# How a cross-border refund is handled without a second conversion When a cross-border deal unwinds, a second FX conversion can quietly erase value for every party. This article explains how onchain routing eliminates that problem at the architecture level. --- There is a specific type of loss that professionals working on cross-border deals almost never see as a single line item. It does not arrive as a fee notice. It does not appear in the wire confirmation. It surfaces only when someone reconciles the numbers weeks later and realises that the refund — the amount sent back to the buyer, or redistributed among parties — is materially smaller than the amount originally paid. The culprit is the second currency conversion: the one that happens on the way back. This article is for the settlement agents, title officers, closing attorneys, OTC desks, and brokers who coordinate the money flows on these deals. It explains exactly how the double-conversion problem occurs, why it compounds with time, how it cascades when multiple parties share a distribution, and how structuring a deal payment through onchain routing eliminates the problem structurally rather than managerially.
USD $3,800shortfall on a USD $126,000 deposit refund wired back to an Australian buyer
6 to 10 daysof constrained working capital per payment cycle when a payment is converted twice
Up to USD $90in opaque deductions on a partial return routed through three correspondent banks

Figures from the worked examples detailed in this article: a USD $180,000 deposit unwind and the correspondent banking timings and fees it describes.

## Why cross-border refunds cost more than the original payment International payments must bridge differences in time zones, currencies, and regulations, which is why international transfers often take longer and cost more, even when the transaction itself is straightforward. That friction is well understood on the outbound leg of a deal. What is less discussed is how it doubles when a deal unwinds. Consider the mechanics. If the transaction involves different currencies, FX conversion is applied based on provider rates. When a payment must be converted, legacy banks and processors rarely give the true interbank exchange rate. Instead, they add an FX markup, frequently between 1% and 3%, on top of the base conversion fee. That markup applies on the way in. On a partial refund or a full unwind, an equivalent or worse markup applies on the way back out — because the market has moved, because a different correspondent bank handles the return leg, and because credit card networks do not refund their cross-border assessment fees when a merchant processes a customer return. In fact, processing a refund can sometimes trigger an additional transaction fee, meaning you lose money on both the initial purchase and the return processing costs. For wire-based flows, the problem is structural rather than exceptional. FX losses rarely stem from bad luck — they come from how traditional systems handle money movement: forced conversions where buyers pay in one currency but banks automatically convert it to another before settlement; double FX hops where funds pass through multiple intermediary banks, each adding its own markup; and delayed settlements where holding periods expose funds to rate fluctuations before conversion. Those three dynamics — forced conversion, double hops, and rate timing — are the engine of refund leakage. And because the amounts look small per transaction, FX leakage is often missed in reporting as it doesn't show up as a single, obvious fee. Currency-related losses are usually scattered across small refund variances and reconciliation adjustments created by multi-currency settlement and timing differences. Since the amounts can look insignificant per order, they rarely trigger investigation — until volume grows and the cumulative impact becomes hard to ignore. ## The compounding timeline: why delay makes refunds worse Every professional who has managed an international unwind knows that refunds rarely happen the same day the deal breaks. There are disputes to resolve, legal sign-offs to obtain, and in many jurisdictions, contractual notice periods to observe. Every cross-border payment made in a foreign currency carries the risk that exchange rates could shift between when it's initiated and when it settles. Even the smallest fluctuations can erode profitability. The timeline problem is compounded by the mechanics of correspondent banking. Each currency conversion leg takes three to five business days. Double conversion means six to ten days of constrained working capital per payment cycle. On a large deal — say, a USD $2,000,000 (AUD $3,100,000) cross-border acquisition — each of those conversion windows represents days during which the exchange rate is live and unpredictable. Multi-day settlement windows expose businesses to FX fluctuation risk. A 2% exchange rate shift during settlement adds another layer of unpredictable currency conversion costs. Add in the correspondent bank chain, and the arithmetic becomes uncomfortable. International payments often involve multiple middleman institutions called correspondent or intermediary banks. Each intermediary along the path can deduct a fee ranging from $10 to $30 directly from the principal