Faster Payments

37.0 What this chapter gives you#

1. You will be able to explain why a payment that reaches the plumber’s account in one second does not move a penny between the two banks until Monday.
2. You will be able to say what deferred net settlement means, and why the Faster Payment System sits in exactly the same category as Bacs.
3. You will be able to work out why a payment made on Good Friday is spendable that afternoon and is not settled between banks until the following Tuesday.
4. You will be able to explain why the £1 million scheme cap is almost never the limit a customer actually meets, and name what binds instead.
5. You will be able to describe the three participation models and say which one requires a settlement account at the Bank of England.
6. You will be able to explain what prefunding and the Net Sender Cap are for, and why the July 2026 flexible model lowers the cost of entry for non-banks.
7. You will be able to distinguish the four transaction types and say why standing orders break the assumption that everything runs 24/7.
8. You will be able to state precisely what “irrevocable” means here, and which remedies still exist after a payment has gone.
9. You will be able to explain why open banking is a way of instructing a rail rather than a rail in its own right.
10. You will be able to size the netting efficiency — roughly six to nine pounds of customer payments discharged for every pound of central bank money that moves.

At 11:47 on a Sunday night in July, a man in Leeds opens his banking app, taps in a sort code, an account number and £340, and pays a plumber who fixed his boiler that afternoon. The plumber’s phone buzzes before the man has locked his screen. The money is there. She can spend it.

Nothing about that sentence would have been possible in Britain in 2007. Before May 2008, a payment instructed by telephone or internet banking went into the Bacs cycle and arrived three working days later. A Sunday-night instruction would have reached the plumber on Wednesday. There was no Sunday, because Bacs does not run at the weekend, so in practice it would have been Thursday. The plumber would have gone to bed on Sunday with an invoice and nothing else, and would have chased the man on Tuesday to ask whether he had paid.

The Faster Payment System changed that, and it changed it more thoroughly than most people realise. It did not merely speed up an existing thing. It created a new expectation — that a payment between two ordinary British bank accounts is a real-time event, available at any hour, on any day, with no batch, no cut-off and no working-day arithmetic — and that expectation has since spread underneath a great deal of British commercial life. Payroll advances, gig-economy earnings, marketplace payouts, insurance settlements, refunds, the entire account-to-account payments industry and every open banking payment initiation in the country all sit on top of it.

And almost none of it is true in the way it appears. The plumber’s £340 did not travel from one bank to another on Sunday night. Nothing travelled. What travelled was a message, and the plumber’s bank paid her out of its own money on the strength of it. The actual money between the two institutions moved during Monday, in one of three settlement cycles, netted against every other payment those institutions had exchanged. The seconds you can see and the settlement you cannot are two different systems bolted together, and the gap between them is where the risk, the liquidity cost and most of the interesting engineering live.

This chapter explains both halves: the fast one you experience, and the slow one underneath it that makes the fast one affordable.

The plain version#

Imagine two villages either side of a wide river. There is one bridge, and it is a long way round.

Each village has a post office that looks after everybody’s money. If Mrs Ali in the north village wants to send £180 to Mr Grant in the south village, the old way was simple and slow: the north postmaster counts £180 out of his drawer, puts it in a locked box, and waits for the van. The van crosses the bridge once a day. Mr Grant gets his money the following afternoon, or the one after that if the van has already gone.

Now somebody installs a telephone line between the two post offices.

The north postmaster picks up the phone and says: “Mrs Ali is sending £180 to Mr Grant. Pay him out of your drawer now. I’ll square it with you when the van comes.”

The south postmaster puts down the phone, opens his own drawer, and hands Mr Grant £180 of his own cash. Mr Grant has his money in under a minute. Nothing has crossed the bridge. There is no van in this story at all yet.

That is the whole trick, and it is worth sitting with for a moment because everything else follows from it. The speed does not come from moving money quickly. It comes from not moving money at all and having somebody at the far end pay out on a promise. The message is fast because messages are cheap and light. The money is slow because money is heavy, and the clever part of the design is realising you do not have to move the money in order to make the payment happen.

What the van still does#

The van has not been abolished. It runs three times during each working day, and it carries the difference.

Suppose that during Monday morning the north post office phones the south fourteen times, for payments totalling £8,420, and the south phones the north nine times, for payments totalling £7,955. At lunchtime the two postmasters compare notes. North owes south £8,420; south owes north £7,955. Subtract one from the other and north owes south £465.

