The Challenge.
- An e-commerce client had no visibility on how to target higher-value customers who had a higher chance of purchasing from them.
- This led to a lot of one-and-done customers and low-value customers.
- There was no targeting of customers based on predictive analytics, nor was there a data-driven understanding of which factors impact a customer's probability of purchasing more than once.
The Solution.
- Designed and built a propensity to purchase model to identify customers with a high probability of purchasing within the next 30 days every month.
- Derived feature importance to identify which factors impact a customer's probability to purchase.
- Segmented the customer base each month into low, medium and high probability of purchasing to customize targeting.
The Results.
- List of customers with probability to purchase was used by the marketing team to target a subset of the customer base each month and optimize the marketing budget.
- Targeted campaigns led to up-selling of customers with medium and high probability of purchasing.
- Understanding the factors that influence purchasing behaviour led to better profiling of customers and increased customer loyalty.