SUCCESS STORY — TOYOTA FINANCIAL SERVICES
Toyota Financial Services — Anticipating contract renewal with machine learning
To move from reactive end-of-contract handling to proactive retention, KPC built a renewal prediction model for TFS, developed and deployed inside Snowflake, with an explainable score for every contract.
The impact in numbers
76%
Overall model accuracy
83%
Of non-renewing customers identified
8
Enrichment stages on the study dataset
The impact in numbers
76%
Overall model accuracy
83%
Of non-renewing customers identified
8
Enrichment stages on the study dataset
The impact in numbers
76%
Overall model accuracy
83%
Of non-renewing customers identified
8
Enrichment stages on the study dataset
RESULTS
Three concrete advances for Toyota Financial Services
A reliable model, built around retention
On a test sample the model had never seen: 76% overall accuracy, and 83% of non-renewing customers correctly identified.
Three concrete advances for Toyota Financial Services
The key factors behind the decision, identified
The model confirms that geography and financial terms (rate, deposit) are the main determinants of the renewal decision.
Three concrete advances for Toyota Financial Services
A decision-support tool for the business teams
Every active contract gets a probability of non-renewal and an automatic explanation of the factors behind it.
RESULTS
Three concrete advances for Toyota Financial Services
A reliable model, built around retention
On a test sample the model had never seen: 76% overall accuracy, and 83% of non-renewing customers correctly identified.
Three concrete advances for Toyota Financial Services
The key factors behind the decision, identified
The model confirms that geography and financial terms (rate, deposit) are the main determinants of the renewal decision.
Three concrete advances for Toyota Financial Services
A decision-support tool for the business teams
Every active contract gets a probability of non-renewal and an automatic explanation of the factors behind it.
RESULTS
Three concrete advances for Toyota Financial Services
A reliable model, built around retention
On a test sample the model had never seen: 76% overall accuracy, and 83% of non-renewing customers correctly identified.
Three concrete advances for Toyota Financial Services
The key factors behind the decision, identified
The model confirms that geography and financial terms (rate, deposit) are the main determinants of the renewal decision.
Three concrete advances for Toyota Financial Services
A decision-support tool for the business teams
Every active contract gets a probability of non-renewal and an automatic explanation of the factors behind it.
Background & core challenges
Moving from end-of-contract as something that happens to you to proactive retention
Toyota Financial Services France
For TFS, the life cycle of a vehicle financing contract is a key moment in the customer relationship: the decision to renew is a major indicator of loyalty to the brand. Every non-renewal means lost revenue and the risk of the customer moving to a competing brand.
Already running the imagino CDP, TFS wanted to work the data to extract scores from it, starting with renewal.
Predict the renewal of each contract at maturity
Identify the key factors behind the decision
Turn the analysis into recommendations for marketing, sales and customer service
Prioritize retention actions with a risk score
Background & core challenges
Moving from end-of-contract as something that happens to you to proactive retention
Toyota Financial Services France
For TFS, the life cycle of a vehicle financing contract is a key moment in the customer relationship: the decision to renew is a major indicator of loyalty to the brand. Every non-renewal means lost revenue and the risk of the customer moving to a competing brand.
Already running the imagino CDP, TFS wanted to work the data to extract scores from it, starting with renewal.
Predict the renewal of each contract at maturity
Identify the key factors behind the decision
Turn the analysis into recommendations for marketing, sales and customer service
Prioritize retention actions with a risk score
Background & core challenges
Moving from end-of-contract as something that happens to you to proactive retention
Toyota Financial Services France
For TFS, the life cycle of a vehicle financing contract is a key moment in the customer relationship: the decision to renew is a major indicator of loyalty to the brand. Every non-renewal means lost revenue and the risk of the customer moving to a competing brand.
Already running the imagino CDP, TFS wanted to work the data to extract scores from it, starting with renewal.
Predict the renewal of each contract at maturity
Identify the key factors behind the decision
Turn the analysis into recommendations for marketing, sales and customer service
Prioritize retention actions with a risk score
Our approach
A data science programme in three stages
A structured approach following the CRISP-DM standard, from building the data foundation through to deploying the model in production.
Data science foundation
A study dataset built in Snowflake
Building a study dataset through successive enrichments: contract and vehicle, financial data, customer history and services taken, interactions with customer service and sentiment analysis. Each stage is materialized to guarantee traceability and reproducibility.
Exploration & modeling
A robust XGBoost model, with no data leakage
Exploratory analysis to audit data quality and rule out the traps: information leakage, multicollinearity, the risk of geographic overfitting. Creating business indicators (financial ratios, rate gap against the local market), then an XGBoost model optimized by cross-validation.