The van crosses the bridge carrying £465. Twenty-three payments worth £16,375 have been settled by moving four hundred and sixty-five pounds.

This is why the arrangement is cheap enough to give away free. If the van had to carry every payment separately it would need to cross twenty-three times with sixteen thousand pounds aboard, and somebody would have to pay for that. Instead it crosses three times a day with the leftovers.

What the south postmaster is worried about#

Here is the part a bright twelve-year-old spots straight away, usually before an adult does.

The south postmaster handed Mr Grant £180 of his own cash on the strength of a phone call. What if, between the phone call and the van, the north post office burns down? Or the north postmaster turns out to have been lying, or has run out of money, or has simply disappeared? The south postmaster has given away real cash in exchange for a sentence spoken down a telephone line.

That worry is the single most important fact about this system, and it is invisible to everybody using it.

The way it is dealt with is not trust. It is a locked box. Before the north post office is allowed to use the telephone at all, it must take a quantity of its own cash to the county courthouse — a building neither village controls, which will not lose the money and will not go bust — and lock it in a box there. The box is not the north post office’s spending money. It sits untouched. Its only job is to be there if the north post office fails, so the south postmaster can be paid from it and nobody who took money out of a drawer that morning is left short.

The size of the box is set by a rule: the north post office may never, between one van and the next, run up a debt to the other village larger than what is in the box. Once it has phoned through enough payments to reach that figure, the telephone stops working for outgoing calls until the van has been and the slate has been cleared.

So the picture is: a telephone that is always on, a van that runs three times a working day, and a locked box at the courthouse that makes the telephone safe.

The rules on the telephone#

There are three more rules, and each of them has a reason.

The first is that there is a ceiling. The south postmaster will accept a phone call for anything up to one million pounds. Above that, the money goes by armoured van, immediately, with two guards and a signature — that is a different service, and it costs money, and it is what people buying houses use. The telephone is for ordinary amounts, and one million pounds is where “ordinary” is judged to stop.

The second is that each post office may set a lower ceiling of its own for its own customers, and they all do. The village rule says a million. Your own postmaster may say that you personally may phone through twenty-five thousand pounds and no more, because he has known you for two years and that is as far as his nerve goes. Both rules apply, and the lower one is the one that stops you.

The third is that the telephone is on all the time, and the van is not. The telephone works at three in the morning on Christmas Day. The van runs on working days only. That is fine, because of the locked box — but it means that over a long bank holiday weekend the north post office can be phoning payments across for four days before a single van crosses the bridge.

The worked example#

Take the man in Leeds and the plumber.

He instructs £340 at 11:47pm on Sunday 12 July 2026. His bank checks he has the money, checks the payment against its fraud rules, and sends a message to the central system. The central system routes it to the plumber’s bank, which decides in under a second whether it can accept the payment, and answers. The whole round trip — sending bank to central system to receiving bank and back — is designed to complete inside fifteen seconds, and normally takes a small fraction of that. The plumber’s bank credits her account immediately, out of its own funds. She can spend the £340 at 11:48pm on Sunday night.

The two banks square up on Monday. Three times during Monday’s working hours, every bank in the system compares what it owes with what it is owed and one number per bank moves at the Bank of England. The man’s £340 is a rounding error inside that number. Neither he nor the plumber will ever know which of Monday’s three cycles carried it, and neither of them needs to.

Now scale it up. In 2025 the Faster Payment System carried 5.55 billion payments worth £4.84 trillion. That is roughly fifteen million payments a day, averaging about £870 each. Yet the money that actually had to move between the banks to settle all of it averaged around £2.3 billion a day in the first half of 2026. Somewhere between six and nine pounds of customer payments are being discharged for every pound of real money that moves. The rest cancels out — north owing south, south owing north, and the subtraction doing the work.

You should now be able to say the following at dinner and be right. When you send money by bank transfer in Britain, nothing is transferred. A message goes across in about a second, the receiving bank pays its customer out of its own pocket, and the banks square up with each other three times on each working day by moving only the difference. It feels instant because it is instant for you. Underneath, it is a promise with a locked box behind it.

Where the plain version stops being true#

The village is a good picture of the mechanism and a misleading picture of the risk, the availability and the guarantees. Four corrections.

