Customer Data Profiling: Problems with Personas (Part 2 of 4)
Even an advanced CRM system can break down when one customer profile is capturing the attributes and behaviors of more than one real individual.
- By David Loshin
- June 21, 2016
Companies collect many types of data about their customers. Conventional customer profiles provide an inventory of descriptive attributes about each individual, incorporating both demographic data (such as age, location, job title, education level) and psychographic attributes (such as hobbies or style preferences) provided by third-party aggregators who collect data through surveys and direct contact with individuals.
In Part 1 of this series, we discussed how more sophisticated analytics tools can infer individual characteristics based on behaviors.
For example, a retail vendor can analyze an individual’s purchase history to dynamically draw conclusions about a customer’s preferences. Many retailers are beginning to use collaborative filtering and recommendation engines to compare one customer’s profile to customers with similar profiles so they can make product offers with some expectation of increased sales.
This kind of analysis should deliver benefits as long as the company’s representation of the customer accurately reflects the characteristics of a unique customer. However, the system breaks down when one customer profile is capturing the attributes and behaviors of more than one real individual. This happens when multiple individuals, or “personas,” conduct their business through a single entry point.
Some examples include:
- Multiple individuals sharing an on-demand TV service (such as Netflix or Hulu). In this case there may be multiple viewers within a single household sharing a single account.
- Multiple individuals sharing an ecommerce account, such as all members of a family using the same Amazon Prime account.
- Multiple companies with a collaborative purchasing agreement using the same account to order office supplies.
There are two key issues with having multiple personas associated with a single customer profile:
Skewed profiles: Because the interactions are likely performed by more than a single individual, the inferred characteristics are going to be biased and unbalanced.
For example, when one individual using a TV service likes to watch horror movies while the other likes to watch romantic comedies, you might infer that the customer likes both genres. Of course, recommending a romantic comedy to a horror movie fan would probably be a mistake. These skewed profiles impact the company’s ability to manage the customer’s relationship and experience.
Imprecise analyses: Because the profile combines the preferences of multiple individuals, there is reduced precision in analytical results.
To continue the example, the recommendation engine might try to find movies that are appealing to individuals who like both horror movies and romantic comedies, which may not be appealing to either of the individuals sharing the account. Imprecise analyses can diminish the value of recommendations and dampen the potential benefits.
To address these issues, a company must accomplish four tasks:
- Determine that multiple personas are associated with a unique customer profile.
- Disambiguate the personas.
- Configure the customer data management environment to accommodate personas linked to a customer account.
- Adjust the analytics to narrow focus by persona in real-time scenarios.
The next article in this series will explore ideas for disambiguating personas associated with a unique customer profile.
Articles in this series on customer data profiling:
Part 1: The Precision Problem
Part 2: Problems with Personas
Part 3: Persona Disambiguation and Behavioral Analytics
Part 4: Persona Analytics
David Loshin is a recognized thought leader in the areas of data quality and governance, master data management, and business intelligence. David is a prolific author regarding BI best practices via the expert channel at BeyeNETWORK and numerous books on BI and data quality. His valuable MDM insights can be found in his book, Master Data Management, which has been endorsed by data management industry leaders.