There is a specific kind of dread that arrives on a Tuesday morning. The money is there. You can see it in the dashboard — sixteen thousand dollars, earned, settled, real. And next to it, a sentence written in the flattest possible language: A portion of your balance is being held as a reserve. No phone number. No name. Just your own money, standing behind glass, while payroll runs on Friday.

Most merchants read that sentence as an accusation. They assume Stripe has decided they're a fraudster, and they respond the way accused people do: they explain themselves. They write long emails about their product quality and their happy customers. It almost never works, because the reserve was never a verdict on their character. It was arithmetic. And the thing that makes the arithmetic so cruel is that it doesn't measure how many customers disputed you. It measures how many disputed you relative to something else — and that something else is the number you control least when business is bad.

A reserve is a risk instrument, not a punishment

When you accept a card payment, Stripe fronts you money it has not finished being responsible for. A cardholder can dispute a charge months after it clears. If that dispute succeeds and your account is empty, someone still has to make the cardholder whole — and that someone, contractually, is the payment processor. Stripe is not holding your funds because it thinks you're a criminal. It's holding them because it has quietly become your creditor, and its exposure just changed.

That exposure is estimated, not observed. Stripe cannot know which of your future customers will dispute. It uses the only proxy anyone has: your recent dispute behavior, expressed as a rate. A rolling reserve — typically a percentage of each day's volume, released after a fixed window like 60 or 90 days — is the standard shape. A fixed reserve holds a flat amount until conditions change. Either way, the trigger is a number, and the number is a fraction.

Understanding that fraction is the whole game.

The denominator is doing more work than you think

Your chargeback ratio is disputes divided by transactions. Simple. But the two halves of that fraction move on completely different clocks, and almost nobody notices until it hurts.

The numerator — disputes — is a lagging indicator. A cardholder can file weeks or months after the sale. The disputes hitting your account this month were largely earned by sales you made two or three months ago. You cannot change them. They are already in flight, like light from a dead star.

The denominator — transactions — is a leading indicator. It reflects today. It reflects the ad campaign you paused, the seasonal slump, the launch that slipped, the week your checkout broke.

Now run the scenario. You had a great quarter: 10,000 transactions, 40 disputes. That's 0.4% — comfortable. Then sales soften. Next month you do 2,000 transactions. But the disputes arriving are still the tail of that great quarter: 35 of them. Your ratio just went from 0.4% to 1.75%. You did nothing wrong. You sold less, and the math punished you for it — right at the moment your cash position was weakest.

This is procyclical by design, and it is why reserves so often land on merchants who are already struggling. It is also why the standard panic response — pause the ads, cut spend, slow down — is precisely the move that makes the ratio worse.

Denominator neglect, and why smart operators miss this

There's a well-documented cognitive pattern behind the blind spot. Psychologists Valerie Reyna and Charles Brainerd have studied what they call denominator neglect: when people evaluate a ratio, attention gravitates toward the numerator — the vivid, countable, emotionally loaded number — while the denominator stays abstract. In their experiments, people asked to draw a winning marble from bowls will often prefer 9 winners out of 100 over 1 winner out of 10, because nine feels like more than one. The larger absolute count crowds out the proportion. It's part of a broader account of how we reason in gist rather than in precise numbers.

Disputes are exactly the kind of number that hijacks attention. Each one has a name attached, an angry email, a reason code, a fee. Transactions are a faceless total on a dashboard. So merchants track disputes obsessively and never once ask: what did my volume do this month? They optimize the number they can feel and ignore the number that's actually driving their standing with the network.

The card networks are not neglecting it. Visa and Mastercard both run merchant monitoring programs — Visa's dispute monitoring framework and Mastercard's Excessive Chargeback Program among them — that evaluate merchants on ratio thresholds calculated monthly, with counts and percentages that must both be exceeded before enforcement kicks in. Thresholds change; check your processor's current documentation rather than a number you read in a blog post. What matters is the shape: both a count and a rate. A tiny merchant with six disputes usually isn't in trouble. A tiny merchant with six disputes and forty transactions is.

The part that feels most unfair: winning doesn't erase the count

Here is the fact that reliably knocks the wind out of merchants who've been diligently fighting every dispute.

In the monitoring programs, a chargeback generally counts when it is filed, not when it is decided. You can win representment, recover the funds, prove the customer was lying — and the dispute still sits in the numerator for that month. Winning gets your money back. It does not get your ratio back.

This is not a loophole or an oversight. From the network's perspective, the ratio is measuring friction — how often your customers reach for the dispute mechanism at all — not how often you were right. A merchant whose customers dispute constantly but who wins every case is still generating enormous cost across the system. The ratio is a smoke detector, not a court.

Which reframes the entire strategy. Fighting disputes protects your revenue. Preventing disputes protects your ability to process at all. These are two different jobs, and merchants under a reserve almost always over-invest in the first and ignore the second, because winning feels like progress and prevention feels like nothing happening.

Your next moves

  • Calculate your real ratio the way the network does — monthly, not lifetime. Pull disputes filed in the last calendar month, divide by transactions processed in that same month, and write it down. Do it again for the three months prior. If the line is climbing while your sales are falling, your problem is the denominator, and no amount of rebuttal-writing will fix it.
  • Read your reserve notice for the release terms, not the reasoning. Find three specifics: the percentage held, the rollout period before funds release, and the stated conditions for review. Those are the only levers. Emailing a defense of your product quality is not a lever.
  • Rebuild a 13-week cash forecast with reserved funds excluded. Treat held money as if it does not exist until its release date. Most reserve-driven business failures are liquidity failures, not revenue failures — the company was solvent and still couldn't make payroll.
  • Fix your billing descriptor and your refund path this week. Two of the cheapest dispute-prevention moves in existence: make sure the name on the customer's statement is recognizable, and make sure canceling or refunding is easier than calling the bank. Every customer who reaches your support inbox instead of their issuer is a dispute that never enters the numerator.
  • Do not cut sales volume to look safer. It has the opposite effect. If you need to reduce risk, tighten the inputs — screen high-risk orders, pause the traffic source your disputes cluster around — rather than shrinking total transactions.

The uncomfortable symmetry

A reserve is what happens when a processor stops trusting your future and starts collateralizing it. You get that trust back the same way you'd get it back from anyone: not by arguing, but by producing a boring stretch of time in which nothing bad happens. Steady volume. Falling dispute counts. Months, not weeks.

That's the hard part. Because the merchant under a reserve is the merchant with the least cash, the least sleep, and the least bandwidth to respond to disputes on time — and every dispute that lapses unanswered is a loss recorded against them, deepening the exact number that caused the hold. The trap closes from the inside.

Which is why the mundane discipline matters more than the heroic one. Every dispute answered on time, with real evidence, before the deadline, is a small vote against that spiral. Argeback exists for merchants living in that squeeze: it pulls your Stripe disputes in automatically, assembles an evidence-backed response from what you already have, and files it inside the 7-day window — from your phone, on a Tuesday morning, when you have nine minutes and no bandwidth. It won't lift a reserve by itself. Nothing does but time. But it keeps you from losing the disputes you should have won while you wait.

If your payouts are held and your deadlines are slipping, see how it works.