The first correction is that “instant” describes the customer experience, and only the customer experience. No money moves between institutions in seconds. Not some of it, not most of it — none. The Faster Payment System is a deferred net settlement system, which is precisely the same category as Bacs, and the only reason it does not feel like Bacs is that the receiving institution is contractually obliged to credit its customer immediately rather than waiting for settlement. The interbank leg happens three times each business day in the Bank of England’s real-time gross settlement system, and the Bank’s own description of the lag is worth quoting because it is more candid than anything the industry says in public: the gap between a customer payment and its settlement “could be several hours for Faster Payments during the week”, but over an Easter weekend “the gap between settlements is currently from between Thursday evening and Tuesday morning”. Read that again. A payment made on Good Friday is spendable on Good Friday and is not settled between banks until the following Tuesday. For four days, one institution has paid out real money and holds nothing but an obligation. That is not a defect and it is not a secret; it is the design, and the prefunded cash at the Bank of England exists precisely because of it. But anybody who tells you Faster Payments “settles in real time” has confused two layers of the system, and if they are designing a treasury function on that belief they will get their liquidity model wrong.

The second correction is that “24/7” is true of the service and false of the settlement infrastructure. The Faster Payment System is available twenty-four hours a day, every day of the year. The Bank of England’s RTGS system is not. As of the time of writing, RTGS operates from 06:00 to 18:00, and the Bank has announced an early-morning extension to 01:30 from September 2027; a move towards near round-the-clock settlement, including a possible additional weekend settlement day, is at the consultation stage, with the consultation paper published on 18 May 2026. So the always-on property that everybody markets belongs to the messaging layer. Beneath it sits a settlement layer that keeps banking hours, does not work weekends, and is only now being asked whether it might. The mismatch between the two is not a temporary lag in modernisation — it is the reason the prefunding regime exists, and any change to settlement hours changes the economics of prefunding directly.

The third correction is that the £1 million cap is almost never the limit you actually hit. One million pounds is the central scheme ceiling per transaction, and it has been since 10 February 2022, when it was raised from £250,000. Every participating institution sets its own lower limits, and those limits vary enormously by institution, by channel and by whether the account is personal or business. On Pay.UK’s own published comparison, as last updated in July 2026, HSBC’s online personal limit was £25,000 against a business limit of £1,000,000; Lloyds was £25,000 personal and £99,999 business; Barclays £50,000 personal; NatWest £100,000 personal; Metro Bank £21,000 personal. Starling, Monzo and several others publish the full £1,000,000 for both, subject to their own additional checks. So “Faster Payments can carry a million pounds” is a statement about the scheme, not about your account, and in most retail cases the binding constraint is two orders of magnitude below the headline. The cap is also per transaction rather than per day, and institutions layer daily aggregate limits, new-payee cooling-off periods and value-triggered verification on top of it. None of that is in the rulebook figure.

The fourth correction is that “irrevocable” and “guaranteed” are being used to mean four different things, and none of them is quite what the customer thinks. It is true that once a Faster Payment has been sent, the payer cannot cancel it — Pay.UK states this plainly, and it follows from the real-time design. But that is not the same as saying nothing can be done. Where a payment has gone to the wrong account by mistake, the industry operates a Credit Payment Recovery process. Where it has gone to the wrong account because the payer was defrauded, a mandatory reimbursement regime has applied since 7 October 2024. And “instant” itself carries an exception: the scheme’s own published expectation is that funds are usually available almost immediately but can take up to two hours, and a very small proportion of UK accounts — Pay.UK puts it at under 0.1%, including some savings accounts — do not accept Faster Payments at all. A receiving institution that cannot post immediately is expected to tell the customer when the funds will arrive, not to fail the payment. So the honest formulation is: unstoppable by the sender, usually instant, occasionally two hours, sometimes recoverable afterwards, and increasingly reimbursable if it was a scam.

The technical version#

What the system is, and who runs what#

The Faster Payment System is one of the retail interbank payment systems operated by Pay.UK, the recognised operator and standards body for UK interbank retail payments. Pay.UK owns the scheme: the rules, the participation criteria, the standards and the liability arrangements. It does not build or run the machines. The central infrastructure is supplied and operated under contract by Vocalink, a Mastercard company, which also provides the central infrastructure for Bacs and for the Image Clearing System. On 17 December 2025 Pay.UK announced multi-year extensions to all three of those infrastructure contracts, providing continuity into the early 2030s.

Settlement is provided by the Bank of England. The Bank is not the operator of Faster Payments and has no role in the scheme rules; it is the settlement agent, holding the accounts in which the interbank obligations are finally discharged in central bank money.

