Household debt is at a staggering $48 trillion globally with sharp growth in both Norway and China. Effective debt collection will become increasingly important as default rates rise because of the economic slowdown.
Traditionally, collections practices have been mostly manual, batch based and reactive, with the number of touchpoints and communication increasing as delinquency ages. However, with today’s increasingly digital consumer, a manual collections process isn’t sustainable.
Now, collections need to become more automated and intelligent than ever before to improve promise to pay rates and reduce default rates. While this may seem like a daunting task, artificial intelligence and cloud technologies can help.
Improving promise to pay rate with better communication
The collections process typically involves various steps depending on which stage of the collections process the customer is in, ranging from pre-delinquency (usual reminder notes) to past due (late fees, credit bureau reporting), and finally late stage collection/write off (which usually involve a third party debt collection agency).
One of the major challenges is a real time view of the customer’s activity so that communication can be optimized. The amount of data exchanged can quickly become unwieldy and hard to manage if processing isn’t automated.
Finally, debt collection can be a very sensitive issue, and customers don’t want to be handled like numbers or criminals – they want personalized communication that takes into account their circumstances for the debt, such as loss of job from the COVID crisis. By analyzing demographic, behavioral and transactional data, businesses can provide tailored communication and payment plans to the customer. This improves customer response and ultimately promises to pay.
Adopting a proactive approach to delinquency
Machine learning with real time information can be a powerful tool to predict delinquency. Assessing internal transactional data from the customer as well as external information enables lenders to better predict who will default and deploy preemptive strategies to reduce default rates.
For example, deploying these predictions with real time data, lenders can be both more proactive with their communications, but also potentially take steps to adjust lending terms based on the current situation of the customer to prevent a default on the loan. Similarly, a model can be used to predict the likelihood and even amount of the repayment. It’s the speed and intelligence that can be a step change for lenders.
Unlocking customer data to make better decisions
In order to provide a personalized experience, we need to bring together the data for that customer, to get as much detail about their past history, current behavior and predictions as to what they may do in the future. For this, the right integration tools are required to connect data from these disparate sources.
Oftentimes, machine learning models are used to provide insights from the data to create the best experience under the current circumstances for the customer. Data models change as behaviors change, so over time having an open, flexible and cloud-native toolset enables a best-of-breed approach that helps data scientists make the most out of the data.
It is also important to understand that, in addition to the data intelligence, the best judgment of a knowledge worker is also critical in ensuring the system can face the changing market needs. Use of an optimal decision tool that gives them the transparency they need, with the ability to easily modify decisions are key to staying on top of policy changes and refinements.
Taking an open source approach
Open source has emerged as the prominent way software is created across the world. It is powering the digital revolution and has accelerated innovation in both cloud and artificial intelligence technology. These projects are managed in foundations that are designed to protect the intellectual property for both the contributors and consumers.
The Apache Kafka and Apache Spark are two notable open source communities in the Apache Foundation that provide the ability to stream and analyze large datasets in real time. They are both essential tools in any organization’s machine learning toolkit. The Cloud Native Foundation with communities such as Kuberntes, Isitio and Kogito, has likewise been the driving force behind cloud technology.
Open source is also about putting you in control. It gives you the ability to choose from a range of partners who include these capabilities in their collection platform or commercial vendors who provide training and support for the components you need. It is your choice on whether you want a more hands off approach or want to co-create together.
Flourishing in the new normal
Embarking on a roadmap to update both collection processes and the technologies that support it will be critical, as an increasing number of customers default on their debt. Fortunately debt collectors can take incremental steps today to improve overall collection performance. Cloud and artificial intelligence technology can help.