amount without warning. On a refund involving a USD $500,000 (AUD $775,000) partial return routed through three correspondents, that means up to USD $90 in opaque deductions — before the FX spread has even been applied. For a closing attorney or settlement agent, this creates a practical problem: the refund amount agreed upon in the mutual release document is not the amount that lands. The parties sign a clear number, but the infrastructure delivers a different one. Reconciliation becomes a negotiation. No professional wants to explain to a client why the wire came up short. ## A concrete scenario: the international property deal that unwinds Picture this. A buyer based in Australia and a seller based in the United States agree on a commercial property acquisition at USD $1,800,000 (AUD $2,790,000). The deal includes a USD $180,000 (AUD $279,000) deposit, already wired from the buyer's Australian bank account. Three parties are to receive distributions at closing: the seller takes the net proceeds, a buyer-side broker receives a success fee, and a title company holds a service retainer. Six weeks before the scheduled close, the deal falls apart. In a mutual release, the deposit return becomes a negotiation point: in some cases, buyers receive a full deposit return; in others, parties might split the deposit if both contributed to the transaction failure. In this scenario, the parties agree on the following split. | Party | USD | AUD | Agreed in the release | | --- | --- | --- | --- | | Buyer | $126,000 | $195,300 | Goes back to the buyer | | Seller | $36,000 | $55,800 | Retained for costs incurred | | Title company | $18,000 | $27,900 | Covers work already performed | | **Deposit** | **$180,000** | **$279,000** | **Already wired by the buyer** | Under a traditional wire flow, here is what actually happens. The title company initiates the return wire to the Australian buyer. That wire converts USD to AUD at the prevailing rate. FX rates can fluctuate, and the time when a currency conversion occurs can make a difference in the final amount. The FX rate can indeed be different at the time of purchase compared to when the transaction is actually settled. The buyer's bank applies its own spread on receipt. The USD $126,000 agreed in the release document becomes AUD $189,400 instead of the AUD $195,300 implied by the spot rate — a USD $3,800 shortfall that neither party explicitly authorised but that neither bank will reverse. This scenario involves conversion fees at the time of both the original purchase and the refund. The seller's retained USD $36,000 (AUD $55,800) stays domestic and lands correctly. But the title company's USD $18,000 (AUD $27,900) involves its own wire to a third-party law firm in a different state, arriving slightly short due to intermediary deductions. The closing file now contains two separate reconciliation discrepancies across a deal that, contractually, was completely clean. ## Why a split refund makes the problem worse The scenario above involves one refund to one cross-border recipient. Professional deals are rarely that simple. A more realistic multi-party structure — a syndicated acquisition, a cross-border JV, a commercial sale with multiple agents — might route a single incoming payment to four or five parties simultaneously. When that deal unwinds, the refund logic has to reverse or redistribute across the same structure. This is precisely where the architecture of traditional wire rails fails. Sending wire transfers is a relatively opaque process once the transfer is sent. Neither the sender nor the recipient receives detailed in-process status reports before the wire transfer is complete. In a multi-party unwind, the settlement agent must initiate four or five separate return wires, each of which travels a different correspondent chain, each of which applies its own conversion rate, and each of which arrives on a slightly different day. The split that was clean and proportional on the term sheet is distorted by the time it reaches recipients' accounts. Even a 1–2% FX spread across large volumes can cost hundreds of thousands annually. Scaled across dozens of unwound or partially restructured deals per year — a normal volume for any active closing firm or OTC desk handling cross-border transactions — those basis points become a material number. More importantly, they create relationship friction: clients who signed clear documents receive unclear amounts, and the professional in the middle is left explaining infrastructure they do not control. ## How onchain routing changes the architecture of refunds The root cause of every problem described above is that traditional infrastructure separates the denomination of a deal from the mechanism of its payment. A deal is agreed in USD; the payment traverses a chain of institutions each converting, marking up, and moving value in ways invisible to the parties. When the deal must reverse, the chain runs again in the opposite direction, with a different FX rate and a different set of correspondents. Onchain routing on Ethereum, as used by shaka.deal, removes that separation. The deal is denominated in a USD-pegged stablecoin. The payment is routed — once, atomically — to all parties simultaneously, at the shares preset in the deal structure. There is no conversion on the