Your payment reached the bank an hour ago. It is not in the account.

A client in Rotterdam rang her bank on a Thursday to ask where fifty thousand euros had gone. The bank told her, correctly, that the payment had been sent and confirmed as received. The supplier told her, also correctly, that nothing had arrived.

Both were right.

That is the whole problem in four words.

The money had reached the receiving bank. It had not reached the supplier's account. Those are different events, on different clocks, run by different institutions and reported separately.

Almost nobody outside the industry knows there are two numbers.

The two clocks

Swift publishes how fast payments cross its network, and the headline is genuinely good: “90% of cross-border payments sent over the Swift network reach the destination bank within an hour”. That is well ahead of what the G20 asked for.

Then the same release adds the number that matters to the person waiting for the money: “43% of all cross-border payments over Swift reaching the end customer's account within an hour”.

Ninety, then forty three.

The gap is not the network. It did its job in under an hour, nine times out of ten. What follows is the receiving country, where the transfer stops being international and becomes domestic, and gets treated like any other domestic credit.

The regulator counts it differently again

There is a third number, and it comes from the people setting the target.

The Financial Stability Board tracks progress against the G20 goal, which is stated plainly: “75% of cross-border wholesale payments to be credited within one hour of payment initiation” by the end of 2027. On its measure, the share credited within an hour was 50.6% in 2024, with 92% inside one business day.

That is lower than Swift's 90% and higher than Swift's 43%, because it counts a different population of payments in a different way.

The FSB is also honest about why its own number moved: the speed “decreased marginally due to technical factors unrelated to underlying settlement times”, one of which was a change in how Swift defines a non-business day. Nothing got slower. The ruler changed.

I find that reassuring rather than annoying, and it is worth saying why. An industry that quietly restates its own figures is one thing. An industry that publishes the restatement, names the cause and shows the effect on the headline is doing something harder and more useful.

Where your corridor sits

Averages hide the thing you actually need, which is your particular pair of countries.

Swift names the movers directly: “92% of payments to the Eurozone settling within an hour (+3%)”, with the Middle East at 88% and Africa at 87%. Those are the regions that improved. The variation between destinations is not random and it is not mainly about technology: it tracks how quickly the receiving country's banks and regulators let money through the last mile.

Inside the euro area there is a further layer. Instant euro transfers now run at about 30.7% of all euro credit transfer volume, and nearly every eurozone provider is signed up to the scheme. So the domestic leg that costs a day elsewhere can be ten seconds here, which is why a euro corridor often feels like a different product rather than a faster version of the same one.

Why the last mile is slow

Several things happen after the money lands, and none of them are visible from your side.

The receiving bank runs its own checks. Sanctions screening, fraud rules, sometimes a regulatory report before a foreign credit is released.

It was checked on the way out. It gets checked again on the way in, by an institution you have no contract with.

Then the cut-off. Domestic clearing runs to a timetable, and money arriving after it waits for the next one. An hour of network speed does nothing about a batch that runs at nine tomorrow.

And then there is the account itself: currency conversion, an internal hold, sometimes a manual release for amounts above a threshold nobody publishes.

Swift's own chief business officer put the industry's task in one sentence, and it is the sentence to quote at anyone who tells you the problem is solved: “we all - the industry and global policymakers - need to focus our efforts firmly on the domestic beneficiary leg of transactions”.

Reading a tracking status honestly

Tracking exists and it tells the truth.

It answers a question you did not ask.

A status of completed on the sending side means the instruction left and was accepted. It does not mean credited. Swift is precise about which leg its headline covers, describing payments that “reach the destination bank within an hour”, and that phrasing is doing exact work: destination bank, not destination account.

So when a status looks final and the supplier says nothing arrived, both statements can be true at once, and arguing about which one is wrong wastes the day you could have spent asking the receiving bank.

Ask for the reference. Pass it to the beneficiary. Let them put it to their own bank, because that bank can see its own queue and yours cannot.

