All financial institutions aim to deepen customer relationships. Traditionally, customer satisfaction surveys have been the go-to method for gauging engagement. While intuitively appealing, these surveys have serious limitations when it comes to assessing the complex dynamics of behavioral loyalty—such as actual customer attrition—by relying on attitudinal loyalty, or the stated intention to remain loyal.
Customer satisfaction surveys ask respondents to self-assess their satisfaction using a five- or seven-point ordinal scale. However, these ordinal responses are not continuous, and in practice, they tend to cluster into three broad categories: “good,” “bad,” or “indifferent.”
The Net Promoter Score: Popular but problematic
The Net Promoter Score (NPS) is a widely used metric based on a single survey question: How likely are you to recommend this company to a friend? Developed by Fred Reichheld of Bain Consulting and popularized through a 2003 Harvard Business Review article, NPS categorizes respondents as "promoters" (9 – 10), "passives" (7 – 8), or "detractors" (6 or below). The score is calculated by subtracting the percentage of detractors from the percentage of promoters.1
However, NPS has several methodological flaws. It arbitrarily discards responses rated 7 or 8—eliminating a significant portion of the population—and lacks predictive validity. Academic research has shown that NPS does not reliably correlate with financial outcomes like revenue growth.2
Understanding the drivers of satisfaction
To truly measure and monitor behavioral loyalty, banks must develop robust predictive models of customer attrition. Tools like ClearScape Analytics® support this by enabling the use of econometric methods (e.g., logistic regression) and machine learning techniques (e.g., gradient-boosted decision trees).
In my banking career, my team was tasked with extracting actionable insights from customer satisfaction surveys. When we overlaid survey results with attrition risk model scores, the findings were striking: Customers with the lowest attrition risk were overrepresented in survey responses, while those at highest risk were underrepresented. In other words, the voices the bank most needed to hear were the least heard.
We also examined how service issue resolution impacted engagement. By comparing behavioral profiles with responses to the “willingness to recommend” question, we found that satisfaction scores were influenced by three key factors: the customer, the channel used, and the nature of the task. For example, resolving fraudulent transactions elicited stronger emotional responses than checking a savings account balance.3
Unlocking insights from free-text responses
Many surveys include free-text responses, which are difficult to categorize manually and often inconsistently interpreted. Fortunately, the Bring Your Own Model capability in ClearScape Analytics® allows banks to apply advanced language models to analyze these responses—identifying topics and sentiment with precision. These insights can then be appended to customer records, enhancing root cause analysis and enabling more targeted engagement strategies.
1. Frederick F. Reichheld, "The One Number You Need to Grow," Harvard Business Review, 2003, https://hbr.org/2003/12/the-one-number-you-need-to-grow.
2. Timothy L. Keiningham et al., "A Longitudinal Examination of Net Promoter and Firm Revenue Growth," Journal of Marketing, 2007.
3. Gary Class, “Customer Journey Analytics in Banking,” 2024, https://www.teradata.com/resources/white-papers/customer-journey-analytics-in-banking.