Are Chargebacks Still the Black Sheep of Payments, or Are They Finally Evolving?

ChargebackX Session Recap

Brian Davis, Host, The House of Fraud

Barbara Redaelli, Fraud & Dispute Strategy Manager, Lastminute.com

Dave Gestaut, Principal Technical Program Manager, Microsoft

Roenen Ben-Ami (moderator), Co-Founder and Chief Risk Officer, Justt

Key Takeaways

• Chargebacks are now a strategic lever for cost control and customer experience, if you treat them with nuance rather than a monolith labeled “fraud.”
• Post-COVID shifts (especially in travel) mean service-related disputes now dominate many merchants’ mixes; which means that preventing fraud alone won’t contain chargebacks.
• Modern programs blend deflection, prevention, and scaled representment with better data access/normalization and ML/AI to learn from issuer feedback at volume.
• Regulation and network rule changes (e.g., Visa programs and fee/timeline updates) reward speed, accuracy, and automation; slow processes now carry real cost.
• Cross-functional alignment (payments, CX, product, finance) is the unlock: trace chargebacks back to business decisions, pricing, billing, policies, not just fraud signals.

Introduction

The discussion, hosted live at ChargebackX, the first conference dedicated exclusively to the world of chargebacks, brought together four experts from across the payments ecosystem to share their experiences developing and leading chargeback programs across a variety of industries.

This session debated whether chargebacks are still the overlooked “black sheep” of payments, or if they’ve gained recognition as a critical part of the payments practice.

From “black sheep” to business lever

Ronen opened with an observation that many in the payments ecosystem can probably relate to: over the past decade, anti-fraud tooling leapt ahead, yet the way merchants handle chargebacks still feels stuck in the past. Manual, slow, and oddly invisible, chargebacks were treated as a cost of doing business. But now all of a sudden, that’s changing fast. 

Each panelist entered the space at a different time and from different industries, but they’ve converged on the same reality: chargebacks can no longer be ignored or viewed only through the lens of fraud. They’re a systemic problem that can point to issues throughout the company, from supply chain and fulfillment to customer service. And yes, fraud too.

The shift is cultural as much as technical. At large enterprises, teams now invite chargeback leaders into roadmap conversations because they’ve started to see the twin upside, lower cost and higher customer satisfaction, when disputes are handled proactively.

The #1 Misconception: “All Chargebacks Are Fraud”

Dave Gestaut, Principal Technical Program Manager at Microsoft, called out the biggest mindset trap: treating everything chargeback-related as fraud. That flattens the nuance that makes or breaks dispute outcomes. For one, “fraud” itself is complicated. It splits into third-party and first-party (malicious, opportunistic, or confusion-driven). Then there are merchant/service disputes (billing clarity, cancellations, delivery, policy friction), which are not always tied to fraud.

Here’s the real example he gave: A business line saw a spike and assumed fraud was the culprit. But the real root cause was a recent price increase and refund/cancel friction, a customer experience problem, not a fraud ring. The fix would have to do with improving communications and creating more intuitive customer flows, not tweaking the risk models.

Barbara Redaelli, Fraud & Dispute Strategy Manager at LastMinute.com, added another hard truth: “If you’re right, you’ll win” is not a strategy. While it’s idealistic to think each case will be reviewed thoroughly and completely, reviewers only have a couple minutes to scan your chargeback dispute packet. You need to explain your business, evidence, and entitlement clearly, as if they’ve never heard of you. There is no room for assumptions here.

Travel’s new reality: service disputes eclipse fraud

Pre-COVID, LastMinute.com saw a fraud-heavy mix of chargebacks. Today, however, roughly 60% are service-related, a reflection of how consumer expectations shifted during the pandemic. When travel restrictions and uncertainty made refunds commonplace, customers grew used to instant, flexible resolution. At the same time, overwhelmed support centers trained consumers to seek faster outcomes through their banks. Easier online dispute tools and stronger consumer protections reinforced that habit, while lingering mistrust from canceled trips and fine-print frustrations eroded confidence in merchant fairness. As a result, many customers now see filing a dispute not as fraud, but as standard customer service.

