Say you’ve to do something you don’t want to do. Or that you find difficult to do. A complicated, time-consuming chore, maybe. Now, if a third person comes in trying to convince you to do it, are you more likely to listen:
a) If they’re harsh and loud? Or
b) If they’re calm, friendly, and rational?
For most of us, the answer is probably the latter. After all, when we’re faced with a difficult task, the last thing we want is someone berating us and breathing down our necks.
It’s human nature 101 - and that’s why it’s time to apply the same logic to debt collection communication.
Consider this - a survey of customers in the UK revealed that 48 percent would respond to a repayment reminder if the collection message was friendly, helpful, and delivered through a trusted source.
Unfortunately, this is still not the case when it comes to many digital lenders, at least in India. When it comes to collections, a number of digital lending apps have resorted to harassment by contacting borrowers’ relatives and employers, threatening them with legal notices and even manufacturing pornographic content. The mental anguish resulted in a number of suicides across the country.
It’s almost as though come recovery time, lenders forget they’re dealing with real people on the other side. Goes without saying, but it’s time these malpractices are shown the door - for the sake of empathy, but also for better business outcomes.
However, it’s a task that’s beyond what humans are capable of. It requires the right technology interventions - specifically, Artificial Intelligence (AI) and Machine Learning (ML). And as mentioned in blog five of this series, risk assessment processes, powered by AI and ML, play an essential role even at the collections stage.
For starters, they help lenders build nuanced customer profiles based on demographic, social and economic data, through which they can distinguish between those who can’t pay, i.e. don’t have the ability, and those who don’t want to pay, i.e. don’t have the intent.
Of course, the way you communicate with both these segments will differ - as it should. And that’s where the personalization aspects of AI and ML truly come into play.
How do AI and ML enable personalized communication to borrowers?
Back in the day, the only ways to communicate with borrowers were through the phone, in person, and through post. Fortunately or unfortunately, in 2022, things look quite different. Lenders now have several communication channels available to them - but the right one depends on the context and several variables. ML algorithms can analyze these variables - e.g. which channel does the borrower use the most? Where are his response rates the highest? - to suggest the most optimal one.
Since AI and ML algorithms are self-learning, they can analyze text messages, notifications, and phone call scripts to decipher what kind of tone and messaging worked best for different borrower cohorts and on what mediums. For example, one borrower may have responded better to a casual, friendly message on WhatsApp, while another may prefer a polite, to-the-point message on a channel that’s slightly less personal, such as email. As data is continually fed into these algorithms, their recommendations become increasingly precise for better customer engagement.
AI-driven tools such as chatbots and virtual assistants can be utilized across platforms for uniform, omni-channel communication. This ensures the borrower’s experience is consistent and unified. Some advanced AI chatbots can go so far as to analyze and mimic the tone of the borrower’s communication style for higher relatability and effectiveness.
In addition to customizing communication, AI and ML can improve the overall effectiveness of debt recovery in the following ways:
AI-driven data analytics allows collections systems to leverage real-time cash flow data to generate precise insights on their financial wellbeing. Based on this, it identifies borrowers with risk factors such as low balance and other upcoming dues. As a result, lenders can allocate collections resources more effectively, bring down costs, and see an improvement in their Collections Effectiveness Index (CEI) i.e.a measure of the amount collected during a specific period of time against the amount of total receivables during that same period.
Customer segmentation traditionally involves homogenous delinquency buckets that don’t account for individual circumstances. With data intelligence, lenders can personalize these buckets further, making sub-categories within at-risk groups. Lenders can also take a value-at-risk approach to prioritize collections from high-risk, high-balance customers.
In summary, meet people where they are, and speak to them like you would like to be spoken to. Tech-driven data analysis goes a long way in making this happen, distinguishing between borrowers in genuine financial strife and those who are wilful defaulters. At the end of the day, no one wants to be in debt - all it takes to remember that is a little empathy, kindness, and a healthy dash of data!