Toyota financing, customer renewal study
Predicting who comes back
Toyota Renewals
For Toyota financing clients, I studied renewal behavior among auto loan customers: clustering profiles, analyzing feature importance, and building ML models with feature selection to predict who would renew and help the team tailor outreach accordingly.
The context
Toyota's financing arm serves customers on auto loans with renewal opportunities at contract end. Understanding who renews, who churns, and which factors drive that decision is critical to retention strategy and communication planning.
The challenge
Renewal decisions depend on a mix of financial, behavioral, and contractual signals spread across customer records. The team needed more than descriptive stats: actionable segments, interpretable drivers, and predictive models tight enough to guide targeted communications.
- Heterogeneous customer profiles with no clear renewal archetypes
- Many candidate variables, unclear which ones actually matter
- Need for interpretable insights, not just a black-box score
- Communications strategy requiring renewal probability at individual level
What I delivered
From exploratory clustering to feature importance and renewal prediction models built for operational use.
Customer clustering
Applied clustering to group financing customers by shared renewal patterns and behavioral profiles, revealing distinct segments with different retention dynamics.
Feature importance analysis
Identified which variables most influence renewal decisions, giving the business interpretable levers beyond model scores alone.
Renewal prediction with feature selection
Built ML models with feature selection to predict renewal probability per customer, balancing predictive power with a parsimonious, explainable feature set.
Communication targeting
Translated model outputs into renewal risk tiers and segment-level insights to adapt messaging, timing, and offers for customers approaching contract end.
Business impact
The renewal study gave marketing and customer teams a structured basis for retention actions, from broad segments to individual-level scores.
Actionable customer segments
Clustering revealed renewal archetypes the team could address with differentiated strategies.
Interpretable drivers
Feature importance highlighted what moves renewal decisions, supporting both modeling and business conversation.
Individual renewal scores
Predictive models ranked customers by renewal likelihood to prioritize outreach.
Targeted communications
Risk tiers and segment insights informed how and when to contact customers before contract end.
Tech stack
Python, BigQuery, Tableau for exploration and reporting
Python
BigQuery
TableauToyota Renewals turned a retention question into a practical toolkit: segments that make sense, drivers the business can discuss, and scores that help prioritize who to reach before they leave.