In this essay
A cross-border payment is a payment where the payer and the payee hold money in different countries, and usually in different currencies. Between them sit two regulatory regimes, at least two banks, often a network or aggregator in the middle, and an exchange of one currency for another. Every one of those is a handoff, and every handoff is a place where the money can stop.
This essay walks one payment through the handoffs. The example is a marketplace paying a seller in Pakistan, one of the corridors Simpaisa operates; the shape is the same for a remittance, a business invoice or an app store paying a developer.
Step 1: Initiation
The payer (the marketplace) instructs a payment: pay this seller, this amount, in the seller's currency, by this date. The instruction carries the seller's identity, the destination account or wallet, the purpose, and the payer's reference.
The product question at this step is data quality. A large share of cross-border failures are caused by what is entered here: a wrong account number, a name that does not match the bank's record, a missing purpose code the destination regulator requires. The best corridor products validate the destination at initiation rather than discovering the problem three handoffs later.
Step 2: Compliance screening on the sending side
Before money moves, the sending institution screens the payer and the payee against sanctions lists, checks the purpose against what the payer is permitted to do, and applies its own risk rules. In regulated corridors both parties' identities must be verifiable; this is where know-your-customer and know-your-business data collected at onboarding is used.
A hit here stops the payment for review. The product question is how fast a human can clear a false positive and how the payer is told.
Step 3: Funding and FX
The sending institution needs to have, or obtain, the destination currency. There are two models. In the pre-funded model, the operator holds a balance in the destination currency (or with a local partner) and draws it down; payments are fast and the operator carries the funding cost. In the just-in-time model, the operator buys currency for each batch; funding cost is lower and speed depends on the FX counterparty.
The rate applied to the payer is set here, and so is the margin. Most of the fee in cross-border is FX margin rather than the visible transfer fee, which is why comparing corridors on the fee alone is misleading.
Step 4: Routing
The payment now has to reach the destination country. There are three main routes.
Correspondent banking: the sending bank instructs a bank it holds an account with, which instructs a bank in the destination country. Messages travel over SWIFT; money moves through the correspondent accounts. Reliable and universal, slower and more expensive, and the route most affected by de-risking, which I cover in correspondent banking and emerging-market corridors.
Network or aggregator: the payer's provider hands the payment to a cross-border network that has its own local partners in the destination, which pay out over domestic rails. Faster, priced per corridor, dependent on the network's local coverage. Our aggregator clients such as dLocal, Thunes and Boku operate this way and use Simpaisa as the local partner in our markets.
Local rails directly: for operators with their own licence and local presence, the payment can be paid out directly onto the destination's wallet, bank-transfer or real-time system. Fastest and cheapest per transaction; requires the licence, the local entity and the local compliance programme.
Which route a payment takes is a product decision, made per corridor and sometimes per transaction, based on cost, speed, the destination's rails and the payer's expectations.
Step 5: Compliance screening on the receiving side
The destination institution screens again, under its own regulator's rules. A payment cleared in the sending country can be held in the receiving country: different lists, different purpose-code requirements, different thresholds for enhanced checks. This is the step that surprises product teams who assume compliance is done once.
Step 6: Payout
The destination institution credits the payee: a bank account, a mobile wallet, a cash pickup, a card. In the markets we operate in, wallets and bank transfers dominate, and each rail has its own cut-offs, limits and failure modes. A payout can fail because the wallet is inactive, the account name does not match, or a daily limit is exceeded. Each failure needs a return path and a message to the payer.
Step 7: Settlement
Somewhere behind the payout, the institutions settle with each other: the operator funds its local partner, the network settles with its members, correspondent accounts are debited and credited. Settlement happens on its own cycle, often daily, in batches, and it is where a corridor's cash needs are decided. A corridor that pays out instantly but settles daily is being funded by someone for the gap.
Step 8: Reconciliation
Finally, every party checks that what it sent, what it received, what it paid out and what it settled all agree. Breaks (a payout that succeeded but was recorded as failed, a settlement that does not match the payout total, an FX rate applied differently at two steps) are found here and worked by an operations team.
This is the step where a corridor is actually run. A cross-border product without a reconciliation process is a corridor that will lose money it cannot find. Our settlement and reconciliation work is described in the settlement and reconciliation case study.
The companies at each step
Sending institutions: banks, licensed payment institutions, marketplaces and platforms with payments licences. FX providers: banks, specialist FX firms, the networks themselves. Routing: correspondent banks and SWIFT, cross-border networks and aggregators, local payment infrastructure companies. Receiving institutions: local banks, wallet operators, payout partners. Compliance tooling: screening and monitoring vendors used at steps two and five. Reconciliation and treasury tooling at steps seven and eight.
A "cross-border payments company" can be any of these. The useful question is which steps it performs itself and which it hands off.
What a corridor team does every day
Watches the funding balance against the day's expected payouts. Clears compliance holds on both sides. Handles payout failures and returns. Reconciles yesterday's files and works the breaks. Monitors partner performance per rail: success rate, time to credit, failure reasons. Adjusts routing when a rail degrades. Answers the payer's "where is my money" with a specific step rather than a guess.
That daily work is the product. I have argued elsewhere that corridors are operating systems, and this is what the operating looks like.
FAQ
How does a cross-border payment work? The payer initiates it; the sending institution screens it and obtains the destination currency; the payment is routed via correspondent banks, a cross-border network or local rails; the receiving institution screens it again and pays the payee; the institutions settle with each other; and every party reconciles.
Why do cross-border payments take so long? Each handoff can add time: compliance holds on either side, FX and funding cycles, correspondent-bank processing, destination cut-offs, and reconciliation breaks. Payments routed through networks with local partners are usually faster than correspondent banking.
Why are cross-border payments expensive? Most of the cost is FX margin rather than the visible fee, plus correspondent charges, compliance cost and the funding cost of pre-positioned balances.
What companies are involved in cross-border payments? Sending banks and payment institutions, FX providers, correspondent banks and SWIFT, cross-border networks and aggregators, local payout partners and wallet operators, and the compliance and reconciliation vendors that support them.
What is a payment corridor? A specific sending-country to receiving-country pair, with its own rails, partners, compliance rules, FX arrangements and settlement cycle. Operators build and run corridors one at a time.
Closing thought and further reading
A cross-border payment is six handoffs and two regulators. Each handoff is a place the money can stop, and each stop is somebody's job.
Building through similar complexity?
Discuss the operating decisions behind the essay, or explore where my experience can help.


