Artificial intelligence could help banks reduce their cost-to-serve by 20-30% as lenders increasingly deploy the technology across underwriting, collections and customer servicing, Tirtha Chatterjee, partner, Boston Consulting Group (BCG), said in an interaction.
“Cost-to-serve, we have estimated, it can come down by 20-30%,” Chatterjee said. The estimate factors in the use of AI across areas such as underwriting, personal discussions with borrowers and collections, he said. The findings are part of BCG’s latest report – Balance: Thriving in the Age of AI – which examines how financial institutions can balance growth and innovation with safety and governance as they adopt AI.
While banks have made significant investments in AI, the industry is still at an early stage in measuring the actual returns from these investments. However, bank executives are increasingly confident that returns will become visible over the next year. “80% of the CXOs that we interviewed believe that in the near term, let’s say in one year or so, they would realise returns on their AI investments,” Chatterjee said.
According to the report, process efficiency is currently the biggest area where banks are seeing benefits from AI, with more than 70% of respondents reporting gains. Nearly one in two respondents are also seeing cost reductions in select areas.
Collections is one area where AI is already helping lower costs, with lenders using AI-enabled bots and other tools to reduce the workload on tele-calling and field collection teams. AI could also help banks expand lending to customers who were previously difficult to underwrite because of limited credit histories. By using alternative data and AI-based models, lenders could assess customer behaviour and identify creditworthy borrowers who may not have been served earlier.
Chatterjee said this could be particularly relevant for the MSME segment, where lack of data and high costs of serving customers have traditionally limited credit access. Platforms such as the Unified Lending Interface (ULI) can provide additional transaction and business data on MSMEs, while AI can be used to identify patterns in that data and improve underwriting, he said. “If you take it together, the cost-to-serve becomes lower, and hence this customer becomes viable to address,” Chatterjee said.
BCG estimates that an “agentic bank” could unlock a 100-basis-point uplift in return on assets (RoA), with the gains shared between institutions and customers. Such a model could also deliver nearly twice the assets per employee, while bringing the cost-to-income ratio down to 25-30%, significantly improving the economics of banking, according to the report.
While AI is expected to automate some repetitive roles in banking, Chatterjee said the Indian banking sector’s relatively low credit and insurance penetration means the technology could primarily act as a growth accelerator rather than simply replace jobs. “India’s story can be slightly different from the global story,” he said, adding that AI could help banks address the large unmet demand for credit and financial services while augmenting existing roles.