On April 1, 2026, Visa lowered the merchant threshold in its Acquirer Monitoring Program from 2.20% to 1.50%, and it now folds fraud and disputes into one number. For a payments product manager, that single rule change is a useful lens on a whole interview topic. Disputes and chargebacks sit at the center of risk, compliance, and merchant experience, and they come up in payments PM loops more than most candidates expect. This guide covers what the process is, the metrics that govern it, and the questions an interviewer is likely to ask.
What a chargeback actually is
A chargeback is a forced reversal of a card payment, started by the cardholder's bank rather than the merchant. The flow has a fixed shape. A cardholder sees a charge they want to contest, the issuing bank files a dispute through the card network, the network passes it to the acquirer, and the acquirer debits the merchant. The merchant can accept the loss or fight it with evidence, a step called representment. If the two banks still disagree, the network can move the case to arbitration and rule on who pays. Each dispute carries a reason code, such as fraud, goods not received, or a canceled subscription, and the reason code sets the evidence a merchant needs to win the case.
The metrics that govern it
Card networks watch ratios, and a payments PM lives inside those ratios. Visa's program now uses a single count-based number: fraud reports (TC40) plus disputes (TC15), divided by settled transactions (TC05), measured on card-not-present volume. Cross the excessive line and Visa charges $8 for every fraud and dispute item, on top of the money already lost to the reversal. Mastercard runs its own monitoring programs with similar intent. The job is to keep two rates healthy at once: the fraud rate and the dispute rate. Push either one too high and the result is higher fees and added network scrutiny.
Knowing how the ratio is built matters in an interview, because it tells you which lever moves the number. The denominator is total settled transactions, so a merchant with rising good volume gets some headroom. The numerator combines fraud and disputes, so a product that cuts fraud reports also helps the dispute side of the same ratio. That shared denominator is the kind of detail that separates a memorized answer from a worked analysis.
First-party fraud is a large share
Not every dispute comes from a stolen card. Mastercard and Javelin research put first-party fraud, often called friendly fraud, at about 20% of disputes in 2025. Some of it is honest confusion: a billing descriptor the cardholder did not recognize, a renewal they forgot, a family member who used the card. Some of it is a customer who received the goods and filed anyway. The product response differs by cause, so a PM has to separate the cases before designing a fix. Mastercard's 2025 State of Chargebacks report expects the global cost of chargebacks to merchants to reach $42 billion by 2028, with close to half reported as fraud.
What a payments PM owns here
The work splits into prevention, detection, and recovery. Prevention reduces the reasons a dispute starts: a clear billing descriptor, an emailed receipt, address and CVV checks at authorization, and 3-D Secure on the riskiest transactions, which can shift fraud liability to the issuer. Detection scores transactions and flags the risky ones before they settle, the space Stripe Radar and similar tools occupy. Recovery is the representment pipeline: gathering evidence, matching it to the reason code, and submitting it inside the network's deadline. A payments PM usually owns the data model and the automation across all three areas, because manual review does not scale with volume.
These three areas trade against each other in budget and attention. A team that pours everything into detection can still lose revenue if its representment pipeline is slow. A team that automates recovery but ignores billing descriptors keeps paying for preventable disputes. The interview-ready view treats the three as one system with a shared metric.
How interviewers probe this
Expect open design questions. "Design a chargeback management product for small merchants." "A merchant's dispute rate just crossed Visa's threshold; walk me through how you bring it down." "How would you reduce friendly fraud without hurting conversion." A complete answer names the metric first, then walks the lifecycle, then proposes specific levers tied to causes. It also states the tradeoff in plain terms. Stricter fraud rules cut losses but also decline good customers, which lowers the authorization rate and revenue. More friction at checkout, such as a 3-D Secure step, reduces fraud but can cost you conversions. Interviewers want to hear you size that tension with numbers rather than talk around the problem.
The tradeoff to anchor on
The center of this topic is a balance between approving good payments and blocking bad ones. Every control carries a false-positive cost. A model tuned to catch more fraud will also block more legitimate buyers, and each blocked buyer is lost revenue plus a support ticket. So the right metric set pairs the fraud rate and dispute rate with the authorization rate and conversion. A PM who reports only one side of that ledger will miss the cost of their own rules. Evidence automation earns its keep on the recovery side, because a won representment recovers revenue that prevention alone would have booked as a loss.
What to study before the loop
Learn the dispute lifecycle end to end, including common reason codes and representment deadlines. Know the network monitoring programs by name and how their ratios are calculated, since the April 2026 VAMP change is a current, concrete example you can cite. Understand the liability shift that 3-D Secure creates, and where it helps or costs you. Be ready to talk about first-party fraud as a measurement problem rather than only a fraud problem. And practice sizing the prevention-versus-approval tradeoff out loud, because that reasoning is the real test.
Disputes are not a corner case in payments. They are a daily input to risk, revenue, and merchant trust. Treat the topic as a system you can measure, and the questions in the loop get much easier to answer.