Travel, specifically, has two structural challenges:

1. Intangibility: You “deliver” a service, not a package. Evidence must prove service fulfillment (usage, boarding, or consumption), not just shipment. In many cases, the customer cannot preview the service beforehand, leading to potential gaps in expectations, like discovering upon arrival that your “small group” wine tour includes 25 people and a bus, not quite the intimate experience you imagined.
2. Long lead times: Chargeback clocks for service disputes often start at service date, not purchase. A February trip booked the prior May is still in the risk window long after authorization. That complicates accruals and operational planning.

Add multi-acquirer complexity and supplier dependencies (airline bankruptcies, refund policies), and you get a sector where data completeness and normalization are table stakes.

A much-needed strategy change: using data, automation, and AI

The House of Fraud Host Brian Davis’ inflection point came when his team realized that manually fighting chargebacks on low-price subscriptions cost more than it recovered. The labor and data prep simply didn’t pencil out. Early automation tools flipped the math. Suddenly, it became viable to contest high volumes of small-value disputes efficiently. Since then, the model has evolved from outsourced, manual rebuttals to automated, data-rich, template-aware representment, plus proactive alerts and deflection (e.g., near-real-time cancellation/refund paths that stop disputes before they’re filed).

Dave Gestaut from Microsoft highlighted the step-function: with hundreds of thousands of cases, LLM-powered analysis can mine issuer loss reasons and longform notes at scale, something infeasible for humans. The payoff is tighter templates and faster learning loops without replacing people; it supercharges them and in many cases produces better representments and better outcomes.

Data is the moat, and the bottleneck

As chargeback operations evolve from manual workflows to automated, AI-assisted systems, data becomes both the competitive advantage and the constraint. Barbara Redaelli from LastMinute.com highlighted two of the biggest blockers standing in the way:

• Accessibility: Acquirers deliver TC40s, chargeback data, and reason codes across SFTP, APIs, emails, mailboxes, or not at all. It’s difficult to access in one place and make sense of.
• Normalization: Every acquirer speaks a different dialect (fields, formats, reason-code mapping). Without unifying those streams, you’ll never see a trustworthy picture of liability, win rate, fees, or root causes.

Her solution is “translation: invest early in a data pipeline that standardizes inputs and enriches evidence. Everything else, deflection, prevention, rebuttal – depends on it.

Prevention vs. representment: run it as a flywheel

Too often, teams treat prevention and representment as separate tracks. In reality, the strongest programs run them as a flywheel, where each motion strengthens the next. Here’s how that looks:

1. Deflect: Eliminate avoidable disputes by solving billing confusion upfront, clear descriptors, instant receipts, click-to-cancel flows, and self-serve refunds where appropriate.
2. Prevent: Tune fraud controls and business policies to reduce friction without inviting abuse. Monitor leading indicators, as Brian Davis from The House of Fraud noted, even macro shifts like sudden currency devaluation can predict regional chargeback spikes.
3. Represent at scale: Use automation and machine learning to build consistent, data-backed cases, using a tool such as Justt. Role-specific templates and narrative support improve accuracy without slowing volume.
4. Learn: Feed issuer feedback and loss reasons back into CX, product, policy, and risk. Every dispute teaches you something about expectation gaps or friction points.

When this loop runs smoothly, you don’t just fight chargebacks better, you learn from each to prevent the next.

Rules and fees: timelines are tightening (and time is money)

Visa and Mastercard’s latest updates, stricter program thresholds, higher fees, and shorter response windows, are raising the bar for dispute management. The message is clear: move faster or pay for it. Manual workflows that once passed as “good enough” now cost real money in missed deadlines and compliance fees. Merchants that automate documentation, tracking, and evidence submission are saving time and protecting margin.

From cost of doing business to cross-functional discipline

Brian Davis from The House of Fraud contrasted the past’s “write it off” culture with today’s best-in-class programs that map chargebacks to specific product and policy drivers (free trials, returns, billing changes, global pricing, translation/localization). That specificity unlocks targeted fixes and credibility with finance and product.

