Home>Upsell and Churn Modeling

Upsell and Churn Modeling in the Timeshare Industry

Client

A Vacation Timeshare Company Headquartered in Nevada ​

Business Problem

Our client was looking at innovative methods to improve their upsell rate and reduce churn, which was at an average of 2% and 13.5% respectively

Objective

Use ML and DL techniques to help the client’s marketing team:
– Identify high-value and high-risk customers
– Recommend right services along with the propensity of purchase for optimum up-sell​

Abzooba’s Solution:

  • Develop baseline and challenger models
    – Baseline: Logistic Regression
    – Challenger: Random Forest; XGBoost
  • Evaluate the performance to:
    – Predict the probability of churn for each customer
    – – Identify high-value and high-risk customer
    – Predict the probability of up-sell for each customer
    – – Recommendation engine for services
    – – Propensity to purchase
  • Test and fine-tune models to improve accuracy
  • Based on the input from models, conduct A/B testing on real-time customers to measure effectiveness
  • Incorporate the feedback from A/B testing into the model for reinforcement learning

    Business Benefits:

    For the first part of the journey, over 18 months, we have increased the upsell to 10% and reduced churn to 9%, helping our client to generate additional business worth $50k – $80k​

Speak to AI expert