The system launched on 27 May 2008, following a testing period in which very small-value transactions were processed. Its origins lie in the Office of Fair Trading’s Payment Systems Task Force, established in March 2004, which by May 2005 had secured an industry agreement to reduce clearing times for telephone, internet and standing order payments. Vocalink was awarded the infrastructure contract in October 2005 and has operated it ever since. Ten institutions went live at launch: Abbey (now Santander UK), Barclays, HSBC, Lloyds TSB, The Royal Bank of Scotland, Citi, Bank of Scotland, Clydesdale Bank, The Co-operative Bank and Nationwide Building Society.

The programme intended to replace the underlying infrastructure — the New Payments Architecture — has been through a substantial reset. Following the cancellation of the NPA procurement and the government’s National Payments Vision, the work was renamed Interbank Infrastructure Renewal. The Payment Systems Regulator revoked Specific Direction 3 in May 2025 and Specific Direction 2 in August 2025, and closed its NPA project page on 14 October 2025. The stated long-term destination remains a single central infrastructure clearing and settling Bacs, cheques and Faster Payments, rather than three separate ones. The Vocalink contract extensions announced in December 2025 should be read against that: the existing infrastructure is now expected to carry the traffic into the early 2030s.

Scale, as reported by Pay.UK: 5.55 billion transactions worth £4.84 trillion during 2025. For comparison, the system handled 967.6 million transactions worth £771.4 billion in 2013.

The four transaction types#

Pay.UK defines four categories of Faster Payment. They differ in initiation channel, availability and timing, not in the underlying clearing mechanism.

Several points in that table repay attention.

The fifteen-second figure is a property of the central infrastructure, not of the end-to-end customer experience. It is the window within which the sending PSP receives confirmation that the receiving PSP has accepted or rejected the payment, and where the answer is a rejection, the receiving PSP supplies a reason. This is the mechanism that makes the customer-facing “sent” and “received” states meaningful. The two-hour outer bound covers the cases where the receiving institution cannot post immediately — an operational problem at its end, or an account type that does not permit immediate posting — and in those cases the scheme expects the customer to be given an indication of when the funds will arrive rather than a bare failure.

Standing orders are the outlier and are frequently misunderstood by engineers who assume the 24/7 property extends to everything. It does not. Standing orders execute on working days only, roll forward when the nominal date is not a working day, and cluster heavily in the early morning: Pay.UK’s published expectation is that at least 90% are sent by 6am. If you are building a system whose behaviour depends on when a standing order will land, that early-morning clustering and the working-day roll are the two facts that matter.

Direct Corporate Access is the least discussed of the four and the most relevant to anyone with a bulk payment file. It allows a business customer to submit files of payment instructions directly into the Faster Payment System, using a secure-IP solution comparable to Bacstel-IP and, notably, the same standard file format as Bacs transactions. That last detail is the practical bridge for organisations that already produce Standard 18 files for Bacs: the file format does not have to change to move the same payment run onto a same-day rail. Not every participant offers Direct Corporate Access, and those that do not typically offer an alternative bulk mechanism of their own. A number of software providers are accredited to supply Direct Corporate Access services.

The value cap and its history#

The central scheme limit is £1,000,000 per transaction. It has stood at that level since 10 February 2022, when Pay.UK raised it from £250,000; at the time of the increase there were 38 directly connected payment service providers and around 400 PSPs connected indirectly through sponsors. Earlier in the system’s life the ceiling was substantially lower, having been raised to £100,000 for all payment types on 6 September 2010.

Institution-set limits are what bind in practice. Pay.UK maintains a published comparison of the limits set by individual banks and building societies; the figures below are drawn from that page as last updated on 24 July 2026, and are accurate at time of writing. They will change.

Monzo’s entry carries a rider worth generalising: payments above £10,000 are subject to further checks. Value-triggered friction of that kind is now routine across the industry and is not visible in any published limit table. A £1,000,000 headline limit and a £10,000 verification threshold coexist comfortably, and the second is what a customer will actually encounter.

Access: three participation models#

Pay.UK defines three ways to participate, and the distinction between them is the most commercially consequential thing in this chapter for anybody building a payments business.

A Directly Connected Settling Participant holds its own technical connection to the central infrastructure and its own settlement account at the Bank of England, and performs its own settlement. It sets its own transaction limits. This is the full-fat model, appropriate to institutions with high volumes and eligibility for a Bank of England settlement account.

