Will Machine Learning Revolutionize Healthcare (and other) Collections?

The intersection of behavioral psychology, classic marketing strategy and tech for the masses—including machine learning—has added tremendous value in the business world at large. We’ve long known what motivates behavior, how that translates to the purchase journey, and how to use computers to focus that intelligence. You see this kind of innovation used everywhere…except the collections industry.

Prime time for change

Some will argue that if it ain’t broke, don’t fix it. But by the numbers, the old collection agency model of sending letters, followed by dialing for dollars is more than ready for change. Partly, this is because the old model is not built around today’s tap & click consumers. The business is overdue to invest in the idea that debtors are still customers (except they haven’t paid their bills). A “payment journey” designed specifically for them is working, especially when it’s supported by machine learning.

Machine learning uses data to predict behavior, and it helps deliver custom messaging (in the right way and at the right time) in order to encourage a desired future behavior. Using machines to “learn” who debtors really are, what their position is, how they behave and what motivates them, we can determine what is most likely to encourage them pay. Machines can make it easy to develop payment terms (with limits that are coded according to issuer/lender/provider preference) that work for each customer’s unique situation. They help deploy tactics that nurture the relationship with carefully curated, compliant and turnkey communications. Machines don’t fail to protect and secure sensitive consumer information, and they don’t inadvertently blank on compliance protocols. Identifying which debts are most likely to be repaid also enables agencies to focus resources most effectively. Machine learning is allowing even small agencies to scale operations and get more done in less time.

Innovators on the Scene

New companies are using big data to learn not just about the delinquency, but about the whole customer lifecycle. Innovative start-ups are using this information to bring the collections business online. TrueAccord, whose CEO, Ohad Samet, was recently selected to represent the ARM industry on the CFPB’s Consumer Advisory Board, is a few years into its digital collection model debut. And it’s gaining traction. Another newer player, Collectly -- a collection service for healthcare providers and medical billing companies -- uses aggregated machine intelligence to forecast behavior and adjust each patient’s collections journey to encourage the best outcome. According to Collectly, their recovery rate is about twice that of traditional collection agencies.

Transformation as Opportunity

Does all this spell the end of the collection call center as we’ve so far known it? From an operational perspective, making it easy to tap away debts in the middle of the night, or set up a payment plan, make changes to an existing plan, and explore other options online has the potential to transform the call center into more of a benevolent resource for the relatively few customers who prefer live contact. 

As the laws change, there is less of a learning curve for call center agents; new rulings or other changes and regulatory guidance can be easily coded into an array of materials, digital properties and outreach communications, and the customer experience is adjusted seamlessly. This can also mean less of a burden for call monitoring and compliance departments. Part of the tremendous upside is the elimination of waste and worry the industry faces today around hot-button issues like on-call identity authentication, consent, auto dialers, and dispute resolution protocols (or lack thereof). There is simply less exposure and less risk to manage with a call center that’s supportive, instead of central, to a collections operation.

The tight margins and regulatory environment of the collections business make it a great candidate for the use of machine learning to improve operational efficiency and manage risks.

What’s missing here is the regulatory framework to allow digital collections to fully scale. Privacy concerns and concern of major creditors about negative exposure prevents many from fully embracing new technologies.

We’ll keep watching this space to see whether innovation and regulation can find a reasonable place to meet.