payment leg because there is nothing to convert: both sides transact in the same denomination. If foreign exchange conversion is required, it typically occurs at one or both endpoints rather than being embedded throughout a chain of correspondent banks. This is the critical architectural point. When a deal is structured onchain from the outset, a refund is not a reversal of a currency conversion. It is a new transaction, in the same denomination, to the same or redistributed addresses. The stablecoin sent back to the Australian buyer arrives as USDC — the same asset, the same value, without a second conversion. The buyer then converts at a single point, once, at the best rate they can access, on their own timeline. A stablecoin payment is one on-chain transfer, with a regulated on-ramp converting fiat to stablecoin at the start and an off-ramp converting back to local currency at the end. The conversion that does happen — fiat to stablecoin at the on-ramp, stablecoin to local fiat at the off-ramp — occurs transparently, at a point the party controls, and only once per direction of travel. Because settlement occurs directly between participants, institutions gain continuous payment visibility without relying on multiple correspondent banks. Payments are also not limited by traditional banking cut-off times, and every transaction produces a single reconciliable record. ## The finality question: why a refund onchain is a deliberate new payment, not a reversal One question that settlement professionals always raise when considering onchain payment rails is this: if a deal unwinds, can the original payment be reversed? The honest answer, and the professionally important one, is: no. Finality is the guarantee that past transactions in a blockchain network cannot be altered, reversed, or canceled. Its primary purpose is to provide absolute certainty to users, merchants, and smart contracts that a transaction is permanently settled and digital assets are secure. For a professional audience, this is not a limitation — it is a feature of the system that must be correctly understood and planned for. Where traditional settlement relies on institutional rules and operational procedures to establish finality, blockchains achieve it through cryptography and economic deterrence. On a proof-of-stake network like Ethereum, a transaction becomes technically final when it is incorporated into a validated block and subsequently confirmed. That finality means the original payment cannot be recalled by a back-office correction or administrative instruction. The economic cost of reversing even a handful of confirmations makes rollback practically impossible. This is actually cleaner from a documentation standpoint than the traditional model. The finality is a feature for settlement certainty and an operational consideration for refund and dispute workflows. The settlement agent drafts a release agreement, the parties sign it, and the agent initiates a new outbound transaction at the agreed amounts. Every party receives exactly what the document says, instantly, simultaneously, with a transaction hash that serves as immutable proof of settlement. There is no "funds in transit" window, no correspondent mystery, no shortfall to explain. Compare this to the traditional model, where ACH payments can be reversed for days after initiation, and card transactions take days to settle and can be subject to reversal for weeks. That reversibility, which sounds like a consumer protection, is also a source of uncertainty for every professional who has waited to confirm that funds are truly settled before releasing obligations. Onchain settlement replaces that uncertainty with an auditable, timestamped record. ## The mechanics of a shaka.deal refund flow Here is how the same scenario — the USD $180,000 (AUD $279,000) deposit from the Australian buyer — operates when the original deal is structured through shaka.deal.
The refund flow through shaka.dealSetup, distribution and unwind of the deposit
  1. At deal setupThe closing attorney or settlement agent configures the payment distribution in shaka.deal: preset shares for the seller, buyer-side broker, and title company. The Australian buyer on-ramps via a regulated provider, converting AUD to USDC at a single, transparent rate. The USDC equivalent of USD $180,000 (AUD $279,000) is sent to the routing contract.
  2. At distributionshaka.deal routes the total amount to every party simultaneously, in one transaction, at the preset shares. The seller receives their USDC. The broker receives their USDC. The title company receives their USDC. The transaction hash captures the entire distribution in a single, auditable record. No correspondent banks. No multi-day settlement. No per-party mystery.
  3. At unwindThe parties sign the mutual release. The settlement agent initiates a new routing transaction through shaka.deal: USD $126,000 (AUD $195,300) equivalent USDC to the buyer's wallet, USD $36,000 (AUD $55,800) to the seller, USD $18,000 (AUD $27,900) to the title company. These three amounts arrive simultaneously, in one transaction, at full USDC value. No FX markup on the routing leg. The buyer off-ramps USDC to AUD at a single transparent rate of their choosing. What the mutual release document says is what every party receives.
