Fee leakage is small as a percentage of payment volume and large as a percentage of margin. On a $100M book it runs $200,000 to $500,000 a year. But on the thin spreads a cross-border platform actually keeps, that is often more than half of net contribution. The number that should worry a CFO is not the leakage rate. It is what the leakage rate does to the bottom line, and how much of it stops being recoverable the longer nobody looks.
- Fee leakage on a cross-border book typically runs 0.2 to 0.5 percent of payment volume (Bluefyn design-partner data). On $100M, that is $200,000 to $500,000 a year.
- Stated as a share of volume it sounds trivial. Stated as a share of net margin it usually is not. At realistic cross-border take rates, $350,000 of leakage can be 50 to 100 percent of net contribution.
- Leakage recovered drops straight to the bottom line. Recovering it is one of the few "revenue" levers that costs no acquisition spend and no extra volume.
- It compounds. A percentage of a growing book is a growing number, and the same uncaught structural error repeats every period until someone catches it.
- The real cost driver is latency, not the leakage rate. Dispute and billing-query windows are contractually bounded, so unexamined charges age out of recoverability. On a $100M book, waiting until annual close instead of catching continuously can cost roughly $220,000 a year in recovery you can no longer claim.
- Hidden FX fees alone cost businesses an estimated $274 billion globally in 2025 (Wise, 2025 G20 Report). Fee leakage is the slice of that you are personally paying and can personally recover.
Short answer
Fee leakage costs a cross-border payments business 0.2 to 0.5 percent of its payment volume per year: $200,000 to $500,000 on a $100M book. The figure that matters to a finance leader is not that percentage of volume but the same dollars expressed as a percentage of net margin, where it commonly lands between half and all of net contribution because cross-border platforms operate on thin per-transaction spreads. The cost grows with volume and worsens with delay: the longer a charge goes unverified, the less of it can be disputed and recovered. Catching leakage continuously rather than at period close is worth roughly $220,000 a year in recoverable money on a $100M book.
The leakage rate is the least interesting number
Most coverage of fee leakage stops at the leakage rate. Zero-point-two to zero-point-five percent of volume. It is an accurate number and a forgettable one, because percentages of volume do not mean anything to a finance leader until they are converted into the only two units a CFO actually manages: margin and cash.
This piece does that conversion. It uses one platform ($100M in annual payment volume) and follows the money through three questions. What is the leakage worth against margin rather than volume? What happens to that number as the book grows? And how much of it can you still get back?
The leakage rate itself is the least interesting part of the answer.
The number is small against volume and large against margin
Take a platform processing $100M a year. Apply the middle of the leakage band, 0.35 percent. That is $350,000 a year, or $87,500 a quarter, leaving the business through provider charges nobody reconstructed.
| Leakage rate | Per year on $100M | Per quarter |
|---|---|---|
| 0.20% | $200,000 | $50,000 |
| 0.35% | $350,000 | $87,500 |
| 0.50% | $500,000 | $125,000 |
As a share of volume, 0.35 percent is a rounding error. That is exactly why it survives. It is too small to trip a variance check and too large to ignore once you see it in the right denominator.
The right denominator is net margin. Cross-border platforms do not keep their volume. They keep the spread between what they charge end clients and what they pay providers, and that spread is thin. Hold the $350,000 of leakage constant and run it against a few realistic margin profiles:
| Blended take rate | Provider cost | Net contribution on $100M | Leakage as % of net |
|---|---|---|---|
| 1.2% | 0.6% | $600,000 | 58% |
| 1.0% | 0.6% | $400,000 | 88% |
| 0.8% | 0.5% | $300,000 | 117% |
(Take rates here are illustrative; the leakage band is design-partner data.)
The same $350,000 that is 0.35 percent of volume is more than half of net contribution in every one of these cases, and in the thinnest one it exceeds net margin entirely. That is the structural reason fee leakage matters more to a cross-border business than to almost anyone else: the leak is sized against gross volume, but the wound is taken out of net margin.
There is a corollary that finance leaders tend to like. Leakage recovered does not have a cost of goods. It is not new volume you had to win, or a price increase you had to negotiate, or a market you had to enter. It is money already earned and quietly handed back. Recover it and it falls directly to net contribution. At an observed recovery rate of roughly three dollars in four, clawing back $350,000 of leakage adds about $270,000 to the bottom line of a business whose entire net contribution might be $400,000. Few growth levers convert that cleanly.
It compounds, because a percentage of a growing book is a growing number
A scaling platform does not stay at $100M. Model the same business growing 10 percent quarter over quarter (a $100M run-rate at the start of the year, exiting it well above that), and hold the leakage rate flat at 0.35 percent.
| Quarter | Volume | Leakage | Cumulative |
|---|---|---|---|
| Q1 | $25.0M | $87,500 | $87,500 |
| Q2 | $27.5M | $96,250 | $183,750 |
| Q3 | $30.3M | $105,875 | $289,625 |
| Q4 | $33.3M | $116,463 | $406,088 |
| Q5 | $36.6M | $128,109 | $534,196 |
| Q6 | $40.3M | $140,920 | $675,116 |
| Q7 | $44.3M | $155,012 | $830,127 |
| Q8 | $48.7M | $170,513 | $1,000,640 |
The leakage rate never moves. The dollars nearly double, because the rate is applied to a bigger book every quarter. By the eighth quarter the business is leaking at an annualized run-rate of around $680,000, and the two-year cumulative figure crosses a million dollars.
