Our client, a large energy company, was struggling to plan sales and predict customer churn. They had over 2,000 customers and 100+ types of products, data for which resided in CRM, but was not used for any advanced analytics.
The customer approached Vitrin9 to help make sense of all existing data and create a predictive analytics modeling system that could help them measure the probability of customer churn, identify cross-selling opportunities, and predict future opportunities based on various factors such as product type, region, season, commodity prices, etc.
To execute the project, our Data Science team applied the Cross Industry Standard Process for Data Mining (CRISP-DM) model which helped with the data mining and to understand the client's business processes and their data sets. We experimented with different models and after careful evaluation, we selected the Random Forest Classifier as the best model for our client's needs. We used Orange Data Mining along with Python for model building and data preparation.
Our Data Science team worked closely with the client to gather all available data from sources like Salesforce, product raw data, NAICS data, sales financials, etc. We ensured that the data was clean, structured, and ready for analysis.
The system was designed to be easily scalable, providing our client with real-time insights that helped them make data-driven decisions. Our approach helped our client realize over $15M in missed cross-selling opportunities and prevent an estimated $212M in churn in the first 12 months. They were also able to optimize their sales strategy and increase new revenue, resulting in significant business growth.
