AustraliaFinance and insurance2017 – ongoingZip Co
Real-time machine learning credit decision engine
zipMoney ran its real-time credit decisions on machine learning and used language processing on customer messages
Problem
zipMoney, now Zip Co, offered interest-free financing and had to decide on credit applications in real time as customers and merchants grew quickly.
Approach
The company built a credit decision engine with machine learning at its core, a feature it had since its early days. Its chief technology officer said several models ran at the same time so they could be validated before being switched on. The company also used natural language processing to analyse and categorise customer messages, aiming to automate query resolution while keeping a human feel. It was testing chatbots for customer care.
Results
No published figures
Takeaway
Running candidate models alongside live ones let the company check them before activation. The interview gave no accuracy, approval or loss figures for the models.
Categories
- Industry
- Finance and insurance
- Business function
- Finance, risk and legalCustomer service
- Innovation type
- Process innovation
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
An interview with zipMoney's chief technology officer. It says machine learning sits at the core of the real-time credit decision engine, that several models run concurrently for validation, and that language processing is used on customer messages. It gives business figures such as over 400,000 customers but no model metrics.
Full details are in the original sources.