Payment routing: which rail carries which transaction

Two identical invoices to the same supplier can settle a day apart. Nothing about the amounts changed, nobody made a mistake, and the difference was fixed before either payment moved. Payment routing is the part of the infrastructure that makes that choice, and most businesses never see it: they compare providers on price and on the currencies listed in a brochure, while the decision that determines settlement time is made from a different set of inputs entirely.

Payment routing is the choice of which rail carries a transaction and which institutions touch it on the way. It happens inside your provider, before anything leaves, and a routing decision fixes four things at once: how the beneficiary will be identified, how many institutions have to agree, when the money can arrive, and what happens if one field is wrong.

That last one is why routing is worth understanding rather than delegating blindly. A payment on a scheme with a single hop either lands or fails immediately with a reason. A payment on a correspondent chain can stall at the third institution days later, and the client and the supplier both learn about it from silence. We wrote about that failure mode in detail in straight-through processing and the repair queue; routing is the decision that exposes a payment to it or keeps it away.

A payment routing decision is narrower than it looks. Four things decide it, and price is not among them.

The currency being sent. A rail either carries a currency or it does not. There is no SEPA payment in Canadian dollars, and no ACH transfer in euros.

Where the beneficiary’s account sits. Not the country of the company — the country of the account. A Portuguese supplier invoicing from Lisbon may hold the account that receives the money somewhere else entirely, and the account decides the rail.

Whether the beneficiary can be reached on a domestic rail. Domestic schemes resolve the counterparty inside their own system: a transit and account number in Canada, a routing number in the United States, an IBAN inside the SEPA area. If the beneficiary is not reachable that way, the payment leaves the domestic system and the correspondent model takes over.

Value and timing. Batch rails release on a cut-off, so a payment that misses one waits for the next. Item-by-item rails do not have that property. For a supplier run this matters more than the headline speed of either rail.

Price follows from these four. It does not decide them, and a comparison that starts from a price list starts from the wrong end.

Swift carries any currency to any country, which is precisely why it relies on a chain of institutions rather than a single scheme. It is the rail that works when nothing more specific applies.

SEPA carries euro payments to accounts inside the SEPA area, identified by IBAN. Under the SEPA Credit Transfer rulebook, funds reach the beneficiary’s provider by the next business day at the latest. A euro payment to an account outside the area is not a SEPA payment and will not behave like one.

ACH carries domestic United States payments in US dollars, identified by routing and account number, released in batches under the NACHA rules.

Domestic Canadian rails carry Canadian dollars to counterparties reachable inside the Canadian system — which is a prerequisite with a calendar attached, as we covered in what supporting Interac for business actually requires.

Read that list again as a filter rather than a menu. For most payments, the four inputs leave exactly one sensible answer.

This is where payment routing stops being an abstraction. If a provider’s only access is Swift, then a euro payment to an account in Lisbon travels the correspondent model — not because that is the right rail for it, but because it is the available one. The payment picks up an intermediary institution, a currency conversion somewhere along the way, and each institution’s own cut-off time.

Nothing in that is visible when you compare providers. Both can say they send euros to Portugal. One does it on a scheme with a defined deadline; the other does it through a chain where no single participant is responsible for the end-to-end time.

The same asymmetry shows up domestically. A provider without Canadian access can still pay a Canadian supplier — by wire, into an account the supplier reaches through their own bank. It works, and it costs the supplier a reconciliation problem and you a slower cycle.

A non-bank payment service provider usually reaches a domestic scheme indirectly, through an institution that is a direct participant. That word matters less than most people assume, and more than the brochure suggests — it is the part of payment routing that never appears in a comparison table.

What indirect access does not change: the payment travels on the scheme, under the scheme’s rules, and settles the way scheme payments settle. What it does change: there is one more party whose operating window applies before the payment reaches the scheme. A cut-off at the sponsoring institution is earlier than the scheme’s own.

We route through Swift under our own Business Identifier Code, indirect SEPA for euro payments, ACH for domestic US dollar flows, and Interac for domestic Canadian transactions through a Canadian payment provider. Our settlement instructions are published in SwiftRef, which is what lets a sending institution route to us from its own reference data rather than asking. We are registered with FINTRAC as a money services business, registration M23481791, and with the Bank of Canada as a payment service provider under the Retail Payment Activities Act. Neither registration is a license, and we are not a bank in any jurisdiction.

A Canadian company owes a Portuguese supplier EUR 40,000.

Routed on Swift. The payment leaves in Canadian dollars or in euros, picks up a conversion at whichever institution performs it, and moves through the correspondent chain to the beneficiary’s provider. Three or four institutions see it. Each re-derives where to send it next from the message, applies its own screening, and works to its own cut-off. If the beneficiary’s address arrives as one unstructured string, an operator somewhere fixes it by hand.

Routed on SEPA. The conversion happens once, on your side, before the payment leaves. What travels is a euro payment to an IBAN, on a scheme with a next-business-day deadline and one institution between you and the supplier.

Both are legitimate payment routing outcomes, and the second is not always available. When it is, the difference is not a better price on the same product — it is a shorter chain with fewer places to stop.

You do not need access to your provider’s systems to see whether routing is working. Three signals are enough.

Payments to the same beneficiary that settle at different speeds mean the route is not stable. A corridor where currency conversion happens somewhere inside the chain rather than at a point you chose means you are paying an FX spread you never quoted. And a provider that cannot name the rail a given payment will travel is telling you it has one rail and routes everything through it.

Can I choose which rail a payment travels on?

Not directly, but payment routing works from inputs you control. The currency you send, the account details the beneficiary gives you and the provider you work with together leave one or two possible routes, and the rest follows. A supplier who can give you an IBAN and accept euros is a different payment from the same supplier invoicing in US dollars. Ask which rail applies before the first payment, not after a late one.

Why do two payments to the same supplier settle at different speeds?

Usually a cut-off rather than a different rail. Batch rails release on a schedule, so an instruction submitted after the last release of the day waits for the next one. On a correspondent chain every institution has its own window, and the payment sits at each of them until it opens. Identical instructions submitted an hour apart can land a day apart.

Is indirect access to a scheme worse than direct access?

Not in the way the phrase suggests. The payment travels on the scheme, under the scheme’s rules, and settles the way scheme payments settle. What indirect access adds is one more operating window in front of the scheme, and it closes earlier than the scheme’s own deadline. The number worth asking for is your provider’s cut-off, not the scheme’s.

Does a faster rail always cost more?

Price follows the payment routing decision rather than driving it, and the headline fee is not the whole cost. A correspondent chain can be quoted cheaply and still arrive short: an intermediary can deduct its own charge, and a conversion performed somewhere inside the chain is priced by whoever performs it. Compare what reached the beneficiary against what left you — that number contains everything.

What should I ask a provider about payment routing?

Three questions, asked per corridor rather than in general. Which rail carries a payment in this currency to this country? Where does the conversion happen, and who sets the rate? What is your cut-off for that rail? A provider with more than one rail answers all three in specifics; a provider with one rail answers in generalities.

Payment routing decides which rail carries a transaction, and it decides it from the currency, the location of the receiving account, the reachability of the beneficiary on a domestic rail, and the timing of the release. Providers differ less in the prices they quote than in how many of those answers they can act on.

Open a multi-currency business account with Montowire. We route through Swift under our own BIC, indirect SEPA, ACH, and Interac through a Canadian payment provider, across our payout network. Our team will map your corridors to the rails that carry them before your first payment goes out.