The closing file now contains two transaction hashes — one for the original distribution, one for the unwind — and a mutual release that matches both to the cent. Reconciliation is not a negotiation. It is a lookup. ## What settlement professionals gain from this structure The professionals coordinating these deals — title companies, settlement agents, OTC desks, closing attorneys, international brokers — do not gain from refund leakage. They gain from clean files, from clients who receive exactly what the contract specifies, and from deals that close (or unwind) without residual accounting disputes. Onchain routing through shaka.deal gives these professionals a concrete tool for achieving that outcome on cross-border deals. The split is preset before the payment is initiated. The distribution is simultaneous and final. The refund, when it becomes necessary, is a new deliberate transaction in the same denomination, not a reversed conversion. Settlement compresses from days to seconds, fees fall to single-digit cents on most chains, and the dollar balance becomes programmable — meaning a payment can carry conditions and chain into longer treasury workflows. For a settlement agent managing a portfolio of international deals, that programmability means the refund logic can be built into the deal structure from day one, not assembled reactively when an unwind becomes necessary. There is no separate clearing message, and no overnight batch to reconcile against a custodian's records. For a title company that issues a closing statement, that means the statement and the blockchain record are the same document. There is no gap between what was agreed and what was settled. ## The broader context: cross-border payment volume is growing The stakes of getting this right are increasing. Cross-border payments continue to gain momentum worldwide, with their market projected to reach USD $320.7 billion by 2030, growing at a 7.1% CAGR from 2025 to 2030. This is a noticeable increase from 2024's $212.5 billion market size. The increasing volume of B2B payments between businesses and international suppliers is a major driver of this growth. As cross-border deal volume grows, so does the frequency of cross-border unwinds, partial refunds, and mid-deal redistributions. Global corporations move around $23.5 trillion across borders every year and pay an estimated $120 billion in transaction fees, a large share of which comes from FX spreads and conversion costs. Unmanaged FX risk adds unpredictable costs, slows settlements, and makes reconciliation harder. The professionals managing these transactions are the ones absorbing the relationship cost of refund discrepancies that originate in infrastructure they did not design and cannot directly fix. Onchain routing does not replace those professionals. It gives them a payment architecture that produces the outcome their documents already specify. ## Practical considerations for professionals structuring cross-border deals onchain Three questions come up consistently from settlement professionals exploring onchain deal routing. **Who handles the on-ramp and off-ramp?** The parties handle their own conversion, through a regulated on-ramp provider of their choice, before funds enter the routing contract. shaka.deal routes. It does not hold, convert, or custody funds. The conversion layer sits outside the routing contract, at the endpoints, under the parties' control. If foreign exchange conversion is required, it typically occurs at one or both endpoints rather than being embedded throughout a chain of correspondent banks. **What happens if the refund amount is disputed?** The same thing that happens under traditional settlement: the parties negotiate, document the agreed amounts, and then execute. The difference is that once the parties have agreed and the transaction is initiated, the amounts are final. There is no "it should arrive in three to five business days" qualifier. There is no correspondent bank deduction to add to the dispute. When the asset leg and the cash leg are bound into one atomic transaction, neither side can complete without the other. **How does the settlement agent retain visibility and control?** shaka.deal is non-custodial. The routing contract executes the distribution according to the preset shares when the transaction is submitted. The settlement agent configures those shares. Every party can verify the transaction on the Ethereum public ledger. The agent's role in designing, documenting, and coordinating the deal structure is exactly as it was before — with the addition of a verifiable, immutable record that supports rather than complicates their professional work. ## Closing point The second FX conversion is not a law of nature. It is a consequence of building a deal payment on infrastructure that was designed for single-currency domestic transfers and adapted — imperfectly — for cross-border use. When the deal denomination and the payment denomination are the same from the outset, the second conversion simply does not occur. A refund, in that architecture, is what it should always have been: a clean, documented, simultaneous distribution to every party, at the exact amounts the parties agreed, settled in one transaction, with a timestamp that everyone can verify. That is what shaka.deal produces. That is the standard that professionals managing cross-border deals should expect from their payment infrastructure.