What We Covered
In this session, Ed Powers explored regression analysis and its use in the customer success domain. Many customer success (CS) teams rely on relationship-driven approaches and traditional metrics like QBR and NPS scores. But mathematical models can identify risks and opportunities with much more accuracy.
Using Regression Analysis to Drive Customer Success Outcomes
Regression analysis is a technique used to describe relationships between random variables. To illustrate this concept, Ed used the example of a person’s height and weight. By plotting height and weight on a scattergram, one can see a general trend: as height increases, weight tends to increase as well. This becomes a formula that allows you to predict the weight, based on height.
However, the relationship between height and weight only explains about 25% of the weight variation. Other factors such as age, gender, health choices, and genetics play significant roles. By adding more variables to the mathematical model, the accuracy of predictions can improve. This is where regression analysis becomes invaluable in customer success. It enables prediction of key outcomes like logo churn, product churn, expansion revenue, and account contraction.
While regression analysis is a powerful tool, it is not the only statistical model to use. Other methods such as factor analysis, co-variance analysis, and control charts can provide valuable insights. The choice of technique depends on the specific goals and needs of the organization.
Health scores are a typical immediate application. Ed shared a six-step process for developing effective health scores using regression analysis:
- Ask customers why they leave, stay or buy more
- Quantify the reasons
- Identify upstream factors and generate testable hypotheses (fit example)
- Collect data and screen variables using factor analysis
- Develop and test predictive models
- Implement, monitor, and continuously improve
Ed noted that human intuition is valuable but it’s subjective. Regression analysis helps overcome this subjectivity by deriving insights directly from data.
Regression: Eliminating the Flaw in Human Judgement
Ed Powers recommended the book “Noise: A Flaw in Human Judgment” by Nobel Prize-winning behavioral economist Daniel Kahneman and his colleagues, Olivier Sibony and Cass Sunstein. Humans think their judgment is accurate, but it’s not the case. The book highlights the variation in that judgment. GTM teams have to recognize that human judgment may have a negative impact on business outcomes. Data-driven approaches instead will achieve more consistent and effective decision-making.
Teams should tailor metrics and variables to suit the specific needs of each business. While there may be common metrics discussed on platforms like LinkedIn, blindly adopting them without considering the unique context of one’s organization can lead to ineffective outcomes. Instead, you should conduct thorough analysis and experimentation to identify the key variables that matter most to their business. This approach ensures that the chosen metrics align with their goals and provide meaningful insights into customer success.
There are also non-traditional indicators of customer churn. For instance, the departure of an executive sponsor without a smooth transition process is a crucial predictor of churn. Identifying these indicators and using regression analysis enables customer success teams to see the early warning signs and proactively address potential issues. By focusing on all accounts collectively, organizations can drive change and improvement on a larger scale, leading to higher customer retention and growth.
To get started, a solid foundation in statistics is important. Tools or data scientists will take care of the complex math for you… but they lack the subject matter expertise to make practical applications. Ultimately, the real challenge lies in solving the data problem, including data acquisition, cleaning, and architecture setup. Once the data is accessible and well-prepared, selecting and implementing suitable regression models becomes relatively straightforward.
Examples of Improved Outcomes Using Regression
Ed Powers, an expert in data science and statistics, shared a customer success story highlighting the significant impact of regression analysis on improving business outcomes. The case involved a tech-enabled service provider working with SMB accounts and facing a churn problem. Powers was brought in to analyze their data and identify key factors contributing to customer churn.
Ed listened to customers first. They expressed concerns about appointment quality, volume, and no-shows. Based on their responses, four key factors were identified: quality, volume of appointments, length of time with the company, and no-shows.
To determine which factors were truly predictive of churn, regression analysis, specifically logistic regression, was applied. Surprisingly, the analysis revealed that quality, which was initially believed to be a major factor, did not significantly contribute to churn prediction. Instead, the volume of appointments and the occurrence of no-shows emerged as the two primary predictors, accurately forecasting 80% of churn behavior.
Based on these findings, a customer health score was developed, enabling the company to identify customers at risk of churn. By focusing on increasing appointment volume, especially in the early stages, and improving customer attendance rates, the company could drive positive business outcomes. The customer health score also provided clear direction for improvement and resource allocation.
This is a great success story that demonstrates the power of regression analysis. However, the implementation of regression analysis may encounter certain roadblocks. One major challenge is gaining access to relevant data and ensuring its accuracy and quality. Data ownership, permissions, and data cleaning processes also present barriers. Nonetheless, by demonstrating quick wins and showcasing the value of regression analysis through a few key factors, organizations can gradually break through these barriers and gain support from leadership.
By applying statistical techniques like factor analysis and logistic regression, companies can identify the most influential factors and develop targeted strategies to improve business outcomes.
As Ed Powers mentioned, the biggest challenge is bringing data together in an analyzable format, and then making it actionable. And that’s where Immersa fits in. Immersa will unify your data, apply statistical methodologies like regression analysis to find hidden risk/opportunities in your customer base, and then work with you to build the workflow to do something based on that data.
Contact us to schedule a free 1-hour consultation. We will review your current data stack and provide immediate feedback on how it can be used.
Make informed decisions using reliable, proven models that will increase customer satisfaction and long-term profitability for your company.