Most cross-border payment losses don't show up as a single line item. They accumulate in basis points — a few here on FX conversion, a few there on unmatched settlements, a handful more in dispute write-offs that were never escalated. By the time a finance team runs a quarterly revenue reconciliation, the gap between gross payment volume and net cash received is somewhere between 3 and 5 percent. That number rarely surprises the CFO. What surprises them is how hard it is to explain precisely where each fraction went.
This article breaks down the four primary leakage categories we see across merchants operating US-to-Brazil, EU-to-Nigeria, and US-to-Philippines corridors. The categories are not equal in size or equally addressable — which matters for prioritization.
Category 1: FX Conversion Spread and Rate Timing
The most visible cost and the most misunderstood. When a merchant quotes a price in USD and settles in BRL or NGN, the conversion happens at whichever rate the PSP applies at settlement — not the rate that existed when the customer checked out. On a T+2 settlement cycle, that's two business days of FX drift. On corridors with higher volatility, like USD/NGN or USD/IDR, intraday moves of 0.3–0.8% are not unusual during periods of central bank intervention or capital flow events.
The spread itself is a second layer. PSPs typically apply a markup over the interbank mid-rate — commonly 50–200 bps depending on the corridor and the merchant's contracted rate. A merchant doing $2M/month through a PSP charging 150 bps on USD/BRL conversion is paying $30,000/month in conversion markup alone, before any other fees. Many merchants don't have this number precisely because the PSP settlement report buries it in a blended "net settlement amount" field rather than itemizing the FX line separately.
This isn't an argument against using PSPs for FX — it's an argument for having the data to know exactly what rate was applied and when, so the FX cost becomes a known, managed input rather than an unexplained residual.
Category 2: Settlement Exceptions and Unmatched Items
Consider a mid-size digital goods merchant processing approximately $800K/month across four corridors. Their PSP generates daily settlement reports in varying formats: one corridor delivers a CSV with transaction-level line items; another delivers an aggregate batch file with a single net figure per currency. The ops team manually reconciles these against their order management system each week.
In that environment, unmatched items — transactions that appear in the PSP settlement report but have no corresponding order record, or vice versa — typically run at 1.5–3% of transaction count. Most of these are benign: timing differences where an authorization posted to the OMS on day N but settled on day N+1, pushing it into the next batch window. But a meaningful subset are genuine exceptions: duplicate settlements, partial captures that don't match the authorized amount, or refunds that hit the settlement report before the refund event was recorded in the OMS.
Exception items that aren't worked within 30 days tend to age out. Some PSPs have contractual windows — typically 45–90 days — after which disputed settlement items cannot be raised. Exceptions that age out are permanent losses. At 1% unrecovered exception rate on $800K/month, that's $8,000/month in recoverable revenue that simply disappears because no one had a systematic process to work the queue.
Category 3: Chargeback and Dispute Write-offs
Chargebacks in cross-border contexts carry costs beyond the face value of the disputed transaction. The standard chargeback fee (typically $15–25 per dispute in Visa/Mastercard networks) is the smaller issue. The larger cost is the operational overhead of responding to disputes across multiple dispute rails with different evidence requirements and response windows.
Local payment methods add complexity. PIX disputes in Brazil operate through the Banco Central's MED (Mecanismo Especial de Devolução) system, which handles alleged fraud and operational errors differently than a standard card dispute. M-Pesa dispute resolution in the Kenya-US corridor goes through Safaricom's internal process with its own timeline and documentation requirements. A merchant without someone who understands each rail's process will either miss response windows or submit inadequate evidence — both result in losses that were avoidable.
We're not saying chargebacks can be eliminated or that the current dispute rail complexity is the merchant's fault to fix. We're saying that a significant fraction of chargeback losses — industry estimates typically put recoverable disputes at 20–40% of disputed volume for merchants with proper evidence management — are recoverable with the right operational process.
Category 4: Interchange and Scheme Fee Mischarges
This category is smaller in absolute dollar terms but particularly opaque. Interchange fees are set by the card schemes (Visa, Mastercard) and vary by card type, merchant category code (MCC), and whether the transaction is domestic or cross-border from the issuer's perspective. PSPs pass these through either at cost (interchange-plus pricing) or as a blended rate.
Mischarges occur when transactions are bucketed into the wrong interchange category. A cross-border corporate card transaction that should qualify for one interchange tier but is downgraded — perhaps because the authorization data was incomplete — moves to a higher tier. The difference can be 40–80 bps per transaction. On a portfolio with significant B2B volume, systematic miscategorization of corporate card transactions compounds quickly.
Scheme fees — the additional per-transaction assessments that Visa and Mastercard levy on top of interchange — are another area where merchants frequently overpay simply because they lack the transaction-level data to audit them against the published fee schedules.
The Compounding Problem
These four categories don't operate in isolation. An unmatched settlement item is also an FX exposure if it sits in a suspense state while the local currency moves. A chargeback on a transaction that was already improperly categorized for interchange means the dispute evidence must account for the correct transaction amount — which requires knowing the FX rate applied, not just the settlement file's net number.
The compounding is why the 3–5% estimate holds across corridors and merchant sizes even though the composition varies. A merchant with tight FX management but poor exception workflows will still land in the range. So will a merchant with clean reconciliation but no dispute response capability.
Addressing all four categories simultaneously requires a single normalized view across PSP settlement data, order system data, and bank statement data — matched at the transaction level. Without that foundation, teams optimize one category while leakage in the others goes undetected. That's how 3–5% becomes the floor rather than the ceiling.