How Gen AI and Automation Can Help Banks Absorb Chargeback Surge

Chargeback requests have been a feature of retail banking for years but in the last 12 months there has been a tidal surge of customer requests. 

Research firm, Datos Insights, discovered customers disputed $11 billion worth of charges in the U.S. in 2023 alone, with this forecast to rise 40% by 2026. Other countries are seeing similar sharp increases so that in a market like the UK it has been estimated that one in 200 payments results in a chargeback. 

This surge is coinciding with how customers have adapted a wide array of alternative payment methods including mobile wallet payments, peer-to-peer payments, buy now pay later (BNPL), and real/near-real time payment systems.

So, in processing customers’ chargeback requests, banks must engage not only with payment giants Visa and Mastercard but also work with these other systems and payment types, many of which are regional or national. 

There’s also the challenge of navigating a complicated web of changing chargeback policies, spread over siloed systems.

Altogether these factors are making the cost to manage chargeback requests increase in terms of time and resources needed to be allocated. 

Automation is clearly important and can absorb these growing pressures and difficulties in processing complex cases at scale. The prerequisite is automation that can be adapted to how processes and integrations with third party payment systems evolve. The challenge in digitising new workflows has been that many man hours of manual work have been needed to map what needs to be coded. Developments in AI-powered workflow design platforms have now enabled the ability to generate design blueprints that can be easily converted into code. Indeed, these kind of tools are going to be key in how new payment types can be supported in just hours and days instead of weeks and months, using a combination of automation with some guided intelligence. 

In their response, banks recognise that their customers are often looking to engage with humans and not systems only. Customers are often in distress or confused when they request a chargeback payment and expect access to help that’s empathetic and effective. The importance of  how well customer service is integrated into the process is highlighted by a study from PYMTS Intelligence. 

The study found that one in two satisfied customers cited fast processing as a key factor in their satisfaction and the importance of accessible customer service (i.e. could they get to talk to someone). The voice channel is far from dead with a panacea of digital only self-service, so the challenge is to set up the best of both worlds – digital first and automated within certain threshold conditions, and the ability for a customer at any time to pause and resume or switch channels to talk to someone.

So, any proliferation of automation must keep the human in the loop. Ironically, the need to amplify the human component in these processes is creating a greater role for generative AI to help guide bank and customer interactions through the next best actions and outcomes for everyone. A genuine case for automation and intelligence augmenting and helping both customers and employees.

There are two prime areas where advanced technology can be effective in customer experience around chargebacks. 

One is training generative AI applications with an encyclopaedia of knowledge about the processes and rules involved and thus provide intelligent guidance through the maze of chargeback claims and be an always-on mentor for a human agent to help optimise their work and overcome roadblocks. 

Secondly, of equal importance is understanding the structured approach adopted by Smart Dispute for data retrieval and processing. Unlike systems that deploy generative AI for sifting through policy and procedure documents, Smart Dispute integrates network rules directly into an orchestrated process. This ensures that bank employees have quick and accurate answers to dispute processing question, enhancing customer satisfaction. What’s key here is the controlled procedure that targets the right documentation, maintaining security and reducing the chance of any inadvertent data misuse or errors.

Building on these core capabilities there are an array of other ways advanced technology can overhaul and improve the chargeback claims process for agents and customers. For example, gen AI agents can help employees process disputes smarter and faster, including instant summaries of customer calls or instant summaries of client claims and dispute histories. Summaries and recommendations can trigger straight to automated processes, for example straight through processing a low value card dispute.

As they are confronted with a rapid growth in chargeback disputes, banks know they must be fast and adaptable to keep up with customer demands. As it matures and becomes more relevant to the sector, both process and statistical AI as well as generative AI can be combined with automation capabilities to increase efficiency and bring faster resolution to customers.