Analytics or Human Experience? The Answer is Both

As agencies consider using analytics to enhance their collection capabilities, one of the topics which often comes up is whether analytics can replace administrators in the effective operation of their departments. My answer is that there is no replacement for the decades of experience that administrators have. Analytics is a tool that allows strong managers to supplement and enhance their experience, and work in concert to further enhance their operations.  It does not mean losing control.  Rather it means gaining more control.

Defining Analytics

Analytics can mean different things to different people, so it is important to start with a common definition.  In this context, analytics refers to mathematically derived predictive models which help inform automated decisions that can enhance the performance results of your operations.  At the most basic level, imagine it is fifteen minutes before the end of your business day.  A collector has time to make one more call and has many possible cases that he or she could contact.  A predictive model scores the cases and selects the case with the highest potential to be contacted.  Even if a collector might select the same case after careful review of all the options, the predictive model allows the collector to work the best case without spending the time to review all their options.

Utilizing a predictive model does not mean that the manager loses control or gives up the ability to create their strategy.  Predictive models assist managers to enhance their strategies.  For example, while the model may predict an order to work collection cases, the managers must determine the strategy of the workloads which will be pursued, which collector(s) to assign cases to, and the timing and frequency of contacts.

Analytics Provides Fairness

One area where predictive models can support a program is enhancing consistency across the jurisdiction.  If a collector is working a case, a model can determine if the case has been assigned for a sufficient period of time, and if it is ready for reassignment, the score model can determine the likely return based on the age.  Some cases based on historical experience are unlikely to be collected if they have not had a payment for a period of time.  Resources are limited, and while a collector may not want to give up on a case, there may be other consumers which would be a better use of the collector’s time.

However, collector experience should be utilized.  Just because a consumer comes back with a high score, does not mean it is guaranteed to be worthwhile.  The collector or supervisor could discover information about the consumer that is important in the process which could change the collectability higher or lower than the predictive model suggests.

Predictive models can also support a goal of treating similar taxpayers the same.  When an agency relies on staff, even staff with longstanding expertise, then different staff are likely to make different selections.  If actions are informed by models, then it is more likely that similar taxpayers will be treated the same, (e.g., timing for specific collection actions, write-off or offer in compromise decisions).

Analytics allows Managers to focus on Policy and Priorities

Predictive models allow you to manage your inventory in a more holistic manner.  Managers decide what models are prioritized to maximize overall performance.  This could include for example:

• Collection models to determine which cases are most likely to self-cure
• Predictive models to determine the cases most likely to result in increased payment if assigned to a collector
• Predictive models to determine which payment agreements are likely to success
• Predictive models to determine the best timing to pursue activities

Models help you select the best cases.  Management still sets the priorities and policies.  The analytical models act as a tool to increase overall results.  In addition, because the models typically utilize machine learning, they can also adjust their scoring calculations as conditions change.

Change Management Challenges

One important area to consider, is that one of the hardest aspects of incorporating analytics into a collection operation is the associated Change Management.  Some managers are fiercely protective of their processes and decision making.  They need to understand that utilizing analytics and models is adding another tool into “their” toolbelt.  Using scores is not giving up control, but it is taking additional input to enhance their decision making.  They set the policy and strategy, and the model scores are a way of automating their goals.  Once they see it as their models and their tool, they are more likely to embrace analytics to help their organization.

To learn more about FICO’s capabilities in collections, analytics and communications visit https://www.fico.com/en/products/fico-collection-recovery-models and contact or follow me on LinkedIn at https://www.linkedin.com/in/tedlondon/.