Debt collection and being friendly don’t go together, do they? But what if this was possible? Like other aspects of the BFSI industry, debt collection too has been disrupted by digital, with AI/ML tools sifting through massive amounts of data such as customer info, financial documents, competitor analysis, and regulations. Data allows lenders to better segment their customers, drive down costs, and improve the collection process. Innovative banks worldwide are using data and AI to unlock key customer insights, mark improvement areas, and change how debt collection works.
Challenges in Debt Collections
Getting results without damaging customer experience
Customers have high service expectations from any service provider. Personalised communication, prompt responses via instant messages, social media etc., quick action to complaints, and other such seemingly small efforts can go a long way in how customers perceive the brand. Similarly, an intimidating or ugly collection experience can result in disgruntled customers, even if their experience at the time of loan origination was pleasant. Collection officers struggle with ensuring they maximise collections without impacting customer experience or undermining customer acquisition efforts.
Following linear processes
Most collection departments work with linear prioritisation of collection and borrower profiling methods. Their follow-up actions are determined by customer responses, without reading the customer’s true intent. In the absence of technologies like NLP, collection officers cannot ascertain the anomalies and patterns in a borrower’s behaviour, thus increasing their strike out rate.
Expensive nature of collections
Collecting dues has always been tricky, involving relentless calls, soft persuasion, and even cold-hearted compelling follow-ups. At the same time there are financial targets to achieve. Many customers tend to default or delay payments. And their challenges have been further intensified by financial difficulties caused by the pandemic. According to the Business Insider, 14.7 Mn credit card customers in the US have defaulted in 2020, while for mortgage, the number is at 2.7 Mn. Lending firms usually have an in-house collection management team. They can also outsource the entire collections process to a third-party vendor or engage them to augment the internal team. In all cases, collections management is expensive to operate, and is made more complicated by regulatory and other risks.
How AI/ML Improve and Simplify Debt Collection
With bank operations getting digitised, financial firms can now access a variety of customer data. AI/ML algorithms can analyse this data to cull out insights that can help optimise collections, help reduce delinquency, enhance the ability to recover debt, and maintain a good customer experience. Data analysis optimises operations and reduce cost of debt recovering, helping agents become more productive and effective. Let’s understand in greater detail how AI technologies will transform the lending and debt recovery industry in 2023.
Predict customer delinquency
Forewarned is forearmed. From analysing a customer’s spend history, consumption patterns, income history, to credit repayment history and parameters like FICO score (where applicable), AI tools can understand a borrower in detail. Banks can even consider the impact of external factors like micro- or macro-economic changes, weather fluctuations, customer demographics etc., on a borrower’s ability to repay debt. According to McKinsey, AI-enabled banks are better placed to engage with customers and ensure they stick to repayment schedules. Some lenders use AI-powered decision/ game theory or MinMax logic along with statistics and psychology to minimise possible defaults. They use scores to determine collection strategies to accurately predict collections as well as cost of debt recovery.
Improve risk assessment
One huge advantage of applying AI to debt recovery is that it allows lenders to take faster, better credit decisions for borrowers of all kinds. A continuous learning model makes it easier to bucket customers into different segments and estimate their risk profile. This helps underwriters approve and reject loans based on predictive scores thus reducing non-performing loans by looking at accurate predictive risk models.
Debt recovery is perhaps one of most hated roles in the market. Ironically, AI can help bring back the human element into debt recovery and improving customer response. By using AI/ML tools, banks can:
- Choose which communication channel works for which customer segment
- Pick the appropriate tone and sentiment to solicit the desired response
- Deliver personalised communication at the right time
- Drive more recoveries by offering personalised advice and insights
Accurately profile borrowers
In difficult economic scenarios – like the COVID-19 pandemic which has highlighted the nuances that exist in the BFSI industry – a deeper understanding of borrowers becomes even more critical for effective debt collection. With AI, banks can build equally nuanced borrower profiles to understand who can default, who will pay on time, and who will need proactive intervention.
Improve customer engagement
Phone calls have always been the most common tool that collection officers use to follow-up with borrowers. Today, banks have access to multiple communication channels like in-person meetings (where possible), emails, social media channels, text messages, web channels, etc. But debt collection is not just about having a plethora of channels to reach borrowers. It’s about choosing the right channel at the right time with a high impact message. For instance, by analysing the activity on the bank’s app and website, they can identify the right time to call a customer. And as client data is fed into the AI tool repeatedly, the customer engagement process becomes increasingly personalised and effective.
Apply behavioural science
Younger lending customers – Millennials and Generation Z – are looking for more innovative ways to complete payments. This means no phone calls during working hours, or multiple letters asking for a repayment with increasing urgency. AI help banks make smarter choices on when and how to contact customers based on their behaviour and other aspects such as demographics and socio-economic data. In case of a customer refusing to pay, legal action can be initiated, which can be an expensive proposition. AI can help suggest alternative repayment models and plans tailored to each borrower’s financial ability to pay.
Reduce cost of debt recovery
In-house collection teams and outsourced call centres to recover debt are very expensive to maintain, especially when they are not very effective. AI-powered bot agents are a cheaper way to communicate with customers via their preferred channel and when it is convenient for them. Bot agents can be trained to gather data from sources such as social media, financial records, and insurance history, and fine-tune their responses based on individual profiles. This allows the bots to suggest relevant repayment options to customers, recommend plans and give advice if required, eliminating the need for underwriters or collection agents.
Lending, be it individual or institutional, has always been risky, with delinquencies, defaulting and inefficiencies complicating the mix. AI/ ML combined with Big Data can help banks make debt recovery more efficient, cost effective, and more successful. According to Business Insider, the application of AI/ ML could help the banking industry save USD 447 billion by 2023, and up to 1 trillion by 2030. Sadly, just 7% of banks have adopted digital transformation successfully. Early adopters of these technologies are already seeing results and are ushering in a new era in debt collection. An era of early delinquency warnings, more relevant categorising borrowers, and optimised customer engagement.