When Iris Friedman, Director of Product for Centrix Dispute Management at Q2, was in her 20s, she stood in a grocery store parking lot, sitting on a curb, panicking over a $200 withdrawal she didn’t recognize. She called her bank immediately to dispute it.
“Disputes are not operational events, even though they often live with operations teams,” she said at CONNECT26, Q2’s annual customer conference. “Disputes are emotional events. They’re very vulnerable situations.”
That framing cuts to the heart of why dispute management matters so much and why getting it wrong is so costly. Research shows that when a dispute isn’t handled efficiently, 60 to 70% of account holders are likely to move to another financial institution. What’s striking is that the outcome doesn’t even have to go against them. Slow, opaque, frustrating process alone can be enough to erode the relationship.
That’s the challenge Q2’s dispute management team is squarely focused on solving. Here’s what intelligent dispute resolution looks like and where Q2 Centrix is taking it.
Data from Q2 Centrix Dispute Management—with more than 500 financial institution customers—confirms the trend: Disputes are not only growing in volume, they’re growing in complexity. The conclusion Emily McDougall, Product Manager for Centrix Dispute Management at Q2, draws from that data is direct: “If we don’t lean on technology more, you will have to throw more people at the problem. And that’s not scalable.”
The financial institutions that have been living with this problem longest know what she means.
“If I had to describe the landscape of fraud in one word, I would use relentless,” said Evelyn Morris of SouthState Bank.
“Fighting fraud still seems reactionary,” added Keith Welch of PlainsCapital Bank. “We’re always just a half step behind the fraudster.”
It’s not just that there are more disputes. The disputes themselves are harder to resolve. True fraud is getting smarter and more automated, while friendly fraud, or first-party abuse, is growing even faster. Stripe data shows a 6.2x increase in abusive free trials detected between November 2025 and February 2026. The 2025 Global Payments & Fraud Report puts the cost of managing first-party abuse at $35 per $100 in disputes.
Meanwhile, the tools most FIs rely on to process disputes weren’t built for this scale.
“Right now we have to use at least four or five systems, counting email, to get through the process,” said Michelle Lutz of First Bank. “That creates a lot of friction. If we can’t respond to the customer in a timely manner, it erodes their confidence in us.”
The new model: Let computers do what computers do best
The shift Q2 is building toward isn’t about replacing human judgment. It’s about protecting it.
“Let computers do what they do best—processing large amounts of data and repeatable, scalable logic—so that humans can do what they do best, which is judgment, empathy, and decision making,” McDougall said.
The foundation for this shift is Q2’s Fraud Intelligence Platform: a closed-loop system that connects Q2 applications, partners, and financial institutions into a shared defense network. Rather than treating each dispute as an isolated event, the platform coordinates signals across the entire fraud lifecycle—detecting threats, intercepting them in real time, resolving cases with better information, and learning from every outcome.
Applied to dispute management, that platform translates into four interconnected capabilities.
One of the most persistent problems in dispute management is that cases arrive stripped of context. A transaction shows up. A customer calls. No one knows yet whether it’s true fraud, friendly fraud, a duplicate charge, or simple confusion. The investigator starts from zero.
What Q2 is building changes that. Rather than looking at disputes in isolation, Centrix Dispute Management will draw on a broader set of signals: the account holder’s behavior across their entire relationship, device and network data, merchant risk profiles, and cross-institution fraud patterns from across Q2’s network.
During a demo, Friedman walked through what that could look like in practice, via a case in which a hypothetical account holder had a series of card transactions posting from Washington state. In this scenario, the system would have already flagged that the account holder’s digital banking login data placed him in Austin at the same time those charges hit. The system would then surface that signal automatically and also flag that the transaction amount exceeded the account holder’s typical daily spending, adding a second layer of evidence pointing toward potential fraud.
The connections run both ways. If a reported account takeover event is already in motion, the system can automatically connect it to incoming disputes without needing a human to connect those dots. And disputes themselves can surface account takeover activity that hasn’t been detected yet. Sometimes, Friedman noted, the first sign that an account has been compromised is a dispute.
During a customer visit, McDougall and Friedman sat in on a call center handling a straightforward duplicate charge dispute. The account holder had already waited 45 minutes to reach an agent. Once connected, the conversation went well, but then both parties waited in silence for 15 minutes while the dispute form was located, sent, signed, and returned.
“Those are the moments of ambiguity and uncertainty that hurt the customer’s ability to trust their financial institution,” McDougall said.
The fix isn’t faster humans. It’s a smarter intake process. Q2 is building an AI-powered dispute intake that predicts the dispute reason based on transaction data, validates that the right attachments are included, coaches agents on what questions still need to be answered, and flags inconsistencies in real time. The goal is to get everything the back office needs on the first pass to eliminate the back-and-forth that makes disputes drag.
The same capability creates an important layer of friction for friendly fraud. If a dispute narrative doesn’t add up, the system prompts for clarification rather than letting it slide through. The intake can be configured to flag inconsistencies and prompt agents to ask the right follow-up questions.
Once a dispute reaches the investigator, the current experience often means opening multiple systems, pulling disparate data, and making a judgment call without a complete picture. It’s slow, inconsistent, and stressful, especially for newer employees.
Q2’s answer is an AI-powered recommendation model that does the synthesis work for the investigator. Rather than starting from scratch, they arrive at a case that already shows a recommended decision, a fraud confidence score, and the full stack of evidence behind it: card processor data, account holder activity, merchant risk signals, dispute history, and cross-network patterns from Q2’s 500+ customer base.
Critically, it’s designed not to be a black box. “We want to show you our evidence so you can come to your own conclusions and build trust in that model,” Friedman said. Investigators see exactly why the system landed where it did and can override it when their judgment calls for it. Over time, that visibility serves two purposes: 1) It expands the type of signals that can be automated and 2) It helps converts a tool into a trusted partner.
Each time the Centrix Dispute Management System goes through this process, it’s also steadily an incredibly valuable true fraud database.
Every dispute process, regardless of how it’s decided, is a data point. Accepted disputes, denied disputes, cases that turned out to be confusion rather than fraud—all feed the model. What the account holder didn’t dispute tells a story as well.
Q2 is building the platform to capture all of those signals and use them to continuously improve, not just within a single institution, but across the entire network. The 500+ financial institutions using Centrix Dispute Management represent an enormous collective dataset. Patterns that emerge at one institution can sharpen detection at all of them.
“The institutions that choose to learn from all of these signals will actually eventually outpace fraud,” McDougall said. “It’s going to allow them to react and change as they’re going through.”
Q2’s roadmap for Centrix Dispute Management includes large investments in automation, data science, and integrations. Some integrations, like Plaid merchant enrichment, are already in place, while merchant collaborations like Visa Verifi, Fiserv EFT, and MasterCom are either in early adopter phase or are on the roadmap. Some of the innovations in the coming months include using language models to predict dispute reasons, and the fraud confidence score.
To learn more about Q2 Centrix Dispute Management and where it’s headed, visit the resources below or reach out to start a conversation.