Town & Country: The Address Standard Swift Cannot Postpone


On 25 March 2026, Swift told the market that no contingency was possible, because address data must be sourced at origin, and that "the November deadline must be adhered to". On 27 August, eleven weeks before the cutover, it deferred the requirement anyway. The rule would have stopped cross-border payment messages carrying fully unstructured postal addresses from 14 November 2026; Swift has instead promised an update on timing by December, and no date now stands. Yet the date was only ever a forcing function. The requirement beneath it comes from the revised Financial Action Task Force (FATF) Recommendation 16 (R16), which Swift's own March announcement cited in support of the change and which Swift cannot postpone. More awkwardly, the minimum Swift was about to enforce, a town and a country in dedicated fields, is precisely what R16 treats as the exception for the originator.
A single UK cross-border payment now answers to three address standards. Swift's network rule, whenever it arrives, requires town name and country in their own fields, and nothing more. The revised R16 asks more. As the FATF's draft implementation guidance of June 2026 explains, it expects the originator's standardised postal address, with town and country sufficient only in the absence of one. The beneficiary needs town and country alone.
UK law asks less than either. Under Article 4(1)(c) of Regulation (EU) 2015/847, the assimilated funds transfer regulation, the payer's address is one of four alternatives, alongside an official document number, a customer identification number, or date and place of birth. Article 4(2) requires no payee address at all.
The network minimum is the standard's exception
Swift's floor and the FATF's fallback look alike and do different work. Swift treats town and country as the floor of an acceptable message. The FATF treats them as a fallback for originators without a standardised postal address, and its guidance ties that concession to people living in rural or informal settlements where structured addresses are unavailable. A London customer with a full address on file, whose payment leaves carrying only London and GB, passes the network test. The same payment falls short of the standard.
Swift's format does not force that outcome. Its hybrid option keeps town and country in dedicated fields and allows two free-text lines for the rest of the address. Yet the network checks only the two fields. A full hybrid address and a bare town and country pass the same test, and nothing in the validation tells them apart.
The tooling built for the deadline reinforces the gap. Swift has released an AI address structuring model, free of charge, to help institutions convert legacy free text. Its output is limited to town and country predictions, and Swift states plainly that it is not designed to extract full postal addresses. That is a sensible design for clearing a network rule. It is the wrong measure of success for a compliance programme, because a remediation that ends where the model ends has proved only that two fields can be found. Whether the street and postcode travel, and in what form, is left untested.
Structure is also not accuracy. Swift's adoption figures carry their own caveat: they say nothing about the completeness of information beyond the mandatory elements. A wrong address parsed into the right fields is still wrong, and now looks authoritative.
Downstream cannot see the difference
The fallback creates a blind spot that only the sender can close. The FATF's draft guidance tells intermediaries that a payment using the permitted fallbacks, town and country in place of a full address among them, should be treated as compliant and complete. It goes further, strongly encouraging supervisors to discourage practices that undermine those simplified options.
That position protects financial inclusion, and rightly so. Its consequence is that no intermediary or beneficiary institution can tell, from the message alone, why the fallback was used. A customer with no standardised address and an ordering bank that truncated a full one look identical on the wire. The justification exists only in the ordering firm's customer records, if it exists at all.
The fallback also costs something beyond compliance. Discussing legal persons, the FATF's guidance says address information should identify the location clearly enough for effective sanctions screening and monitoring. A town and a country without a street or postcode do little to separate a customer from a designated person of the same name in the same city.
The evidential burden therefore sits at origin. The guidance lets ordering institutions rely on information that is consistent with customer due diligence, reasonably believed to be accurate and validly formatted, meaning prima facie capable of being a real address. A firm that sends the fallback for customers whose full addresses it holds will struggle to explain the choice. It will not be able to point to the network, because the network never asked the question.
The FATF is still testing how far address verification can go. Its consultation in June 2026 asked industry directly how feasible it is to verify an originator's address against reliable and independent sources, and by what means. That uncertainty argues for recording the method, not for skipping the step.
UK firms already police an optional field
Receiving firms appear to have expected more than UK law requires for some time. The Financial Conduct Authority (FCA) shows why in its Freedom of Information response of 21 September 2026 covers 365 notifications received since 2020 about payment service providers repeatedly failing to supply required information. Of those, 142 concerned a missing or incomplete payer or originator address. Once the FCA's categories are combined, address is the defect named most often, a finding OpusDatum noted in "Return to Sender".
The tension is in the law. Article 4(1)(c) lets a payer's provider send an identification number, a customer number or birth details instead of an address. The FCA's categories do not record whether such an alternative was present, so the figures are not evidence of breach, and the yearly mix swings too sharply to show a trend. They do suggest that the firms doing the reporting already expect an address, a position closer to the revised R16, or to their own policies, than to the regulation they notify under.
The law has not moved either
The UK text sits where it was. The official revised text, published by The National Archives, shows no change to Article 4 since the end of 2020 and no outstanding amendments to the regulation. The FATF expects full implementation of the revised R16 by the end of 2030. Asked in a parallel request in August 2026 whether it had considered amending, replacing or revoking the regulation since 2023, HM Treasury declined to answer on cost grounds.
The beneficiary side shows the distance most clearly. Swift's July figures recorded no postal address at all for almost one in seven beneficiaries on its network, alongside almost three in five originator addresses still unstructured. Under UK law the beneficiary gaps are lawful, since no payee address is required. Under the revised R16 they will not be, because town and country become mandatory for the beneficiary of any cross-border payment above the threshold.
UK firms therefore face a network date that has gone, a domestic law that permits an address to be omitted, and an international standard that expects a full one by 2030. Swift's deferral removed the only near-term date. It did not remove the destination.
Build to the standard, not the date
The deferral is an opportunity to correct the target rather than pause the programme. For originators whose full address is held, the outbound message should carry it, structured to the extent possible. Town and country should be reserved for customers who genuinely lack a standardised address, with the reason recorded at customer level so that the choice can be evidenced later. Parsing tools that stop at town and country should be treated as an aid to that work, not as its completion.
Receiving and intermediary firms face the opposite risk. The FATF's guidance treats a payment using a permitted fallback as complete and warns that systematically demanding more than R16 specifies is overcompliance. Procedures under Articles 8 and 12 of the regulation should therefore distinguish a permitted fallback from a genuine gap. Yet pattern still matters: a counterparty that consistently sends only the fallback for corporate originators with registered addresses is disclosing something about its data, whatever the network accepts.
Governance should follow the standard. Business cases that rested on avoiding Swift rejections will look weaker now that the rejections have been postponed. Programmes justified by R16 and by the quality of sanctions screening and monitoring will not, and boards should be shown that case before the budget is reallocated.
Conclusion: Two fields describe a place
Swift's extension has given the industry time, and time is easily mistaken for relief. The network rule asked for two fields, and its validation will check for nothing else. The revised standard asks for an address wherever one exists. A town and a country describe a place, not a person. Swift can postpone its rule; it cannot postpone the question a supervisor will eventually ask about whose address that was.
Do you know how many of your customers with a full address on file leave your systems as a town and a country?
If you cannot answer that with evidence, we can help you establish the position before the network rule returns; contact us.
At OpusDatum, we examine how payment data is captured at source and what a firm can show for each field it sends, including when and why a fallback was used. Our focus is the evidence behind the message rather than its format.


