Machine Learning and Predictive AI
AI portfolio management
A decision engine that analyzes the history of millions of collection attempts to determine which accounts to work, when to contact them, through which channel and with what priority.
The problem
Collections organizations have to decide every day which accounts to work, when, through which channel and with which resources. When those decisions are made by hand, they depend on the experience of a few specialists, eat up hours of analysis and often lead to inefficient allocation of portfolios, agents and contact channels.
Who it’s for
- Banks
- Lenders
- Retail
- Insurance
- Telecommunications
- Companies with large data volumes
Why predictive management
Risk, propensity and scoring models are trained on the activity history stored in MARA, not on someone else’s data.
It takes demographic and behavioral data and sets the WhatsApp, SMS and email strategy that maximizes the likelihood of payment.
In about two months the engine gets to know a newly placed portfolio and adjusts the strategy on the fly.
Models that propose the payment plan most likely to be kept, even on severely delinquent portfolios.
Before and after
Before
- Manual assignments
- One-size-fits-all strategies
- Lower effectiveness
- Higher risk
After
- Automated predictions
- Smarter prioritization
- Lower risk
- Higher operational return
In action
Every day the engine decides which accounts to work, when and through which channel: agent occupancy reached 85% with the same contact effectiveness as the experts’ strategy.
- 85% agent occupancy
Next step
Predictive management is built on the data MARA, the chatbot and the voice agent already generate; the dashboards show its impact.
- Collections CRMMARA
- AI chatbotARIA
- AutomationRPA
- Voice agent
- AI portfolio You are here
- DashboardsMERY Next