Why international payments still take days

An international payment takes days because it crosses systems that were never built as one. The instruction travels in seconds; the money moves in steps, across separate banks in separate time zones with separate rules, and every step has a clock, a checklist, and a funding requirement. Six causes account for almost all of the delay.

This is not a niche complaint. The G20 set targets for cross-border payments to become faster and cheaper by end-2027, and in October 2025 the Financial Stability Board concluded that meeting them at the global level is unlikely.

Cut-off times and time zones

Every bank in the chain has a cut-off: the time after which today’s payments become tomorrow’s. Cross a time zone or two and cut-offs compound. A payment leaving São Paulo after lunch has already missed the London afternoon; by the time it is processed in London, Singapore is asleep. BIS analysis of SWIFT gpi data identified banks’ offline hours as one of the strongest predictors of slow payments, because a payment that arrives outside business hours simply waits.

Weekends multiply the effect. Banks’ books and the domestic systems they settle through close, and because holidays differ by jurisdiction, a three-country chain can lose three separate days to calendars in a single week.

The correspondent chain

Most cross-border payments travel through correspondent banking: a sequence of banks holding accounts with one another, each moving the payment one leg further. The journey, message by message and account by account, is mapped in how SWIFT works.

The surprise in the data is that the chains are short: the BIS found cross-border payments involve just over one intermediary on average, and payments with three or more intermediaries are under 1% of volume. The delay is not mainly the number of hops. It is what happens inside each hop: validation, screening, funding checks, batch cycles, and the wait for the next institution’s business day.

The beneficiary leg

The same BIS study, built on roughly 20 million payments across 141 countries, found the biggest single block of time at the very end of the journey. The beneficiary leg, from the last intermediary paying the beneficiary’s bank to the customer actually being credited, absorbs almost eight hours on average, which SWIFT itself flags as the industry’s priority for getting faster.

The reasons live locally: the beneficiary bank credits through domestic rails with their own batch windows, applies its own compliance checks, converts currency where needed, and does all of it during its own business hours. Capital controls and reporting requirements in the destination country add stops that no upstream bank can remove.

Compliance stops

Every institution in the chain carries its own regulatory obligations, so sanctions and AML screening run again at every hop. Each screening pass can throw a false positive, and a false positive parks the payment in a manual review queue until an analyst clears it. The historical driver was data quality: legacy message formats truncated names and addresses, making innocent payments look suspicious. The move to ISO 20022, completed for cross-border payments in November 2025, attacks exactly this, carrying structured and complete party data. It reduces the number of stops. It does not change the fact that the checking is duplicated at every institution, because each one is separately accountable.

Funding and liquidity

A correspondent pays out against the account its client bank holds with it, which means that account must be pre-funded. If the balance is short, the payment waits for funding, and funding waits for the same cut-offs and business hours as everything else. This is why banks park working capital in nostro accounts around the world: capital trapped in transit, priced against every payment that needs it. Where balances and hours do not line up, the payment queue is the shock absorber.

Batch processing

Underneath the chain, much of the machinery still runs in batches. Domestic ACH-style systems run on deferred net settlement, settling at scheduled windows rather than continuously, and banks’ internal processing often cycles overnight. A payment that arrives between windows waits for the next one, and the clearing-versus-settlement gap inside each system adds its own lag (the mechanics are laid out in clearing vs settlement).

What fast looks like today, and what it costs

None of this means every payment is slow. By SWIFT’s gpi figures, nearly 60% of payments reach the end beneficiary within 30 minutes and almost 100% within 24 hours. The averages hide the tail: the payments that miss a cut-off, hit a screening queue, land on a weekend, or terminate in a slow beneficiary market are the ones that take days, and businesses cannot predict in advance which payments those will be.

Speed is also only half the bill. The World Bank still measures the global average cost of sending remittances at 6.36%, and the corporate version of that cost, stacked intermediary and FX fees you can’t see, arrives as a payment that lands short of the invoiced amount.

Where Frame fits

Frame is the settlement layer for global finance: one integration to orchestrate payments at scale across fiat rails, stablecoins, and tokenized deposits, with compliance enforced on every transaction and settlement in seconds instead of days.

Each cause on this page is a property of settlement running across disconnected systems. Frame’s answer is to make settlement itself programmable: a payment enters through one integration, Frame routes it across whichever rail fits the corridor and the policy that governs it, and Frame’s Rules Engine enforces those policies inside settlement, so the checking happens once, as a condition of the transfer, rather than repeatedly around it. Every settled transaction produces verifiable evidence that its conditions were met, without exposing the underlying business data.

See how a rail-neutral settlement layer works: The Frame Blueprint.