About the Project

A fast-growing e-commerce marketplace was looking to quantify and enhance customer lifetime value across its user base. Existing methods were static, based on averages, and failed to capture the nuances of customer cohorts, churn probability, and changing purchase behavior.

The client needed a predictive CLTV engine that could model value at an individual customer level and enable strategic activation across loyalty, marketing, and retention programs.

Key Issues

• CLTV models based on outdated static rules, ignoring behavioral evolution

• Inability to tailor retention and loyalty interventions based on predicted value

• Fragmented systems limited integration between CLTV scores and action workflows

• Lack of an adaptive framework to update CLTV in response to market conditions

Our Approach

ML-Powered Dynamic CLTV Engine

We built a predictive, cohort-aware CLTV platform using advanced machine learning models, updated iteratively to reflect changing customer lifecycles and monetization patterns.

• Data Engineering: Unified customer, transaction, engagement, and churn probability signals

• Model Development: Built a 12-month predictive CLTV model factoring in churn likelihood, transaction frequency, AOV, and margin

• Activation Strategy: Segmented users into value tiers and integrated insights with loyalty, CRM, and campaign platforms

• Program Design: Designed and launched a "Power Users" loyalty program for top decile users offering early access, exclusive perks, and delivery benefits

• Feedback Loop: Regularly refreshed CLTV scores with new behavioral inputs and campaign responsiveness

Impact