This is the part operators consistently underestimate. They assume scale brings an advantage: bigger book, tighter pricing, fewer surprises. On the cost-verification side the opposite is true. More volume means more corridors, which means more provider contracts, more invoice formats, more contract versions straddling a quarter boundary, and more places for the same structural errors to recur uncaught. The finance team most exposed to fee leakage is the one with the most volume, and it is the one with the least spare capacity to chase it down by hand.
The real cost driver is latency
Here is the number that should change behavior. The cost of fee leakage is not the leakage. It is the portion you can no longer recover by the time you find it.
Provider disputes are not open-ended. Billing-query and dispute windows are written into the contract, and they are usually measured in months, often a quarter, sometimes two. A charge you flag inside that window, with the underlying transaction, the contract clause, and the expected-versus-actual variance attached, is recoverable. The same charge surfaced at year-end close, eight months stale, usually is not. The money is identical. The recoverability is gone.
Run the first year of the growing book through two detection regimes.
| Detection regime | Recovered | Permanently lost |
|---|---|---|
| Continuous (catch each period, ~77% recovered) | ~$313,000 | ~$93,000 |
| Annual close only (most quarters already stale) | ~$90,000 | ~$316,000 |
Same business. Same $406,000 of leakage. The only variable is when you looked. Catching it continuously recovers roughly $313,000; waiting for the annual close recovers under $90,000, because by then three of four quarters have aged out of the practical dispute window.
The gap between those two outcomes, about $220,000 on a $100M book, is the pure cost of latency. It is money that exists, that you are owed, that you could prove, and that you will not get back for the single reason that you reconciled too late. No leakage rate produced that loss. The reconciliation cadence did.
This reframes the whole problem. The question is not "how much are we leaking," which is a rate you cannot easily change. It is "how fast do we detect," which is an operating decision entirely within your control.
Stop calling it a leak
A leak is an accident, a one-time event you patch and forget. Fee leakage is neither. It is a recurring cost line that scales with volume, compounds with growth, and converts into permanent loss on a contractual clock. On a $100M cross-border book it is a low-six-figure annual item today and a high-six-figure one within two years of normal growth. It belongs on the same page as your largest infrastructure costs, rather than in a footnote nobody reconciles.
Treating it that way means one thing operationally: every transaction gets priced against the contract that should have governed it, continuously, with evidence attached, so a discrepancy surfaces inside the window where it can still be recovered, rather than at a year-end close where it can only be regretted. That is the entire function of fintech infrastructure verification, the category a system like Bluefyn is built to run. The point is not that the leakage exists. Every cross-border platform leaks. The point is whether you find it while the money is still yours to claim.
The bottom line
Fee leakage costs more than its rate suggests for three compounding reasons. It is sized against volume but taken out of margin, where it commonly runs from half to all of net contribution. It grows as the book grows, because a flat percentage of a rising number rises. And it converts into permanent loss the longer it sits unexamined, because the windows to dispute it are contractually short.
For a platform processing $100M a year, the live cost is $200,000 to $500,000 annually, most of it recoverable if caught in time and most of it gone if not. The companies that scale cleanly through the next phase of cross-border growth will be the ones that treat that figure as a managed line item rather than an invoice to be glanced at, and that measure their performance against it not by how much they leak, but by how fast they catch it.
Frequently asked questions
How much does fee leakage cost a cross-border payments business?
Fee leakage typically runs 0.2 to 0.5 percent of payment volume per year. For a platform processing $100M annually, that is $200,000 to $500,000. Because cross-border platforms operate on thin per-transaction spreads, the same dollars usually represent between half and all of net contribution.
Why is fee leakage worse for cross-border platforms specifically?
The leak is sized against gross volume but taken out of net margin, and cross-border margins are thin. A cost that is 0.35 percent of volume can be 50 to 100 percent of the spread the platform actually keeps. Scale makes it worse, not better: more corridors mean more contracts, more invoice formats, and more contract versions where structural errors recur uncaught.
Does fee leakage compound?
Yes, in two ways. The leakage rate is applied to a growing book, so the dollar amount rises as volume rises. And the same uncaught structural error (a misapplied tier, a rate deviation, an omitted rebate) repeats every period until someone catches it. On a book growing 10 percent per quarter, leakage can nearly double over eight quarters even at a flat rate.
Can recovered fee leakage improve margin?
Yes. Recovered leakage has no cost of goods. It is money already earned and handed back, so it falls directly to net contribution. At a recovery rate of roughly three in four dollars, clawing back $350,000 of leakage can add around $270,000 to the bottom line, with no new volume or pricing change required.
Why does waiting to reconcile cost money?
Provider dispute and billing-query windows are contractually bounded, often a quarter or two. Charges flagged inside that window with evidence attached are recoverable; the same charges surfaced at year-end close are usually too stale to dispute. On a $100M book, catching leakage continuously rather than at annual close can be worth roughly $220,000 a year in recovery you can otherwise no longer claim. Latency, not the leakage rate, is the real cost driver.
How big is the broader hidden-fee problem?
Wise's 2025 G20 Report estimates hidden FX fees alone cost consumers and businesses around $274 billion globally in 2025. Against a B2B cross-border flow that analysts expect to grow from roughly $39 trillion in 2023 to $56 trillion by 2030 (Convera), fee leakage is the slice of that cost a given platform is personally paying, and the slice it can personally recover.