A Directly Connected Non-Settling Participant holds its own technical connection to the central infrastructure but does not settle at the Bank of England. A sponsoring participant performs the Bank of England settlement on its behalf. This model exists precisely for organisations that want the operational control and latency of a direct connection but are not eligible for, or do not want, a settlement account.

An Indirect PSP has neither its own connection nor its own settlement. It sends and receives Faster Payments through a sponsoring participant, which also performs the Bank of England settlement. The sponsor allocates the indirect participant a sort code from its own range and registers it with the payment systems and with the industry reference data. The indirect participant still processes its own payment messages, handles its own exception flows, and manages its own settlement and safeguarding accounts with the sponsor — those latter accounts being commercial bank accounts, and not to be confused with a Bank of England settlement account.

The eligibility criteria for the settlement account are set by the Bank of England, not by Pay.UK. To hold a settlement account, a firm must either hold a reserves account or be a financial market infrastructure, or an e-money or payment institution authorised in the UK by the Financial Conduct Authority, and must meet the settlement participation criteria of the relevant payment system.

That last clause is the one that changed the shape of the industry. In July 2017 the Bank of England extended direct access to RTGS settlement accounts to non-bank payment service providers. The consequences arrived quickly: Ebury became the first directly connected non-settling participant in February 2018, and TransferWise — now Wise — became the first non-bank settling participant in April 2018, followed by Equals Money in February 2019. Pay.UK’s own access programme, launched in 2014, has seen the number of direct Faster Payment participants more than triple.

The current shape of that is visible in the published participant lists. As last updated in April 2026, the Faster Payment System participant list named 47 organisations, against 32 for Bacs. Among the 47 are Adyen, Banking Circle, ClearBank, Corpay, Modulr, PPS, Revolut, Soldo, Square, Stripe UK, SumUp, The Bank of London and Wise — a set that would have been unimaginable as direct participants in a UK clearing system fifteen years ago. Faster Payments is now, by participant count, materially more open than Bacs.

Pay.UK also operates an accreditation regime for technical aggregators, solution providers and consultants, with a published fee structure for Faster Payment System technical accreditation. For most new entrants the practical route to a direct connection runs through one of those accredited providers rather than through building the gateway in-house.

Settlement: deferred net, prefunded#

Settlement of Faster Payments takes place three times each business day in the Bank of England’s RTGS system. The Bank’s classification is explicit: this is one of four settlement models it supports, and Faster Payments sits in the prefunded deferred net settlement category, alongside Bacs and the Image Clearing System.

The mechanics are multilateral netting. Between settlement cycles, participants exchange customer payments through the central infrastructure, entirely outside RTGS — the Bank notes that RTGS does not need to be open for this exchange to happen, which is what makes the 24/7 service possible at all. Obligations accrue. At each settlement cycle, the operator calculates each settling participant’s multilateral net position against all others, and a single net amount per participant is settled across accounts in RTGS.

Because obligations accrue between cycles, settlement risk arises between cycles. The mitigation is prefunding, introduced for both Bacs and Faster Payments in September 2015. Each settling participant holds cash in a dedicated prefunding account in RTGS. That balance caps the participant’s maximum net obligation to the system. If a settlement participant defaults, the segregated cash is used to complete settlement. The effect, in the Bank’s own words, is that prefunding “eliminates settlement risk in systems that use it by capping the maximum net obligations of settlement participants” and “eliminates credit risk between settlement participants”. Before 2015, the exposure was managed by a mutualised loss-sharing arrangement among participants; prefunding removed that mutualisation entirely.

The cap itself is the Net Sender Cap. It limits a participant’s indebtedness to other participants within each settlement cycle, and it is backed by the prefunded balance. Historically Pay.UK set Net Sender Caps prescriptively, using a Minimum Net Sender Cap together with a Peak Contingency Value that participants had to hold at all times. The observed problem was that participants held more liquidity than they needed, with the high-usage cases occurring only infrequently — dead capital sitting in a segregated account at the Bank of England against a peak that rarely came.

On 1 July 2026, Pay.UK went live with a flexible model. Participants may now determine their own Net Sender Cap values according to their own business requirements, submitting plans quarterly for Pay.UK review. Adoption is optional; participants may remain on the existing prescriptive framework. Real-time alerts and the existing settlement protections are retained, and Pay.UK states that no technical change is required to adopt the new model. The intended effects are lower liquidity cost for existing participants and a lower barrier to entry for new ones, particularly non-banks, for whom the prefunding requirement is a direct and highly visible cost of participation.