Why the targets exist at all

This whole measurement effort is recent, and it exists because nobody could previously answer a simple question about how long money takes.

The G20 set targets covering speed, cost, access and transparency, and the Financial Stability Board reports against them every year. The wholesale speed target reads: “75% of cross-border wholesale payments to be credited within one hour of payment initiation”, with the remainder inside one business day, by the end of 2027.

The 2024 report put the one-hour share at 50.6% and the one-business-day share at 92%, noting the fall came as the “share of payments over Swift crediting funds within one hour and one business day decreased to 50.6% (-3.2 pp compared to 2023) and 92% (-0.7 pp)”. Two years of runway remain on the first number. The second is close to done.

Read alongside Swift's 90% to the receiving bank, the picture is consistent: the international part of the journey is largely fixed, and the domestic part is where the target will be won or lost. That is also why the FSB writes about the beneficiary leg rather than about networks.

What I could not find out

I wanted to give you a table of last-mile times by country.

It does not exist in public at that resolution, and I spent most of a week finding that out rather than finding the table.

Swift publishes regional aggregates and says it covers the top forty receiving countries. The per-country breakdown behind that is not published in a form I can cite, and I am not going to reconstruct it from press releases and call it data.

I also cannot tell you the threshold at which a given bank holds a payment for manual review. I have asked three banks. Two did not answer and one said it was not something they disclose, which is a fair answer to a question that also has an obvious reason behind it.

What the wait actually costs

A delay is not a nuisance. It is a price, payable by whoever holds the risk while the money is in transit.

Say a shipment worth fifty thousand euros waits three days for credit. The supplier holds the goods. The vessel sails without them. The next sailing on that route is a week later.

The payment was never lost. Ten days of stock were.

That is the ordinary case. It also explains why exporters price certainty above speed: a corridor that always takes two days can be promised to a customer, and one that usually takes an hour and occasionally four days cannot.

The G20 target is written in those terms rather than in averages: 75% credited within an hour, and the rest within one business day. The second half of that sentence is the useful half. It is a promise about the worst case, not about the typical one.

The euro leg is a different animal

Inside the euro area the last mile has largely been solved, and the numbers say so.

By October 2025 there were 2,792 providers signed up to the instant euro scheme. The scheme's own status note records “a share of 79 percent of all SCT adherents”, rising to 92% inside the euro area, with instant transfers running at “close to 30,7 percent of the total credit transfer volume”.

A payment into a German account and one landing outside the scheme are not fast and slow versions of the same thing. They are two products.

Only one can promise a supplier anything about Friday afternoon.

This is worth knowing when you choose where an intermediary sits. Routing a payment so that the final leg lands inside the instant scheme is often a bigger win than shaving a fee, and it is a decision made once rather than negotiated every shipment.

What I got wrong for a year

I had assumed the delay was correspondent banking. A chain of intermediaries, each adding a day.

It is a satisfying explanation. It fits the fee deductions people see.

The published numbers do not support it. Ninety per cent cross the whole network, correspondents included, inside an hour.

The chain is not what makes it slow. The last institution is.

That changed what I tell people to do. Chasing your own bank harder achieves very little when your own bank finished its part before you noticed there was a problem.

How to plan around it

Treat it as two legs with separate promises. Your bank owns the first and can be held to an hour.

Nobody you have a contract with owns the second.

Ask for the tracking reference on every payment above whatever amount matters to you, and ask specifically whether the status means received by the bank or credited to the account. If your provider cannot tell the two apart, that is itself information about the provider.

Build the last mile into the terms rather than into your nerves. A supplier who ships on confirmation of credit and a supplier who ships on proof of sending are asking for different things, and the second one is asking you to carry the last mile as a risk.

And when it goes wrong, ask the receiving side first. The money is usually sitting in a queue at the end you cannot see, which is exactly where nobody thinks to look on the first day.