First-party fraud is putting tremendous pressure on card issuers, with 160 million disputes filed last year. Some studies suggest that as many as 50% of disputes may involve first-party fraud. At this scale, manual reviews are no longer sufficient. Automation is helping issuers improve dispute outcomes while turning the resulting data into actionable insights.
In a PaymentsJournal Podcast, Pallavi Kuppa-Apte, CEO and Co-Founder of Decisionly, an AI-powered dispute automation platform for issuers, and Suzanne Sando, Lead Analyst of Fraud Management at Javelin Strategy & Research, discussed why first-party fraud surged in the post-COVID era and how advanced tools are helping the industry combat it.
Kuppa-Apte and her co-founders previously built Chargehound, the first automated dispute solution for merchants, which automated billions in disputes for some of the world’s largest merchants before being acquired by PayPal. Decisionly brings that technology-first approach to issuers, automating 90% or more of disputes end to end with zero human touch.
Reasons Behind the Rise
The rise in first-party fraud stems in part from consumers’ growing familiarity with the dispute process. Many have learned to use it to seek refunds when they feel they’ve been defrauded or scammed. At the same time, issuers have invested in better digital experiences, making it easier to dispute a transaction with just a few clicks in a banking portal.
The pandemic also helped normalize certain behaviors. Some retailers stopped requiring customers to return merchandise, either out of concern about virus transmission or because they couldn’t manage the surge in online orders and returns. Over time, some consumers came to view these losses as a cost of doing business, rather than a reason to follow the usual return process.
Together, these factors have created an easy path for abuse at scale. Roughly 40% of people admit to committing first-party fraud, and among cardholders who successfully win a dispute, 50% may try it again.
“First-party fraud has become perceived as a victimless crime,” said Sando. “Many consumers think larger retailers are not going to feel the loss from a fraud claim or disputed charge. Others don’t understand that they’re committing fraud by filing a chargeback.”
Kuppa-Apte saw the same shift firsthand. “We were on the merchant side during COVID building Chargehound and saw this massive acceleration in first-party fraud,” she said. “One of our big takeaways now at Decisionly is that issuers often don’t have the tools in place to handle a first-party fraud landscape that is so different from even just a few years ago.”
The Limits of Manual Processes
Many issuers have struggled to keep pace with the influx of disputes, sometimes inadvertently rewarding fraudulent behavior. This creates a vicious cycle.
“When dispute handling is manual and issuers see a holiday spike, release a new product or enter a new market, that’s when these manual processes break down,” said Kuppa-Apte. “You’re coming up against deadlines, you turn to write-offs, and you’re not evaluating these claims and these cardholders. Those cardholders are going to come back and increase your first-party fraud even more.”
Cardholders learn that disputing a transaction is an easy way to get a refund, encouraging repeat abuse and driving dispute volumes even higher. Most dispute operations were built for a smaller, more predictable caseload and aren’t equipped to handle the current scale.
“When issuers have a manual process, everything is happening asynchronously, so the person on the phone has to intake the dispute, write it all down and then hand it off to someone to investigate it,” said Kuppa-Apte. “It’s all happening sequentially, and you don’t have the information that you need upfront.”
Regulatory and network deadlines don’t move when volume spikes, and hiring can’t keep pace.
“You can’t hire and train an analyst fast enough, and every additional headcount is additional fixed cost,” Kuppa-Apte said. “What if you had a massive fraud spike and it’s going to go back down the next month? What do you do with that headcount? People handling disputes inherently does not scale.”
Issuers also feel pressure to deflect disputes at intake, because processing them manually downstream is so expensive.
“It’s going to cost them 50 bucks to process it manually after the fact, and that’s a tremendously high opportunity cost at scale,” said Kuppa-Apte. “Decisionly can process 250,000 disputes a minute. Now you can give your cardholders a great experience on the front end, because you know you can instantly run a consistent, accurate investigation and decision on the back end for pennies. That flips the equation.”