Dave Gestaut’s barometer at Microsoft: teams now proactively invite chargeback leaders to discuss trends and roadmaps. That’s the signal you’re no longer the black sheep. You’re now part of core business hygiene.

A practical blueprint to take with you

• Own the taxonomy: Segment by third-party fraud, first-party fraud (malicious/opportunistic/confusion), and merchant/service issues.
• Instrument CX: Tighten descriptors, receipts, cancellation and refund flows. Monitor friction and post-purchase tickets.
• Normalize the data: Centralize acquirer feeds; standardize fields; reconcile fees and outcomes.
• Automate the middle: Alerts/deflection; template-driven representment; ML-assisted packet creation; SLA dashboards for clocks.
• Close the loop: Analyze issuer loss reasons (structured + unstructured), push learnings to product, pricing, and policy owners.
• Forecast the weird stuff: Macros (currency moves), policy shifts (pricing), and seasonality, so finance can accrue and ops can staff.

Ready to Reduce Chargebacks?

Most teams still treat chargebacks like the “black sheep” of payments, something to push into the fraud bucket and forget. But that no longer works. To truly reduce losses, you need to connect the dots across CX, payments, and risk, powered by a unified data pipeline and an always-on flywheel: deflect, prevent, represent, learn.

Justt brings the automation, analytics, and expertise to make that shift real.

See Justt in action →

FAQs

How can a cardholder file a chargeback months after purchase?

For service disputes, network timelines typically start at the service date, not the purchase date. In travel, a ticket bought in May for a February trip remains within the dispute window long after authorization. Plan accruals and evidence retention accordingly.

What’s the difference between third-party fraud and first-party fraud?

• Third-party fraud: an unauthorized party used the card.
• First-party fraud: the cardholder (or household) authorized the transaction but disputes it, maliciously, opportunistically (e.g., “I forgot my other account”), or due to confusion (billing descriptor, subscription renewals).

We’re “right”, why did we still lose?

Issuers review quickly. If your packet doesn’t teach a stranger what you sell, why the customer is liable, and how you delivered the service, supported by clear, labeled evidence, you can lose a technically correct case. Narrative and formatting matter.

What are the fastest ways to deflect avoidable disputes?

Clarify statement descriptors, send real receipts with item/service details, enable click-to-cancel, expose self-serve refunds where policy allows, and surface renewal notices and price changes clearly.

Which KPIs should we track beyond “win rate”?

• Dispute rate by type (fraud vs. service/merchant)
• Deflection rate (alerts resolved without chargeback)
• Timeliness (response-in-time %, cycle times)
• Net recovery (wins minus fees/ops cost)
• Root-cause mix (policy, pricing, CX, payments auth)
• Issuer feedback themes (structured + unstructured)

Our dispute mix shifted to service issues. What should we change first?

Audit policies and communications: cancellations, refunds, service delivery proof, SLAs, and customer messaging. Then tune representment templates to emphasize entitlement and fulfillment, not just identity.

How does AI help in representment?

LLMs can summarize issuer loss reasons at scale, suggest template improvements, and help assemble case narratives from raw logs. Pair that with human review, and you accelerate learning without sacrificing quality.

Multi-acquirer setup is messy. Any tips?

Centralize feeds, normalize fields, and standardize reason codes. Where acquirers lack APIs, schedule reliable SFTP/email ingestion. Build a single “source of truth” that powers deflection, case assembly, and reporting.

How do upcoming network fee/timeline changes affect us?

Shorter clocks + higher fees for misses mean you need automation, SLA monitoring, and clean evidence pipelines. Slow responses now show up directly as cost.

We’re early – what’s a pragmatic 90-day plan?

Stand up data intake/normalization, fix descriptors and receipts, deploy alerts/deflection, templatize top 5 reason codes, and build a weekly cross-functional review (payments, CX, support, finance) to drive root-cause fixes.