All Case Studies

Customer Churn Prediction

A Fortune 1000 company providing financial technology services across 32 countries wanted to predict probability of attrition for network merchants.


Reduction in attrition rate

Reduced overall merchant attrition rate by 3%-9%


Prediction accuracy

Introduced predictions with up to 83% accuracy using supervised learning 

Identifying attrition causes

Identified factors influencing merchant churn

Bitwise - Customer churn prediction


In today’s highly competitive business environment, merchant or customer attrition is one of the most prevailing challenges, mainly due to the lack of tangible and timely insights into what’s making the customers churn. One of our clients, a Fortune 1000 company offering financial services globally, chose to get ahead of the problem by leveraging prediction models to predict churn probability amongst network merchants. Bitwise designed an efficient and robust predictive model that helped the company stay on top of the issue by accurately predicting the churn probability and gaining insights into factors leading to attrition.


When it comes to solving a customer churn issue, the key challenge lies in identifying the scope of the issue before taking any decisive resolution measures. Being a long-standing business with a significant global presence for years, the company had large reserves of data with insufficient information on factors influencing merchant attrition. This meant the technology solution had to intelligently encompass, define and curate the influencing factors before building an actual prediction model. 

Extensive collaboration with key decision-makers in the company led to identifying and focusing on the following challenges: 

  • Massive amount of non-curated data causing unreliable assumptions 
  • Lack of reliable information to identify factors affecting churn 
  • No concrete definition of merchant attrition 
  • Sedate increasing attrition in the financial industry 


The company had been operating globally for years with complex data coming in from diverse channels. They needed a technology partner with deep expertise and experience in handling such data and collaborating with diverse stakeholders to build the right solution, well-fitted to their requirements.  

The core advantages that were considered for a value partnership: 

Our approach: “Supervised Learning Classification 

Bitwise created a churn prediction model for the company which enabled them to make reliable predictions for a merchant’s propensity to churn. The implementation of the solution was as follows: 

  • Identification of characteristics influencing merchant attrition was crucial to building an effective churn prediction model. Bitwise used merchant transactional data, demographic data, and pricing information to gain insights and plan accordingly. 
  • Bitwise leveraged the huge amount of data to create a concrete definition of merchant attrition by successfully using data exploration techniques
  • The next step was training and testing different models using supervised learning methods with multiple classification algorithms such as logistic regression, random forest, support vector machine, decision tree, Naïve Bayes etc. 
  • On-demand predictions were made by the team using the end-to-end pipeline on Google Cloud Composer(Airflow). 
  • The final step was integration of the predictive model with Google Data Studio for interactive reports which generated actionable insights. 

How Bitwise solution aligned with the client vision 

The diligently designed and vigorously tested prediction model by Bitwise helped the company to accurately identify churn-probable merchants and made it easier to plan and execute targeted retention campaigns. Furthermore, it gave the client great visibility into factors impacting revenue in near future and avenues to prevent that.  

The Bitwise solution led to following advantages as envisioned by the company: 

  • Increased accuracy in predicting merchant attrition rates 
  • Better visibility into factors affecting attrition 
  • Refined customer retention strategies with targeted campaigns 
  • Reduced revenue loss 
  • Improved customer service  
  • More cross-sell & up-sell opportunities 

Transformation that followed 

During heavy customer acquisition phases, many businesses tend to put off resolving churn issues amongst existing customers until it visibly starts affecting the revenue streams. But in a competitive landscape where retaining the market share is of paramount importance and much easier than carving a new piece of customer segments, investing in customer retention pays off in multiple ways. 

By efficiently leveraging prediction analytics, the client company was able to better see within its own processes and systems and spot opportunities to protect its revenue streams and strengthen them for years to come. Using prediction model designed by Bitwise, they successfully reduced merchant attrition while simultaneously tapping cross-sell & up-sell opportunities, adding value to customer experience and increasing customer loyalty which paid off the investment in no time.


Relevant Case Studies

This is a center of discovery and innovation where businesses can enable themselves with the latest insights and perspectives from technology experts to stay relevant in a competitive environment.


Data Fabric: A unifying approach to enterprise data

Perspective | August 20, 2020

Data Fabric: A unifying approach to enterprise data

According to a latest insight by IDC, data is growing at a phenomenal rate. During…

Read more

Self-service BI and Dashboards

Perspective | August 7, 2020

Self-service BI and Dashboards

The rapid advancement of technology over time has propelled the evolution of Data Warehouses and…

Read more


Perspective | July 15, 2020


No one would have predicted a year ago that we would have to go through…

Read more