Determining Intent
Traditional fraud detection focuses largely on whether a transaction appears legitimate. If a customer passes identity verification or uses a familiar device, the transaction may not raise any red flags. But those signals don’t necessarily reveal the customer’s intent. The challenge is determining whether a dispute reflects a legitimate grievance or an attempt to exploit the process.
“That’s very difficult to determine,” said Sando. “A customer can genuinely fail to recognize a transaction. Maybe they’re using a card that’s for the entire family. They’re an authorized user on an account, but they don’t recognize that that transaction came from somebody else. Maybe the information that’s fed back to their online account statement from the merchant doesn’t match what they expect to see.”
Kuppa-Apte said the answer starts at intake. “One of the most difficult things about first-party fraud is that you often don’t know it’s first-party fraud until the result of that dispute,” she said. “If you’re using Decisionly, for example, we can look at all sorts of signals around the cardholder and the transaction to flag immediately if this could be a situation where you want to ask more questions.”
Getting that decision wrong has significant consequences. Failing to identify first-party fraud can lead to repeated abuse and mounting losses. But denying real claims can erode customer trust, raise regulatory concerns, and damage an issuer’s reputation. Frustrated cardholders may even close their accounts and take their business elsewhere.
Poor decision-making can also compound the problem over time. Inconsistent or inaccurate outcomes may feed low-quality data back into fraud models, making it harder for organizations to distinguish legitimate disputes from first-party fraud in the future.
“Oftentimes, because you’re going through this process so quickly, you don’t have the context that you need to make the best decision for this particular dispute case,” Sando said. “You just don’t have time. An organization is forced to deprioritize the lower dollar claims because it just isn’t worth the monetary investment of investigation.”
From the Merchant’s Perspective
From the merchant’s perspective, first-party fraud has long been a costly challenge. Disputes that merchants could successfully challenge often went uncontested because the time and expense of manual processing made it impractical to pursue every case.
Automation has changed that equation. Rather than picking and choosing which disputes to fight, merchants can now process virtually all of them at scale, improving their ability to recover revenue that might otherwise be lost.
“They also started getting access to all of these data signals around cardholder behavior, around first party fraud and real fraud,” said Kuppa-Apte. “They started to understand how many of these disputes are first party fraud versus a real fraud problem. They were able to tune their front-end transaction monitoring and their back-end returns and cardholder experiences. That data was super valuable. Those are things that we’ve taken to the issuing side with Decisionly.”
The Promise of Automation
Manual dispute handling simply does not scale to meet today’s demands. Automation, especially when powered by AI, can support a comprehensive investigation from intake through resolution, potentially with zero human intervention.
“Every issuer, every card product, every card program is different,” said Kuppa-Apte. “Decisionly enables them to investigate and action 100% of their disputes, regardless of volume, instantly, every single time.”
That shifts where people spend their time.
“You get to make a choice about the 5% of cases you want a human to look at,” said Kuppa-Apte. “Not because I’m forced to have humans look at this, but because I think that humans are going to add value to this segment of disputes, whether that is first party fraud, repeat abusers, high-value card holders or high-value transactions. You get to make a choice around your exceptions.”
The benefits go beyond processing efficiency. Automated reviews generate structured data that issuers can use to improve both fraud models and dispute workflows. Rather than treating each dispute as an isolated case, organizations can use the data to identify patterns and make more informed decisions.
These patterns can be difficult for humans to identify across large volumes of claims. AI can analyze more disputes consistently, learn from outcomes, and help refine future decisions. Over time, that feedback loop can bolster detection capabilities and give issuers a clearer picture of downstream risk.
“You can’t just isolate events and evaluate them as such,” said Sando. “You have to combine the important signals like customer history, prior dispute history, their account relationships—all of these things that make a full picture of what’s really happening.”
“That’s what helps you build the intent behind what a customer transaction really is,” she said. “That’s where you can make a stronger decision on whether or not this is a legitimate dispute.”
“Any one dispute that falls through the cracks creates regulatory risk, a missed cardholder experience and more opportunities for fraud,” said Kuppa-Apte. “That’s how you get coverage: by investing in real automation technology that can automate over 90% of your total dispute volume end to end. Now you’re really doing high-value human reviews.”