The client had only 9% subscription of add-on packs for almost two years without increase.
- Built a suite of ML-based propensity models for each of the add-on packs
- The model arrived at a propensity score for each subscriber to buy those respective packs
- Propensity score was determined by customer’s buying behaviour
- Channel preference model was created to recommend the add-on packs over the customers’ most preferred communication channel
- CRM integration was done to help call centre agents recommend the next best pack to subscribers
10 to 12% uplift in revenues from add-on packs.