The numbers give a sense of the netting efficiency. The Bank of England’s published average daily RTGS settlement values for Faster Payments were £2,008 million in Q3 2025, £2,104 million in Q4 2025, £2,287 million in Q1 2026 and £2,298 million in Q2 2026. Set that against Pay.UK’s reported £4.84 trillion of Faster Payments in 2025 — roughly £13.3 billion for every day of the calendar year, and closer to £19 billion for every business day, since payments accrue over weekends and settle on working days. The ratio of customer payments to central bank money moved therefore sits somewhere between six and nine to one depending on how the days are counted. Either way, the great majority of the value cancels before anything settles. It is also worth noting that Bacs, with far fewer participants and a slower cycle, settles a larger daily net value than Faster Payments: £6,205 million average daily in Q2 2026. Faster Payments dominates by volume and by public visibility, not by settled value.

RTGS operating hours are the constraint on all of this. At the time of writing they run 06:00 to 18:00. The Bank has announced that CHAPS will open at 01:30 from September 2027, giving a 16.5-hour settlement window, and on 18 May 2026 published a consultation paper on further extension towards near 24x7 settlement, including the possibility of an additional weekend settlement day, most likely Sunday. The Bank identifies liquidity and risk management as one of the three principal drivers, noting explicitly that longer operating hours “could enable more frequent net settlement cycles, reduce settlement risk, and allow liquidity to be used more efficiently, lowering prefunding and liquidity cost”.

The Bank’s own worked illustration of the current exposure is the sharpest statement of the seam in this chapter. Between settlement cycles on a normal weekday, the lag “could be several hours”. Over an Easter weekend, the gap between settlements runs “from between Thursday evening and Tuesday morning”. In 2026 that means the last settlement before Good Friday on 3 April, and the next on Tuesday 7 April — four full days during which Faster Payments continue to flow at full volume, receiving institutions continue to credit customers, and the obligations pile up behind prefunded balances. The Bank makes a related observation in the same paper that deserves to be read by anyone modelling stress: outflows from banks can already take place at the weekend through Faster Payments, because it operates 24x7, at a time when most other sterling flows are closed.

Messages and standards#

The Faster Payment System’s native message set is derived from ISO 8583 — the same lineage as the card messages in Volume III, which is why an engineer who has worked on card switching finds the FPS message structure oddly familiar. Pay.UK maintains an FPS ISO 20022 Standards Library, described by Pay.UK as “a reference source providing information on the recommended conversion of ISO 8583 messages to the ISO 20022 messaging standard”. The library sits behind registration on Pay.UK’s standards portal rather than on the open website.

Beyond the payment messages themselves, participants exchange status reporting — in Faster Payments terminology, the fate of a payment — and a set of exception messages covering recalls, returns and related error flows. A direct or indirect participant must be able to generate, receive and process all of them; supporting only the happy path is not participation.

Confirmation of Payee#

Confirmation of Payee is an overlay service, not part of the Faster Payment System itself, but no honest account of a UK account-to-account payment in 2026 omits it. Launched by Pay.UK in 2020, it checks the payee’s account name against the account details entered before the payment is initiated, and returns a result to the payer. Its purpose is to reduce misdirected payments and to attack the specific fraud pattern in which a customer is persuaded to send money to an account in a different name from the one they believe they are paying.

Adoption is now broad: Pay.UK reports over 300 organisations participating and more than two million checks completed every day. Participation follows the same shape as the payment systems themselves — direct participants that are account-holding PSPs, and indirect participants reaching the service through accredited PSP aggregators. The Payment Systems Regulator mandated an expansion of coverage in 2024.

Liability: who pays when it goes wrong#

Two regimes matter.

For payments sent to the wrong account in error, the industry operates the Credit Payment Recovery process. It is a best-efforts recovery mechanism, not a right of reversal; the receiving customer’s consent, or a legal process, is generally required to get the money back.

For authorised push payment fraud — where the customer was deceived into authorising the payment themselves — a mandatory reimbursement requirement has applied to Faster Payments since 7 October 2024. The maximum level of mandatory reimbursement is £85,000 per claim, confirmed by the Payment Systems Regulator in policy statement PS24/7; the Bank of England, as operator of CHAPS, set the same £85,000 maximum for CHAPS to keep the two consistent. The PSR’s assessment was that £85,000 covers 99.8% of Faster Payments APP scams by volume and 90% by value. Firms may apply an optional excess of up to £100, but may not apply it to vulnerable consumers. Claims must be reported within 13 months of the payment. Reimbursement is due within five business days of the claim, with the clock stoppable where the firm needs more information, and a decision required within 35 business days. The cost is split 50:50 between the sending and the receiving payment service provider — a design choice that deliberately puts the receiving institution, which onboarded the mule account, on the hook for half.