Deployment & explainability
An explainable score for every active contract
An automated inference pipeline runs directly inside Snowflake on contracts coming up for maturity. Thanks to SHAP values, every score comes with an explanation of the factors behind it, readable by the business teams.
Our approach
A data science programme in three stages
A structured approach following the CRISP-DM standard, from building the data foundation through to deploying the model in production.
Data science foundation
A study dataset built in Snowflake
Building a study dataset through successive enrichments: contract and vehicle, financial data, customer history and services taken, interactions with customer service and sentiment analysis. Each stage is materialized to guarantee traceability and reproducibility.
Exploration & modeling
A robust XGBoost model, with no data leakage
Exploratory analysis to audit data quality and rule out the traps: information leakage, multicollinearity, the risk of geographic overfitting. Creating business indicators (financial ratios, rate gap against the local market), then an XGBoost model optimized by cross-validation.
Deployment & explainability
An explainable score for every active contract
An automated inference pipeline runs directly inside Snowflake on contracts coming up for maturity. Thanks to SHAP values, every score comes with an explanation of the factors behind it, readable by the business teams.
Our approach
A data science programme in three stages
A structured approach following the CRISP-DM standard, from building the data foundation through to deploying the model in production.
Data science foundation
A study dataset built in Snowflake
Building a study dataset through successive enrichments: contract and vehicle, financial data, customer history and services taken, interactions with customer service and sentiment analysis. Each stage is materialized to guarantee traceability and reproducibility.
Exploration & modeling
A robust XGBoost model, with no data leakage
Exploratory analysis to audit data quality and rule out the traps: information leakage, multicollinearity, the risk of geographic overfitting. Creating business indicators (financial ratios, rate gap against the local market), then an XGBoost model optimized by cross-validation.
Deployment & explainability
An explainable score for every active contract
An automated inference pipeline runs directly inside Snowflake on contracts coming up for maturity. Thanks to SHAP values, every score comes with an explanation of the factors behind it, readable by the business teams.
Our clients
Our clients
They trust us.
They trust us.
Equipment rental
Professional Equipment Rental Company | Industrialized tender response 100% Snowflake
Gen AI
Snowflake
→
Transport & Logistics
Global shipping company — Data Science: from R&D to industrialization
AI
Data
→
Industry
Aqualung Group — Transforming the data platform
Snowflake
Data Platform
→
Business services
APAVE — Cross-office capacity planning
Pigment
EPM
→
Organic food distribution
Biocoop — Steering the forecasting processes with SAP BPC
SAP
Retail & Distribution
→
Heating, plumbing and renewable energy distribution
Richardson — Opening up access to data by moving to the cloud
Qlik
Snowflake
→
Equipment rental
Professional Equipment Rental Company | Industrialized tender response 100% Snowflake
Gen AI
Snowflake
→
Transport & Logistics
Global shipping company — Data Science: from R&D to industrialization
AI
Data
→
Industry
Aqualung Group — Transforming the data platform
Snowflake
Data Platform
→
Business services
APAVE — Cross-office capacity planning
Pigment
EPM
→
Organic food distribution
Biocoop — Steering the forecasting processes with SAP BPC
SAP
Retail & Distribution
→
Heating, plumbing and renewable energy distribution
Richardson — Opening up access to data by moving to the cloud
Qlik
Snowflake
→
Equipment rental
Professional Equipment Rental Company | Industrialized tender response 100% Snowflake
Gen AI
Snowflake
→
Transport & Logistics
Global shipping company — Data Science: from R&D to industrialization
AI
Data
→
Industry
Aqualung Group — Transforming the data platform
Snowflake
Data Platform
→
Business services
APAVE — Cross-office capacity planning
Pigment
EPM
→
Organic food distribution
Biocoop — Steering the forecasting processes with SAP BPC
SAP
Retail & Distribution
→
Heating, plumbing and renewable energy distribution
Richardson — Opening up access to data by moving to the cloud
Qlik
Snowflake
→
A DATA SCIENCE PROJECT IN MIND?
What if you could see your customers leaving before they do?
KPC supports you from building the data foundation through to putting explainable predictive models into production, directly inside your Snowflake platform.
A DATA SCIENCE PROJECT IN MIND?
What if you could see your customers leaving before they do?
KPC supports you from building the data foundation through to putting explainable predictive models into production, directly inside your Snowflake platform.
A DATA SCIENCE PROJECT IN MIND?
What if you could see your customers leaving before they do?
KPC supports you from building the data foundation through to putting explainable predictive models into production, directly inside your Snowflake platform.