How Artificial Intelligence Is Changing Debt Collection (and What to Do About It)
Debt collection can feel like sifting for gold: you know there is profit to be made, but a lot of time and effort is spent looking in the wrong places. Even before the pandemic, the average collection rate for banks was below 20%.
Artificial Intelligence (AI) promises a more efficient way to locate those golden accounts that yield meaningful profits. However, 75% of finance executives still believe AI is mostly hype, and just 6% of organizations currently use it in their collections process.
In his article, we explain why this is a mistake – and how AI will transform the accounts receivable professionals in the coming years.
Traditional Debt Collection is Not Efficient
Delinquent accounts fall into three buckets: those that will pay voluntarily; those that will pay if they are litigated; and those that are best being sold off. But determining which bucket any given account falls into is far from simple.
Businesses invest significant time and resources in determining which account can afford to pay. Frequently, they rely on obsolete, unreliable, or insufficient data and engage a broad array of third-party vendors to assist their endeavours. All of this occurs even before commencing the litigation or voluntary collection procedures.
As a result, the debt collection process is frequently very inefficient. Firms spend large sums litigating accounts that were never able to pay, and those costs – both in the form of time and resources – eat into their bottom-line profit.
They need a better way to decide which accounts to pursue, and that is exactly what AI offers.
AI Takes the Guesswork Out of Accounts Receivable
Rather than sifting through the sand to find gold, AI enables organizations to pinpoint the best places to dig. Using information about a consumer’s assets, and comparing it with a large dataset of historical payments, AI can determine the likelihood a consumer will repay their debt and how much they are likely to pay.
In one case study, this process found that 70% of a bank’s collections could be made from just 10% of their accounts. This would dramatically increase the overall profitability of their distressed portfolios, both by reducing the litigation costs and avoiding missed opportunities. But there is also a vital time saving component to the process.
While a typical human might manually analyze a handful accounts in an hour, AI can analyze thousands over a few days. This reduces the labor costs involved in debt recoveries and enables more rapid action – which ensures you can achieve profitability more quickly.
Given these clear benefits, the question arises: why don’t more companies use AI in debt collection?
The Great Compliance Barrier
Financial services are heavily regulated, and many leaders believe AI poses a threat to their compliance. A quarter of CFOs explicitly cite “AI management and ethics” as a concern, and 62% of banks say the complexity and risks associated with AI outweigh the benefits. This helps explain why so few firms have adopted AI as part of their recovery strategy.
However, there is a simple solution to these concerns: leveraging third-party tools that have compliance built into their models. At RDS, our AI debt collection software has used machine learning to model existing regulations so that its outputs are all fully compliant, and we regularly update he system to ensure it meets changing standards. As a result, recovery teams can leverage our tools with confidence, knowing they are not taking on any new regulatory risks.
Learn How RDS Uses AI to Enhance Debt Collection
A major component of the compliance barrier is explainability. Firms fear that AI is a “black box” that will produce decisions they cannot justify – and therefore puts them at increased risk. But at RDS, we pride ourselves on increasing transparency for recovery agencies by ensuring every output is explainable and our models are fully compliant with all relevant regulation.
That belief in transparency is also why we created our popular whitepaper on data analytics in debt collection, where we explain exactly how our Machine Learning models work – and how they produce liquidation rates 2-3x higher than competitors.