Underneath both regimes sits the statutory backstop in the UK’s payment services regulations, derived from PSD2: a payment instructed by mobile, internet or telephone banking must arrive by the end of the following business day at the latest. Faster Payments beats that by roughly a day and a half. The regulation is the floor, not the standard.

What Faster Payments is not#

It is not the rail for large value. Above £1 million, and for anything requiring same-day irrevocable settlement in central bank money on a per-payment basis, the answer is CHAPS, which settles gross across RTGS with an early-afternoon cut-off for the same-day guarantee and carries the large majority of sterling value on a tiny fraction of the volume.

It is not real-time settlement, and describing it as such in a prospectus, an architecture document or a liquidity model is a material error rather than a simplification.

It is not the same thing as open banking. Open banking is an API layer that allows a third party to initiate a payment from a customer’s account with the customer’s consent; the payment it initiates is, overwhelmingly, a Faster Payment. Open banking has no clearing, no settlement, no participant list and no prefunding. It is a way of instructing a rail, not a rail.

And it is not, any longer, a banks-only club. Nearly half the named direct participants are non-banks, they hold settlement accounts at the Bank of England on the same terms as the clearers, and the July 2026 change to the Net Sender Cap regime was explicitly designed to make that easier still.

What this means if you are building on it#

Three practical consequences follow from the technical picture, and they are the ones most often got wrong.

The first is that prefunding is a real and continuous cost of direct settling participation, sized to your peak net sending position between cycles rather than to your average. It is cash, it is segregated, and until the flexible Net Sender Cap model it was sized conservatively by rule rather than by your own traffic. If your business is heavily one-directional — a payout business that sends far more than it receives — your net sender position, and therefore your prefunded balance, is close to your gross volume, and the economics of direct settling participation look very different from those of an institution with balanced two-way flow.

The second is that there is no cut-off to design around, and that is an unfamiliar discipline for teams coming from Bacs. There is no submission window, no processing day, no entry day. There is a queue, a fifteen-second response expectation, a two-hour outer bound, and a settlement cycle you do not control and cannot see. Reconciliation must therefore be built against your own message-level fate reporting rather than against a settlement file, because the settlement file tells you about a net figure that bears no reconstructable relationship to individual payments.

The third is that the weekend is the risk. Volume does not stop on Friday evening; settlement does. Whatever exposure your model tolerates on a Tuesday afternoon, multiply it by the length of the longest bank holiday weekend in the calendar and check that the prefunded balance and the Net Sender Cap still work. That is the calculation the Bank of England is effectively inviting the industry to revisit in its May 2026 consultation, and it is the single most instructive number in the whole system: the measure of how far a payment that arrives in one second can outrun the money behind it.

37.98 Common wrong ideas#

Wrong: Faster Payments settles in real time. Right: It is a deferred net settlement system; the interbank leg settles three times each business day at the Bank of England, and only the credit to the receiving customer is immediate.

Wrong: The whole system runs twenty-four hours a day. Right: The messaging service does; RTGS runs 06:00 to 18:00 on working days, with an extension to 01:30 announced for September 2027 and near round-the-clock settlement still at consultation.

Wrong: You can send £1 million by Faster Payment from any account. Right: £1 million is the central scheme ceiling per transaction; institution-set limits bind first, with HSBC’s online personal limit at £25,000 and Metro Bank’s at £21,000 as published in July 2026.

Wrong: A Faster Payment is always instant. Right: Funds are usually available almost immediately, but the scheme’s own published expectation allows up to two hours, and under 0.1% of UK accounts do not accept Faster Payments at all.

Wrong: Once sent, nothing whatever can be done. Right: The payer cannot cancel, but Credit Payment Recovery exists for payments sent to the wrong account in error, and a mandatory reimbursement regime capped at £85,000 per claim has applied to APP fraud since 7 October 2024.

Wrong: Faster Payments settles more money than Bacs. Right: Bacs settles a larger daily net value — £6,205 million average daily in Q2 2026 against about £2,298 million for Faster Payments — because Faster Payments dominates by volume and visibility, not by settled value.

