Banks urged to improve fraud dispute processes to restore trust
By Pieter de Swardt, Entersekt, SVP Global Customer Success
Fraud resolution remains slow, manual and often opaque, even as digital banking fraud accelerates.
Despite security efforts, South Africa digital banking fraud incidents grew by 86% in 2024 compared to 2023, accounting for roughly R1.9 billion in losses, with banking apps accounting for 65% of reported incidents. Most major banks now have digital fraud dispute reporting channels. Yet the actual fraud dispute process remains cumbersome and largely analogue, often taking weeks to conclude.
Concomitantly, the National Financial Ombud Scheme’s banking division saw record complaint volumes in 2024 (over 15,000 complaints), with nearly 80% of the cases decided in the banks’ favour. It’s not surprising that there is a growing perception among local consumers that the system is stacked against them.
How banks respond is increasingly vital, with research showing that 73% of customers say their loyalty is heavily influenced by how fraud is handled.
Using the same signals twice
The problem is not a lack of data, but how it is used.
When banks authorise a high‑risk transaction, they already have a wealth of contextual data, including device fingerprints, geolocation, behavioural biometrics, step‑up authentication outcomes, and risk scores generated in real time by their fraud engines.
However, most of that intelligence is currently used once to decide whether to allow, challenge, or deny a payment. This data is then effectively lost to the dispute teams who are forced to reconstruct what happened after the fact.
By treating disputes as an extension of risk‑based authentication, not a separate process, institutions stand to significantly assist with their dispute challenges.
Firstly, they can surface more transaction context to customers in‑app and online. Where merchant names are unclear, banks can display additional data such as merchant category, location, and even past‑spend patterns to help customers recognise legitimate payments and avoid unnecessary disputes altogether.
Secondly, the same evidence used for step‑up authentication (device data, behavioural biometrics, geolocation and anomaly scores), can be made instantly available to dispute teams, cutting investigation times and reducing “we have no proof” conversations.
Finally, by reframing disputes as part of risk based authentication, not just an admin queue, disputes become a continuation of the same risk‑based logic, using the same signals, models and audit trails, but with a lot more transparency for customers.
This is not a conceptual leap for South African institutions already using advanced, context‑based authentication platforms to combat phishing, SIM‑swap fraud, and authorised push payment scams. It is primarily an integration and product‑design challenge. Given the regulatory and reputational pressure created by rising digital fraud and public cases where some victims have fought for months to be refunded, it makes excellent sense for banks to take this opportunity to address this challenge.
Informed customers are loyal customers
U.S. banks are acutely familiar with the costs associated with disputes. In 2024, consumers disputed up to 105 million charges with U.S. card companies, worth an estimated $11 billion.
Eager to better understand the business costs associated with how disputes are handled, researchers from Notre Dame and Carnegie Mellon universities examined customer data from a major U.S. bank over five years.
Rather than walking away simply because they had experienced the fraud, the study found that it was the lack of transparency from their bank that prompted the customers’ decision to leave. What’s more, when a bank had identified the fraudster, 62% fewer victims left compared to customers who had never experienced fraud.
Analytics helps keep it transparent
Banks could let existing analytics do the heavy lifting to help deliver provisional outcomes. This injects speed and fairness during a time of uncertainty and adds transparency to the process. However, this comes with risks.
Presenting more context about transactions in digital channels helps avoid unnecessary disputes, but over‑exposing internal signal data could give sophisticated fraudsters clues about how to evade controls.
Looking ahead, the data and decisioning engines banks already operate for fraud prevention give them a ready‑made foundation for faster, more personalised and more transparent dispute journeys.
Partners that already underpin digital authentication and fraud prevention at South African banks are well‑placed to help close the gap between prevention and resolution. Experience shows us that smarter use of risk and contextual signals promises lower handling costs, better ombud outcomes and, perhaps most critically in a high‑fraud environment, a way to rebuild trust in digital banking.
