Predicting churn
Our client wanted to improve their customer service and sales efforts making it more effective in obtaining and retaining subscribers. They handed over a very large customer database of historically active and inactive subscribers as well as a large set of customer characteristics.
The challenge
The client wanted insights on factors that caused customers to cancel their subscriptions. For PwC, the main challenge was to identify churn patterns in a dataset of approximately 1 million active and inactive subscriptions along with several characteristics of each subscription.
The outcome
We communicated two key outputs from our model to the client. First, the model can assess the probability of cancellation for each of the active subscriptions today. Second, the model highlights patterns in customer behaviour. One key insight was that the first 100 days was key to building customer loyalty.
Contact us
