Rolling reserves, chargeback ratios and the numbers that bite

The line on the statement reads "reserve funding," and it's a percentage of everything you sold last Tuesday. Nobody called. Nobody emailed. The money is simply not in the deposit, and it won't be for another six months.

That's how most operators meet a rolling reserve. After it's already started.

So let's do this in the order you'll actually need it: the thresholds that trigger trouble, the arithmetic of what a reserve costs you, and the handful of cheap fixes that move your ratio before anybody's risk team notices you.

Two numbers, and both have to break

Card network monitoring programs run on a count and a ratio. Not one or the other. Both.

Under Visa's dispute monitoring program, as the rules currently stand, you land in the standard tier at 100 disputes in a month and a 0.9% dispute ratio. The excessive tier sits at 1,000 disputes and 1.8%. Mastercard's excessive chargeback program works on similar logic with different figures, historically 100 chargebacks and a 1.5% ratio, with a higher tier above that. The networks revise these. Get the current version in writing from your acquirer rather than trusting a blog post, including this one.

The part operators get wrong is the "and." A merchant doing 300 transactions a month with 12 disputes has a 4% ratio, which sounds catastrophic, and isn't in any program, because 12 isn't 100. A merchant doing 400,000 transactions with 1,600 disputes has a 0.4% ratio and is fine on paper while eating four figures a week in dispute fees.

Your acquirer, though, has its own opinion. Most start asking questions well below the network line. Somewhere around 0.5% you stop being a normal account and start being a file on someone's desk.

The denominator isn't what you assume

Visa measures disputes in a month against sales transactions in that same month. Mastercard has measured chargebacks in a month against the transaction count of the preceding month.

That difference matters enormously if your volume moves. Say you do 20,000 transactions in November and 8,000 in December after the seasonal rush ends. The chargebacks from November's sales arrive in December and January, because cardholders take weeks to dispute. Under a prior-month denominator, December's chargebacks are divided by November's big number and look fine. January's chargebacks get divided by December's small number and your ratio doubles without a single new unhappy customer.

Every business with a seasonal peak or a hard stop on paid acquisition has walked into this. You cut ad spend, volume drops, the ratio spikes on the tail, and you get a letter.

Plan the ratio like a lagging indicator. Because it's one.

What a reserve actually costs

A rolling reserve means the processor holds an agreed percentage of each day's settlement and releases it a fixed number of days later. Common terms are 5% to 10% held for 90 to 180 days.

Run it. Ten percent on $500,000 of monthly card volume, held 180 days. The reserve fills at $50,000 a month and plateaus at $300,000. That's $300,000 of your money sitting in someone else's account, permanently, for as long as you keep processing.

Two things about that plateau. First, once it's full, your cash flow is normal again, because the daily release matches the daily hold. The pain is entirely in the ramp, which is where most businesses get hurt, because the ramp lands exactly when growth is already consuming working capital.

Second, you don't get it back by behaving well. You get it back by stopping. Standard terms hold the balance for six to nine months after your last transaction. If you switch processors, the old reserve sits there until that clock runs out, while the new processor starts its own.

There are variants. A capped reserve holds a percentage until a fixed dollar amount is reached, then stops. That's far better than a true rolling reserve, and it's worth asking for by name. An upfront reserve wants the money now. A delayed settlement, say T+7 instead of T+2, is sometimes offered as an alternative and is usually easier to live with.

Why yours got imposed

Rarely one thing. Usually a stack.

• High average ticket with delivery in the future. A $4,000 order shipping in eight weeks is eight weeks of exposure.
• Subscriptions, memberships, anything that rebills.
• Volume that grew faster than the forecast on your application. Underwriting priced you as a $200k-a-month business and you're doing $900k.
• A chargeback trend heading the wrong way for three consecutive months.
• Your merchant category code, regardless of how you personally behave.
• A refund backlog. Unfulfilled orders are a liability the processor may have to cover.

Notice how many of those are success. Reserves aren't a punishment for being bad at business. They're a hedge against the acquirer being left holding refunds if you close the doors on a Friday.

Getting it reduced

Ask for a written step-down. Not "we'll review it," but a schedule: reserve drops from 10% to 7% after two consecutive quarters below a stated ratio, then to 5%, then to nothing. Processors do agree to this. They rarely offer it unprompted.

Bring evidence. Fulfilment times, refund turnaround, dispute win rate, the specific changes you made and the month they took effect. A risk analyst can approve a reduction if there's a paper trail to point at. They can't approve it off a phone call where you say things are better now.

Offer an alternative form of security. A letter of credit or a personal guarantee will sometimes substitute for cash. Whether either is sensible for your situation is a question for your own lawyer and accountant, not a question to answer in a meeting you didn't prepare for.

And read the reserve clause before you sign anything. Most merchant agreements let the processor change the reserve at its discretion with minimal notice. That's the single most expensive sentence in the document, and it's usually on page nine.

The fixes that move the ratio cheaply

Start with your billing descriptor. A meaningful share of disputes are people who genuinely don't recognize the charge, and the descriptor is the only thing they see. Put the name customers know, not your holding company. If the field allows it, include a phone number. A customer who calls you costs you a refund. A customer who calls the bank costs you a chargeback, a fee, and a tick on your count.

Then the timing of rebills. Send a notice a few days before a subscription charges, with a one-click cancel. You'll lose some renewals. You'll lose fewer than you think, and the ones you lose were going to dispute.

Sign up for pre-dispute alerts through the networks' resolution services. An alert lets you refund within a short window and stop the chargeback from being recorded at all. You eat the sale. You keep the ratio.

Which brings up the trade-off nobody likes. On a $29 order, fighting a dispute is usually a mistake. Compiling evidence, writing the rebuttal and waiting eleven weeks costs more in staff time than the $29, and you might lose anyway. Refund fast, keep the count down, and reserve your fighting for the $800 orders where you have delivery confirmation and a signature.

On the fraud side, note that the networks run separate fraud monitoring measured in dollars, not just counts, based on what issuers report. You can sit comfortably under the dispute thresholds and still trip a fraud program. Ask specifically about both.

If you're already in a program

You'll typically get a workout period to get back under the line, with monthly fines escalating the longer you stay in. Your acquirer can tell you the exact schedule; it's in the network rules, and it gets worse every month you ignore it.

Treat the letter as a deadline with a number attached. Work out how many disputes you must avoid to get under the ratio at your current volume, then go find them. Usually it's one product, one ad creative, one fulfilment partner, or one billing page. It's almost never spread evenly.

Pull your ratio for the last six months. Both networks, both denominators, month by month. If nobody at your processor can produce that inside a day, you've learned something more important than the number.