Serious delinquencies have climbed to 12.31%, and the collection models most institutions rely on were built for a fundamentally different environment. These models were designed when phone calls were the primary contact channel, when consumer behavior was more predictable, and when portfolios were more homogeneous. The assumptions baked into these models no longer hold, and the gap between model expectations and actual performance widens with each passing quarter.
A regional bank we worked with illustrates the problem. The institution was operating 11 legacy systems across its collections operation, each feeding different data into different models at different intervals. Agents were making contact decisions based on scores generated from incomplete data, and the segmentation strategy had not been updated in over three years. The result was wasted effort on low-probability accounts and missed opportunities on accounts that were curable with the right intervention at the right time.
The solution required three interconnected capabilities. Predictive contact optimization uses machine learning to determine the best channel, time, and message for each account based on the customer's historical response patterns and current behavioral signals. Dynamic customer segmentation replaces static risk tiers with continuously updated groupings that reflect real-time changes in payment behavior, financial capacity, and engagement likelihood. Automated decision intelligence removes the latency between insight and action, pushing optimized contact strategies to agents and digital channels without waiting for manual review cycles.
GOBLIN serves as the platform that enables these capabilities by consolidating the fragmented data streams that feed collection models into a single, real-time analytical environment. When models can draw on complete customer data across all products, channels, and touchpoints, their accuracy improves dramatically. More importantly, the platform enables continuous model recalibration, ensuring that collection strategies adapt to changing consumer behavior rather than relying on assumptions from a previous economic cycle.
The institutions that will manage through the current delinquency environment successfully are those that treat their collection models as living systems requiring constant refinement, not static tools that can be set and forgotten. The data and technology to do this exist today. The question is whether leadership is willing to invest in the infrastructure that makes it possible.




