Customer Data Profiling: The Precision Problem (Part 1 of 4)
Over the years, CRM tools have evolved from simple customer databases into increasingly more sophisticated tools for engagement, but the growing amount of customer data brings new challenges.
- By David Loshin
- June 20, 2016
Customer relationship management (CRM) began with the desire to understand who your customers are and how their characteristics influence their buying decisions. Over the years, CRM tools have evolved from simple customer databases into increasingly more sophisticated tools for engagement.
Traditionally, CRM tools provided a customer data repository and standardized processes for engagement during different phases of the acquisition cycle followed by transition into an ongoing relationship. By coupling customer data collection with analytical methods such as clustering, discrete customer segments could be defined to help drive marketing efforts and targeted sales.
Modern customer intelligence environments, however, encompass more than just capturing and segmenting customer data.
As electronic commerce has blossomed and mobile apps provide both continuous and more comprehensive visibility into customer behaviors, analytics applications for customer interaction use increasingly more sophisticated algorithms. These algorithms exploit predictive intelligence to reach out and engage prospects and customers. They take advantage of opportunities to improve the customer's experience at every potential touch point with the business.
These increased analytics capabilities are impacted by the growing volume of available data. Although early-generation CRM tools focused on some key attributes (such as age and location) enhanced with summarized behavior data (such as median income or education level by geography), today's tools have access to a broad array of additional attribute features.
These might be characteristics willingly self-reported by the customer, such as answers to survey questions when registering for a product online. Alternatively, these might be characteristics acquired through publicly available sources (such as the customer's Facebook or LinkedIn page), through third-party data purchased from one of the growing number of data aggregators, or attributes inferred from text analysis of the customer's social media postings.
One might be able to infer much detailed information about individuals based on the content in their product reviews, posts to fan sites, or even their "likes" of other people's posts. These harvested characteristics can be appended to the customer's record, presumably leading to an enhanced customer profile.
In spite of (or perhaps because of) the amount of data that can be collected, there are some challenges in establishing the precision and accuracy of this enhanced customer profile. For example:
- Limitations on capturing the breadth of knowledge: As the number of explicit and inferred characteristics grow, conventional data models may not be sufficient to capture the attributes in ways that are easily analyzed. There may be a need for more dynamic assignment of attributes based on different preferences or behaviors.
- The speed at which data is generated: Not only has the number of sources increased, the rate at which data is being generated has accelerated. There might be multiple sources for a customer's interactions operating simultaneously, such as mobile app, website, and location data communicated from the same smart device.
- The potential for conflicting inferences: Inference engines might derive new attributes differently due to issues with data synchronization originating from different sources at different times.
These issues introduce substantial complexity when dealing with a single customer's profile. However, they are even more complex in the context of a growing phenomenon that cannot be ignored: the subsumption of multiple personas within a single customer entity.
This happens when two or more people use a single "account" for their transactions. Some examples include all members of a household using the same Netflix account, a married couple using the same supermarket courtesy card, or multiple individuals ordering products through the same Amazon account.
In our next article we will explore the concept of personas and the problems they pose for customer profile management.
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.