UK mortgages experienced a significant surge in arrears during the final quarter of 2023, reaching a seven-year high with 93,680 homeowner mortgages in arrears, as reported by UK Finance. Similarly, data from the Finance and Leasing Association (FLA) revealed a two-year high in car repossessions in May 2023. This trend is attributed to high interest rates and a persistent cost-of-living pressure, indicating a likelihood of more debtors struggling to meet payments in the near future.
Unfortunately, lenders are ill-prepared to address the anticipated increase in arrears and provide proactive support to customers, not only to manage their financial obligations but also to prevent families from losing their homes and vehicles.
One UK bank faced a concerning rise in early arrears among unsecured loans, posing a threat of escalating default rates. To mitigate this risk, they sought proactive methods to predict which customers might face financial difficulties and determine the most suitable forbearance options for their credit agreements. These solutions aimed to improve both customer outcomes and business performance. However, existing AI models in the industry lack the sophistication required for this task. Traditional credit scores from bureaus like Experian offer limited predictive capabilities, and developing bespoke AI solutions in-house is prohibitively expensive and challenging due to a scarcity of AI talent. Moreover, current solutions fail to incorporate real-time multimodal data, such as voice tonality, or historical data reflecting periods of financial crisis and inflation, thus hindering their effectiveness in dynamically predicting customer financial distress and recommending optimal interventions like product restructuring or payment holidays.
Zinia AI addressed these challenges by offering hyper-personalized optimization models through its platform, facilitating dynamic decision-making for omnichannel customer experiences.
For this specific use case, the bank wanted to predict customers most likely to default on their mortgages and optimize intervention strategies, such as payment holidays and product restructuring, to both prevent defaults and maintain customers’ financial well-being post-loan repayments.
The bank had two objectives: (1) minimizing defaults and (2) ensuring customers maintain a healthy financial position. All the historical data, including the real-time conversation data with customers and information on available intervention actions was uploaded on Zinia platform. The AI platform automatically gathers additional external data, including economic indicators like unemployment rates and inflation.
After performing data cleansing, transformations, and causal analysis, the platform builds predictive models like default probability and repayment behavior. It then combines and selects these models using specific metrics accounting for precision, accuracy, and speed to optimize intervention parameters that achieve both objectives simultaneously.
The outputs include a customer-by-customer stream of parameters, such as recommended intervention strategies with suggested scripts and tonalities for customer support, aiming to minimize defaults while maintaining financial health. This process involves multi-objective optimization, ensuring optimal customer and business outcomes.
Additionally, the platform provides statistical guarantees on the improved performance of the optimized decisions, based on confidence levels or data-based performance percentiles, ensuring transparency and reliability in decision-making.
The optimal forbearance decisions generated by Zinia AI improved the key business performance indicator (default rate in this example) by 7%, while also reducing operational costs by 20% through increased utilization of digital channels.
By tailoring decisions to individual circumstances and ensuring seamless integration across various channels, Zinia AI enhances the overall customer experience. Moreover, its automated implementation via APIs or file transfers minimizes the burden on IT resources for customers, empowering businesses to rapidly scale on their AI projects to optimise operations.