Wrong: Open banking is a rail competing with Faster Payments. Right: Open banking has no clearing, no settlement, no participant list and no prefunding; it is an API layer, and the payment it initiates is overwhelmingly a Faster Payment.

Wrong: Standing orders inherit the 24/7 property of the scheme. Right: They execute on working days only, roll forward when the nominal date is a weekend or bank holiday, and cluster early, with at least 90% sent by 6am.

Wrong: Faster Payments is a banks-only club. Right: The participant list named 47 organisations as last updated in April 2026, including Adyen, ClearBank, Modulr, Revolut, Square, Stripe UK, SumUp and Wise, making it materially more open than Bacs.

Wrong: You can reconcile individual payments against the settlement file. Right: The settlement file reports one net figure per participant per cycle, with no reconstructable relationship to individual payments; reconciliation must be built against message-level fate reporting.

37.99 Chapter summary in 20 lines#

1. Before May 2008 a British payment instructed by telephone or internet banking went into the Bacs cycle and arrived three working days later.
2. The Faster Payment System launched on 27 May 2008 and created the expectation that an account-to-account payment is a real-time event at any hour on any day.
3. Nothing actually travels between the banks when you send one: a message goes across, and the receiving bank credits its customer out of its own funds.
4. The speed comes from not moving money at all and having somebody at the far end pay out on a promise.
5. The banks square up three times each business day, netting everything they have exchanged and moving one figure per participant at the Bank of England.
6. That netting is what makes the service cheap enough to give away, with somewhere between six and nine pounds of customer payments discharged for every pound of central bank money moved.
7. The exposure created between the message and the settlement is capped by prefunded cash held in a segregated account in RTGS.
8. Prefunding, introduced in September 2015, replaced a mutualised loss-sharing arrangement and caps each participant’s maximum net obligation through the Net Sender Cap.
9. On 1 July 2026 Pay.UK went live with a flexible model letting participants set their own Net Sender Cap values, lowering liquidity cost and the barrier to entry.
10. “Instant” describes the customer experience only, because Faster Payments is a deferred net settlement system in precisely the same category as Bacs.
11. The gap between a customer payment and its settlement can be several hours on a weekday and, over an Easter weekend, runs from Thursday evening to Tuesday morning.
12. “24/7” belongs to the messaging layer, while the settlement layer beneath it keeps banking hours and does not work weekends.
13. The £1 million per-transaction ceiling, in force since 10 February 2022, is rarely the binding limit, because institution-set limits are often two orders of magnitude lower.
14. Four transaction types share one clearing mechanism: Single Immediate Payments, forward-dated payments, standing orders and Direct Corporate Access.
15. Standing orders are the exception to everything always-on, running on working days only, rolling forward, and clustering before 6am.
16. Participation comes in three shapes — directly connected settling, directly connected non-settling, and indirect through a sponsor — and only the first holds its own Bank of England settlement account.
17. Since the Bank opened settlement accounts to non-bank payment service providers in July 2017, the direct participant list has grown to 47 organisations, nearly half of them non-banks.
18. The sender cannot cancel a payment, but Credit Payment Recovery covers misdirection and mandatory reimbursement, split 50:50 between sending and receiving firms, covers APP fraud.
19. Faster Payments is not the rail for large value, is not real-time settlement, and is not the same thing as open banking.
20. If you build on it, prefunding is a continuous cost sized to your peak net sending position, there is no cut-off to design around, and the weekend is where a payment that arrives in one second most outruns the money behind it.

Sources: Pay.UK published scheme material (wearepay.uk Faster Payment System pages covering how Faster Payments work, transaction types, transaction limits, participation, participant lists and the FPS ISO 20022 Standards Library; Confirmation of Payee pages; the 1 July 2026 flexible Net Sender Cap announcement and explainer; the 17 December 2025 Vocalink contract extension announcement; and the February 2022 £1 million limit press release), the Bank of England (payment and settlement pages, payment and settlement statistics, and the consultation paper “Extending RTGS and CHAPS settlement hours — next steps towards near 24x7 settlement”, published 18 May 2026, together with the 2016 RTGS PFMI self-assessment for the September 2015 introduction of prefunding), and the Payment Systems Regulator (PS24/7 on the maximum level of reimbursement, APP fraud reimbursement protections, and the Interbank Infrastructure Renewal project page). Launch history and early participant milestones cross-checked against public record. All limits, timings, figures and participant counts accurate at time of writing.