Facing high acquisition costs for new leads in personal loans, our client aimed to streamline their marketing campaign expenses for personal loan campaigns.
Overview
Solution
To achieve this objective, we undertook the following approach:
We aggregated customer account information, including factors like the count of different accounts, total loan amounts, maximum credit limits, total balances, past due amounts, active and closed accounts, among others.
New account-level features were created, incorporating metrics such as EMI (Equated Monthly Installments), interest rates, revolving accounts for credit cards, and a balance trade index derived from balances and payments over the last 24 months.
We extracted the most recent date of birth, credit score, and zip code for each customer, based on the latest reported data.
Flags were introduced to identify customers who had taken personal loans within the last 6 months from the date of credit report generation.
Leveraging a machine learning probabilistic model, we analyzed aggregated customer-level data.
We optimized the threshold probability for marketing campaigns through decile analysis.
Output
We achieved a 70% coverage of the target population within the top 3 deciles, effectively identifying potential customers with a high probability of purchasing personal loans.
The cost per prime lead was reduced by half, significantly surpassing the cost reductions set by the original business rules.
