September 25, 2026 · Disputed Team
Most chargeback teams measure their success with one number: "What percentage of disputes did we win?"
The problem with this approach is that it hides everything useful.
Imagine you fight six types of chargebacks. Two types you're great at, you win almost all of them. Four types you almost never win. Blend those together and you get a win rate that looks… fine. Maybe 50%.
But that number doesn't tell you which fights are worth fighting and which ones are a waste of time and money for you.
Now, what if your CFO asks you, "Why are we losing 10% of these cases?" and you can't answer them because that blended number hides the information you actually need?
What you need to do is stop looking at one scoreboard number and start tracking the metrics that tell you something actionable, so you know which battles to fight harder, which to fight differently, and which to stop fighting altogether.
Those eight are your Visa VAMP ratio, win rate by reason code, chargeback accept rate, cost per represented dispute, time to respond, recovery rate, repeat-disputer rate, and root cause distribution.
Averages always hide the more interesting stories that your data tells you over time.
So, let's walk through the eight metrics a dispute team should actually watch, why they're important, and what good and bad look like.
What it is: The Visa Acquirer Monitoring Program (VAMP) counts fraud reports and disputes together against settled transactions, on card-not-present volume only.
VAMP ratio = Count of [Fraud (TC40) + Disputes (TC15)] ÷ Count of Settled Transactions (TC05)
What bad looks like: A ratio trending toward the applicable VAMP threshold without a clear remediation plan or downward trend. A single spike from a fraud ring, a product recall, or a billing error is enough to move you into that territory.
What good looks like: A ratio with a meaningful buffer before the applicable threshold, and room to absorb a bad month without triggering monitoring. If you go over, Visa may subject you to VAMP identification and applicable assessments. The specifics depend on your program and volume. If you sell across regions, you can break this out per region.
Two things drop out of the numerator: cases resolved through pre-dispute solutions, and TC40 fraud qualifying for Compelling Evidence 3.0. Neither is guaranteed in a given month. Visa conditions both on timing.
What it is: The percentage of disputes you win, broken out by reason code rather than blended into one number.
Why it matters: A blended win rate can hide the fact that you're winning friendly fraud cases at several times the rate you win true fraud. The single number makes the team look average when it's strong in one category and wasting effort in another.
What good looks like: A table your team reviews monthly. You know which reason codes you're good at, and which ones aren't worth the fight.
What bad looks like: One number on a dashboard that falls apart the moment leadership asks which codes it came from.
What it is: The percentage of everything you made an active decision on that you chose not to fight at all.
Accept rate = Accepted ÷ (Won + Lost + Accepted)
Why it matters: Accepting isn't inherently bad because some cases genuinely aren't worth fighting. But the number is a black box without a reason attached: no evidence, too small, a known-unwinnable reason code, a missed deadline, a client instruction. A rising accept rate means something different depending on which reason is driving it, a staffing problem looks nothing like a data problem, even though both show up as the same headline number.
One wrinkle you want to be aware of: group it by the week the decision was made, or the week the case first came in. Ops wants the first because it's when the call happened. Finance wants the second because it's when the case entered the pipeline. Pick one and stay consistent, or the trend line moves for reasons that have nothing to do with your team.
What good looks like: Every accepted case tagged with a reason, and a monthly read of which reason is growing.
What bad looks like: An accept rate that moves and nobody can say why.
What it is: Total dispute-team cost (labor, vendor fees, network charges) divided by disputes responded to.
Why it matters: This tells you whether fighting is worth it at your volume. If it costs you $45 to handle a $30 dispute, it probably doesn't make financial sense, no matter how often you win.
What good looks like: A number you can put next to your average dispute value.
What bad looks like: Spending more to fight a dispute than the dispute is worth, and not seeing it because nobody tracks the cost side.
What it is: Average days from dispute notification to evidence submission.
Why it matters: When the average creeps toward the deadline, cases get rushed or missed. Some people say earlier responses win more often, which is worth testing against your own data before you accept it.
What good looks like: Consistent and well inside the deadline, with enough room to build real evidence.
What bad looks like: A number that keeps climbing while dispute volume climbs with it.
What it is: Same idea as win rate by reason code, but weighted by dollars instead of case count. A $5,000 win should count for more than a $50 win, and case counts don't do that.
Recovery rate = $ Won ÷ ($ Won + $ Lost + $ Pre-Arbitration + $ Accepted + $ Not Fought)
Why it matters: Case counts can hide a real problem — winning lots of small, easy cases while losing the big ones. If recovery rate sits well below your win rate by reason code, that's the tell: you're winning volume, not value, and it's worth finding where the big losses concentrate.
Subtract program costs like fees and tooling from the numerator and you get the honest version: did this program make money, or just look busy?
What good looks like: A number that tracks close to your case-count win rate. A meaningful gap tells you exactly where to look.
What bad looks like: A healthy win rate on paper sitting next to a recovery rate that shows you're losing the disputes that actually matter.
What it is: The percentage of disputes from customers who have disputed before.
Why it matters: When customers repeatedly dispute charges, it can point to either first-party misuse or ongoing issues with their experience. Tracking these cases surfaces patterns you'd miss looking at each dispute on its own. If this number is climbing, the problem sits upstream of the dispute team.
What good looks like: Stable or declining, with the worst repeat offenders flagged before the next dispute arrives.
What bad looks like: You don't know, because your system doesn't connect disputes to customer history.
What it is: A breakdown of why disputes happen. True fraud, friendly fraud, merchant error, processing error, subscription confusion.
Why it matters: This is the metric that names what to prevent. If a lot of your disputes come from subscription confusion, take a closer look at how cancellations work, how billing is described, what renewal messages say, and how you communicate with customers. If a lot come from first-party misuse, review your abuse controls and check customer history signals more carefully.
What good looks like: A chart that puts prevention on the agenda, with a named owner for the largest cause.
What bad looks like: Every dispute treated the same way regardless of why it happened.
Your VAMP ratio tells you whether you have a compliance problem. Win rate by reason code, chargeback accept rate, cost per represented dispute, time to respond, and recovery rate tell you how the team is performing right now. Repeat-disputer rate and root cause distribution tell you where the problem starts, before it reaches the team at all.
There's one thing worth watching that isn't its own metric: accept rate and win rate together as a paired trend.
Accepts rising while win rate holds steady or improves means you're correctly filtering weak cases, that's healthy triage. Accepts rising while win rate drops means something upstream is degrading. It's the same headline number with two very different situations.
If your metrics live in a spreadsheet you maintain by hand, that's fine. Knowing which numbers to track comes first.
Source: Visa Acquirer Monitoring Program fact sheet. Thresholds effective 1 June 2025; Excessive Merchant threshold for AP, Canada, EU and US reduced to 150bps on 1 April 2026.