When SWIFT launched the global payments innovation (gpi) initiative in 2017, the cross-border payments industry was overdue for structural improvement. Correspondent banking was opaque, slow, and difficult to audit. A corporate treasurer initiating a payment from New York to Singapore could wait three days for confirmation and receive no meaningful status information in between. gpi addressed that problem directly, and the improvements were real. But gpi also has well-defined limits that become important when treasury teams are evaluating whether the correspondent model can meet their operational requirements or whether alternative rails are worth the migration cost.
What gpi Actually Changed
The core innovation in SWIFT gpi is the Unique End-to-End Transaction Reference (UETR), a 36-character identifier that travels with a payment through every correspondent in the chain. Before UETR, each bank in a correspondent chain assigned its own internal reference, and there was no standardized way for the originating bank to query status from an intermediary three or four hops downstream. A treasurer asking "where is my payment" was effectively asking a bank to chase its correspondent manually, which might take 24 to 48 hours to produce a useful answer.
UETR changed that by creating a common tracking thread. The SWIFT gpi Tracker monitors the progress of gpi-enabled payments and provides status updates visible to banks and, via bank API integrations, to corporate treasury systems. Payments in major corridors between gpi member banks now achieve same-day credit in a meaningful percentage of cases. The tracker shows whether a payment is in processing, credited, or has returned, and provides timestamps at each correspondent hop.
That is a genuine step forward. For treasury teams that previously had no status visibility between instruction and final confirmation, gpi tracking converts a process that felt like sending a payment into a void into something that resembles a trackable parcel. The audit value alone is significant: gpi-tracked payments produce a structured record of the correspondent chain and timing that supports treasury audit trails and exception investigation.
The Correspondent Chain Problem Remains
What gpi did not change is the fundamental architecture of correspondent banking. A payment from a mid-sized U.S. company to a supplier in a less-common corridor, say Colombian pesos or Indonesian rupiah, still travels through a chain of correspondents. gpi adds a tracking layer on top of that chain. It does not reduce the number of hops, change the underlying bilateral agreements between correspondent banks, or modify the fee deduction behavior of intermediary banks.
Fee deduction is a persistent friction point for corporate treasury. SWIFT gpi introduced a commitment by members to send payment amounts in full rather than deducting their fees from the principal. The SHA/OUR instruction model that has always existed in SWIFT messaging applies here: under OUR, the sender pays all fees; under SHA, each correspondent deducts from the principal. In practice, compliance with gpi's full-amount commitment has not been universal across all corridors and all members. Treasurers receiving beneficiary complaints about short-credited payments may be encountering either non-gpi intermediaries in the chain or inconsistent OUR enforcement. gpi improved the norm without providing technical enforcement.
Pre-Funding and Liquidity Position
gpi is a messaging and tracking standard. It operates within the existing correspondent banking liquidity model, which requires banks to maintain pre-funded nostro positions in destination currencies. Those nostro balances are what enable same-day or next-day credit at the receiving end. When a corridor has adequate liquidity pre-positioned, gpi tracking shows fast movement. When the corridor has thin liquidity, gpi tracking shows the reality of a payment sitting in queue, which is more transparent than before but does not solve the underlying liquidity constraint.
For corporate treasurers, this means that a gpi-tracked payment is not a guaranteed fast payment. The tracker will tell you that your payment has been received by the next correspondent. It will not resolve the situation where that correspondent does not have sufficient local currency to credit the beneficiary and is waiting for its own liquidity cycle to replenish. In high-volume corridors between major financial centres, this is rarely an issue. In emerging market corridors or corridors with fewer active correspondents, pre-funding constraints surface in gpi tracker data as unexplained queue delays.
Opacity Within Each Hop
gpi gives you timestamps at each hop. It does not give you the reason for delays within a hop. If a payment sits at an intermediary for four hours before forwarding, the tracker shows that it arrived at that bank and later departed. It does not show whether the hold was a routine processing queue, a sanctions screening match that required manual review, a compliance query triggered by a payment detail flag, or a system connectivity issue. That information remains inside the correspondent and is retrievable only through manual correspondent inquiry.
This matters operationally when a payment is approaching a cut-off deadline or when a beneficiary is reporting that funds have not arrived. The treasurer knows the payment is somewhere in the chain, but diagnosing why it has stalled still requires correspondent bank communication, which is exactly the slow manual process that gpi was meant to improve. The tracker narrows the problem to a specific correspondent. Resolving it still depends on that correspondent's responsiveness.
gpi vs. Alternative Settlement Rails
It is worth being clear about what gpi is and what it is not. gpi is an improvement layer on the correspondent banking network, not a replacement for it. Alternative settlement approaches, including local real-time gross settlement (RTGS) networks, regional payment schemes, and direct connectivity to local clearing systems, address different parts of the problem. A payment settled directly through Brazil's PIX system or India's RTGS avoids the correspondent chain entirely for the local leg. gpi's tracking improvements are not relevant because there is no multi-hop correspondent chain to track.
The choice between gpi-routed correspondent banking and direct settlement infrastructure is not a choice between old and new. It is a choice between breadth and depth. Correspondent banking with gpi covers a large number of currencies and corridors through a single network. Direct settlement infrastructure in specific markets offers lower latency, more predictable fee structures, and richer status data for those corridors, at the cost of requiring separate connectivity and compliance infrastructure per market.
At Birch Hill, we started from the premise that the corridors that matter most to a treasury team need to perform reliably and produce complete audit records regardless of the underlying rail. For some corridors, gpi correspondent routing is the right path. For others, direct local settlement infrastructure gives the treasury team the confirmation certainty and fee predictability that gpi alone cannot guarantee. The goal is a routing layer that picks the appropriate rail per transaction, not a mandate to use one network for everything.
What This Means for Treasury Policy
If your treasury policy currently equates "we use SWIFT gpi" with "our cross-border payments are under control," that policy deserves a closer look. gpi is a meaningful improvement in payment tracking and, in many corridors, in speed. It does not eliminate the need to monitor pre-funding adequacy in destination currencies, manage correspondent relationships for exception resolution, or evaluate whether specific corridors would be better served by local settlement infrastructure. Those decisions require data on corridor-level performance, fee deduction rates, and average time-to-credit confirmation, not just the fact of gpi enrollment.