Buhlian

Case Study: NPS Driver Model

Using machine learning to model and identify key drivers for optimization of Onsite Net Promoter Score.

Project Overview

At Miinto, the Net Promoter Score (NPS) is a key tool in monitoring and improving the customer experience. The score is collected on various customer touchpoints - including the order confirmation page, where a survey is prompted with the NPS question and several subtheme questions. Based on the responses to this survey, the NPS score can be calculated and monitored. In addition, the other survey subthemes can be used to identify what is driving the NPS for Miinto customers. This way, it is possible to effectively identify pain areas and prioritize projects.

The Approach & Solution

To model how the different subthemes contribute to the overall NPS performance, I developed a Random Forest model to classify each response into one of the three NPS segments (Promoter, Passive and Detractor) based on the subtheme selections. Based on this model, it is possible to extract the feature importance of each subtheme - i.e. how useful each of them is for the model to make its predicition. From the feature importance, we can clearly identify the two key drivers for the Miinto order confirmation NPS: Navigation and Product Images/Descriptions. Digging further into these subthemes, I was able to identify several, specific projects for optimizing the customer experience - fx changing enhancing navigation, allocate more ressources for optimizing product description and images, tuning the product list filters and solve the onsite search module.

image
image

The Results

Effect
+10
Projects Identified
Effect
+37%
NPS Increase

Based on the model and the subtheme deep dive, +10 pain areas were identified. These pain areas have later on been solved or prioritized accordingly for future work. As a result, the order confrimation NPS has steadily improved - endning with +37% increase in little over 18 months. Thus, this data-driven approach to optimizing customer experience has proven extremely effective.

Want to connect?

Please feel free to reach out, if you are interested in knowing more about me and my projects or want to connect and hear how I might help your organization.