Why Cross-Border Payments Fail

14 out of every 100 cross-border payments get stuck on the first attempt, and most of these are due to failures of data and non-adherence to rules.

There is no universal standard to measure failure in cross-border payments, so all industry benchmarks are directional, not exact. This 14% number comes from LexisNexis's study on failed payments, and it gives us enough confidence to determine the direction and causes of failure. A vast majority of cross-border payments fail because of data and rule issues rather than any issue with funds or routing: a name mismatch, an incomplete address, a truncated field, and so on.

A cross-border payments transaction is a series of handovers between intermediaries. The sending platform receives the funds. The funds pass through a correspondent bank. They get converted to the foreign currency. From there, they reach the local payout rail and eventually land in the beneficiary bank. Each handover is a potential point of failure.

At each handover, the payment is required to carry certain information. This information is about who is sending it, who is receiving it, the address, the purpose, and so on. This information needs to be checked at every handover point, and while the money moves very securely, this information often gets lost in transmission.

Five reasons a payment fails

There are five main reasons why a payment can fail.

The first is wrong data. If the beneficiary name does not match the account, or an incorrect account number is entered, it causes the payment to be stopped.

The second is missing or truncated data. The required details may be incomplete at the time of sending, or they may be cut off during transmission. This happens because a correspondent bank or a rail may have a shorter field, or a different format, than the one the data came from.

The third is non-adherence to rules. Different regulatory environments have different requirements, such as the need for purpose codes and identifiers, different limits, the need for declarations, and so on. If a payment does not include what the receiving jurisdiction requires, it will get held or returned, even if the other details are correct.

The fourth is a hold because of compliance checks. A screening system can flag a payment when the data is inadequate and triggers a false match, or when the customer's funds flow is hidden in a way that the system reads as being obscured. These holds trigger a manual review by a human.

The fifth is due to operational issues. If the account is closed, there are insufficient funds, or cut-off times are missed, payments may be impacted. This, however, is a very small part of the total.

Looking at the overall picture, the first four are data and rules problems. Only the fifth is about money or the operational process. This also means the hypothesis that failed cross-border payments are unavoidable is an exaggeration. If a failure is caused by data and rules, it is avoidable.

Platforms fail more than banks

Payments platforms, compared with banks, typically have higher failure rates. The reasons for this are largely structural.

A payments platform is more likely to use a pooled account for funds transfers, which means the customer's flow is not visible to the bank. When the bank is unable to understand the transaction, it is more likely to flag it for investigation.

Second, a payments platform works with many bank partners and local rails. Each of them has its own data format and field lengths. All of these points of difference are places where a data issue can occur. Add to this that the onboarding form is designed to maximise conversion and reduce sign-up friction, not to capture all the identifiers needed for a clean cross-border payment. This optimisation for sign-ups means there will be higher failures downstream.

Banks have higher-quality payment data because regulations forced them to adapt years ago. Platforms, on the other hand, have been able to operate while capturing the minimum required during their sign-up flow.

What the data says

As per a 2026 study by Datos Insights, only about 40% of cross-border payments are processed straight through, which means 60% of payments need some manual intervention. The largest reason is incorrect beneficiary information, accounting for 21%. Account-number errors and incorrect bank details are second, at around 15% each. Another sharp finding in this study was that despite all the innovation, less than 2% of institutions process cross-border payments with near-full automation.

The global cost of this friction is estimated at around $118 billion a year. The caveat is that these benchmarks are not standardised and are estimates. If you run a payments platform, you don't run your business on industry estimates, but rather measure your own. However, the industry average is enough to confirm three things: most failures are due to data and rules, most of these are fixable, and nobody really adds up what they cost.

A payment failure has high costs. The obvious ones are a repair fee paid to the correspondent bank, the man-hours for manual investigations, and a delay in the customer receiving their money. Most firms don't add it up, because the costs are spread across multiple teams and no single owner.

We will try to add it up in the next piece.