Rolling Reserves Explained: What Processors Hold, Why, and How to Get It Down

The approval email says yes. Then you get to the third paragraph. Ten percent of every batch, held for one hundred and eighty days. Nobody explains that at $100K a month, that clause quietly parks $60K of your money in someone else's account. Most founders sign it anyway, because the alternative is not processing at all. Six months later they are staring at a profitable P&L and an empty operating account. A reserve is not a fee and not a punishment. It is collateral against a liability you keep carrying for months after the card is charged, and it is one of the most negotiable terms in the agreement. Here is what reserves secure, what they cost you in working capital, what typical structures look like at each tier, and what genuinely gets one reduced.

What a reserve is actually securing

When an acquirer boards you, it is guaranteeing your performance to the card networks. If a customer disputes a charge four months from now and you have already spent the money, the acquirer eats it. If you take orders and go out of business before shipping, the acquirer eats every one of those disputes. That exposure is its contingent liability, and the reserve is collateral against it.

The size of that liability is a function of time, not just risk appetite. Under Visa's rules a cardholder generally has 120 days from the transaction or expected delivery date to dispute, and some reason codes stretch considerably further for delayed delivery and interrupted services. Mastercard's windows are similar in shape. That is why 180 days is the default hold period in so many contracts: it is the dispute window plus a margin. It also explains why the verticals with the longest gap between payment and delivery get hit hardest. A travel merchant collecting for a trip nine months out is asking an acquirer to carry risk for the better part of a year, and reserves there are priced accordingly.

The important reframe: a reserve is not the processor's money. It is your money in a restricted account, released back on a schedule, and on your books it is an asset rather than an expense. Only the portion consumed by chargebacks, refunds, and fees ever becomes a cost. That distinction matters when you model cash, and when you argue for a lower number.

The three structures you will be offered

Rolling reserve

A fixed percentage of each settlement is withheld and released after a set period, usually 180 days. It rolls: today you fund a new tranche and receive back the tranche from six months ago. This is the high-risk default and the one most founders sign.

Capped reserve

The same mechanism, but withholding stops once the balance reaches an agreed ceiling, usually a dollar figure or a multiple of expected monthly disputes. After that, releases continue and no new money goes in unless the balance drops below the cap. This is meaningfully better for a growing merchant, because your reserve stops scaling with your revenue. Ask for it specifically, by name.

Upfront or fixed reserve

A lump sum deposited at boarding, held for the life of the account. Common when you have no processing history and the underwriter wants collateral up front. Painful on day one, but it does not tax growth, and it is a good trade if you can fund it.

The structure you accept matters more than a few basis points of discount rate, and it draws far less scrutiny.

The cash-flow math nobody runs

A rolling reserve does not cost you 10 percent of revenue forever. It costs you 10 percent during the ramp, then converts into a permanent balance-sheet lockup. The steady-state balance is simple arithmetic: monthly volume, times the reserve percentage, times the hold period in months.

At $100K a month with 10 percent held for 180 days, that is roughly $60K. At 5 percent held for 90 days, about $15K. At $250K a month on the same 10 percent, $150K.

The first consequence is the ramp. For the first six months nothing is released back to you yet, so it feels exactly like a 10 percent revenue cut. That is the window where merchants who scaled ad spend against gross revenue run out of cash while the dashboard says they are winning.

The second is that growth makes it worse before it makes it better. Money goes in against current volume and comes back out against volume from six months ago, so a doubling leaves you funding larger tranches while receiving smaller ones. Fast growth on an uncapped rolling reserve is a working-capital trap, which is precisely why the cap matters.

There is a redundancy cost here too. Running multiple merchant accounts, as we recommend in the MID load balancing guide, usually means funding a reserve on each one. That is the real and often unbudgeted price of not being single-processor dependent, and it is worth paying, but size it deliberately rather than discovering it.

What reserves look like across the three tiers

Tier	Typical reserve posture	In practice
PSP aggregator (Stripe, PayPal, Square, Shopify Payments)	None stated at boarding; discretionary holds later, often 100 percent of the balance for 90 to 180 days	No friction until the risk review, then everything at once, with no contract term to point to
Domestic dedicated MID via a high-risk ISO	Commonly 5 to 10 percent rolling for 180 days, negotiable, sometimes capped or stepped down	Predictable, priced, reducible with performance
Offshore acquiring	Often 10 percent or higher, longer holds, closure reserves running past 180 days	Highest cost of capital, slowest recourse in a dispute over the balance

Read the aggregator row carefully, because an aggregator's lack of a stated reserve is not generosity, it is deferral. They underwrite after the fact, and the tool they reach for is a discretionary hold on your entire balance, which is far more disruptive than a predictable 10 percent. That sequence is the subject of why processors shut down high-risk accounts, and reserves are why a shutdown hurts long after processing has stopped. A negotiated reserve on a dedicated MID is a known, financeable cost. A 100 percent hold from an aggregator is an unplanned outage of your entire cash position.

Reserve, hold, and freeze are three different things

These get used interchangeably and they should not be. A reserve is contractual, disclosed at boarding, released on a published schedule. A hold is discretionary, applied while risk reviews a pattern, and lifted when the review clears. A freeze is termination: settlement stops and the balance is held through the tail of the dispute window, commonly 180 days after your last transaction and sometimes longer.

Know which one you are in, because the response differs. A hold is answered fast, with fulfillment records and tracking. A reserve is answered with performance data over months. A freeze is answered by having already built somewhere else to process.

What actually gets a reserve reduced

Reserves are reviewed, and the review is winnable on evidence delivered in the language underwriting already uses.

Processing history is the strongest lever. Six clean months on the account outweighs almost any argument you can make at boarding, so the realistic play is to accept a workable reserve now and schedule the review. Ask in writing, before you sign, what the reduction criteria are and when the first review happens. A provider that will not name criteria is telling you the reserve is permanent.

Dispute performance is the second lever, and the one you control. Present your ratios the way the networks calculate them rather than as a blended dashboard figure, which is the exercise in the chargeback ratio guide. A merchant arriving with 0.4 percent against Visa's monitoring threshold, month over month, backed by prevention alerts and clean fulfillment data, negotiates from a different position.

Then the structural asks, in rough order of how often they land: convert an uncapped rolling reserve to a capped one, step the percentage down on a defined schedule such as 10 percent for six months then 5 percent on review, shorten the hold from 180 to 90 days, or substitute collateral like a letter of credit. A personal guarantee is also on the table and is genuinely serious. It can lower the reserve materially, and it moves the liability onto you personally. Do not trade it away casually.

Underwriting quality sits underneath all of this. Clean financials, consistent descriptors, honest product claims, and enforced compliance controls make you cheaper to carry, and that is the entire basis of a lower reserve. The document pack that gets you approved, covered in the high-risk merchant account guide, is the same pack that gets your reserve down a year later.

One warning sign: if your processor raises your reserve mid-term without a matching change in your dispute numbers, they are managing their exit, not your risk. Treat it as a countdown and get a second account boarded.

Practical takeaway

The path that fails is treating the reserve line as boilerplate, signing an uncapped 10 percent for 180 days, scaling spend against gross revenue, and discovering the lockup during a cash crunch you cannot explain. The path that lasts is boring: model the steady-state balance before you sign, negotiate a cap and a written review schedule, run your dispute numbers the way the networks run them, then come back with six months of evidence and take the percentage down. Reserves are the price of honest underwriting in a risky vertical, and they are supposed to shrink as you prove the risk was overstated. If you want a second read on your reserve terms and what a realistic reduction looks like for your volume, apply for an architecture review.