that is attractive and differentiated with a substantial group of these customers.  You might imagine a hip and differentiated “I Hate to Plan” themed experience for the sub-segment of Mass Affluent customers that are Avoiders… or a more conservative, goal-driven experience customized to the customers that are Achievers.   A financial institution that embeds an Adaptive Customer Profiling process in their interactions with customers could more effectively customize the experience to the customers’ goals, behavior, mental model, and temperament.

Personae Driven Experience Design

A persona is a fictitious person created for the purpose of helping designers and decision makers understand how people actually experience their interactions with a product, service, or organization.  The use of personae was popularized by Alan Cooper in the book “The Inmates are Running the Asylum.”  In this critique of the software development industry, Cooper recommends the use of personae to help developers get a practical, visceral feel for the ways users think and behave. Since that time, the use of personae has become very popular in a wide variety of product and user interface design applications.  Personae are given a name (Bob, Sue, etc…) and a set of richly described characteristics, situations, goals, pain points, and behaviors that are relevant to the design.  For any given application, there are usually a relatively small set of personae that characterize the range of users or customers.  Cooper has suggested that one persona is usually sufficient.

The  benefits of personae in understanding and designing distinctive customer experiences are substantial.  Typically executive leaders and functional managers do not have a clear and concrete understanding of how their customers experience the world and, more specifically, their interactions with the client’s organization.   Personae are powerful because they put a specific human face on often abstract customer information.  In this way, they are fundamentally different than customer or market segments, which are generally shared characteristics of categories of customers.  This “human face” makes it easier to make decisions and design tradeoffs with an understanding of how what you do either fits or doesn’t fit for the customer.

One of the best examples of using personae for customer experience design is Best Buy, who made substantial changes to their store design, merchandise assortment, training, etc… based on the definition four customer personae.  In particular, they started to shift elements of  the experience design to work for the persona they called Jill.   Jill is a soccer mom that does most of the shopping for her family but is intimidated by electronics stores.

Unfortunately, the way most organizations develop personae appears to be very loose; more of an art than a science.  The generally accepted best practice is that personae should be based on solid ethnographic research with customers.  However, sometimes persona are just made up based on what the team thinks they know about customers (because they know so much about them already!).  Assuming research is done, the process of turning research findings into personae is also very loose.  Typically, common themes across customers are identified and clustered in a creative process that generates a plausible enough set of personae.   In addition, details are usually added to these personae in a way that “rounds them out” and makes them more believable.

Over the past couple of years, we’ve been trying to address the lack of rigor in personae development.  Our starting place for this was our emphasis on designing from “mental model of the customer” rather than the “mental model of the company.”  Not only does this perspective address the same basic objective as customer personae but the idea of defining personae precisely based on elements of a mental model is appealing.  It also provides a means of deciding how many personae are needed since the only reason to have different personae would be because there were relevant and substantial differences in the mental models of two different types of customer.

Our working definition of Cognitive Customer Personae include models that capture:  what the customer is trying to accomplish; the end-to-end behaviors the customer typically performs to accomplish those things; a structure of beliefs and temperamental characteristics that drive their rational and emotional reactions to their experience.  Each one of these personae is described by four models that are described in more detail with a few examples in the post titled:   Cognitive Ergonomics:  Designing Customer Experiences that Fit with Customers’ Mental Model.