The Outlier Lab

Payee Confirmation and Payment Friction

Checking a payee's name against the destination account before a push payment, and adding targeted warnings, questions and delays at the moment of payment to interrupt scams and misdirection.

What it is

In Authorized Push Payment (APP) Scams, the legitimate customer is the one sending, so authentication controls don't help. The customer passes MFA, uses their own trusted device, and authorizes the payment. The decision point is the payment itself: who is being paid, why, and whether the customer understands what they're doing. This control works at exactly that point, and it is the main transaction-level defense against scams.

1. Payee confirmation (name check). The payer's bank sends the entered payee name and account identifiers to the payee's bank, or to a directory service, which returns a result. Under the EU regulation the outcomes are match, close match, no match, or a fourth "other" result when verification isn't possible; the UK scheme uses an equivalent tiered set. After a close match or no match, the payer can still choose to continue.

  • UK Confirmation of Payee (CoP). PSR Specific Direction 10 (August 2019) required the six largest banking groups, involved in about 90% of Faster Payments and CHAPS transactions, to implement CoP by the end of March 2020. Specific Direction 17 (in force October 2022) extended it: Group 1 firms by 31 October 2023, and about 400 more Group 2 firms by 31 October 2024. A 2026 draft variation would add a Group 3 (deadline 31 December 2026) and keep the direction running past its original November 2026 expiry.
  • EU Verification of Payee (VoP). Required by the Instant Payments Regulation (Regulation (EU) 2024/886) for both instant and standard euro credit transfers, free of charge, with the result given before the payer authorizes. Euro-area PSPs had to offer it from 9 October 2025; PSPs in non-euro member states from 9 July 2027. Non-consumer payers can opt out (for example for bulk files) and carry the resulting risk.
  • US. No mandated equivalent. Some networks and bank consortia offer account-owner or name-match services, and P2P apps often show the registered recipient name. US payers mostly rely on each institution's own payee controls (Bank Account Ownership Verification covers the payout-side equivalent).

2. Risk-scaled payment friction.

Friction levelWhenForm
NoneEstablished payee, normal amount, no scam signalsNothing added
LightNew payee, low amountName-check result plus a generic reminder
Dynamic warningNew payee and scam-pattern indicatorsWarning specific to the scam type ("Your bank will never ask you to move money to a 'safe account'")
Purpose questionsElevated scam score"What is this payment for?" The answer routes to tailored warnings; some answers trigger a hold
Cooling-off delayHigh scam score, large or irreversible paymentDelay of hours to a day, with a notification through a trusted channel
Human interventionVery high risk; vulnerable customer; repeat warnings overriddenOutbound call from trained staff; in-branch conversation; trusted-contact outreach
BlockKnown scam payee; confirmed mule account; consortium hitDecline, with an explanation and support
Diagram: Customer initiates push payment, Payee name check, Send, Show actual name; customer confirms, Strong warning; extra checks, Scam score (behavior, payee risk, context), Purpose question + dynamic warning, Delay + trusted-channel contact / human call, Block

What it stops (and what it doesn't)

  • Strong against:
    • Misdirected payments caused by typos
    • Payee-detail substitution, where a fraudster swaps account details in an invoice or message, because the name won't match
    • "Safe account" scams that send victims to accounts in third-party names
    • Many first payments to mule accounts, where the name the victim was told doesn't match the account holder
  • Moderate against: impersonation scams. Dynamic warnings and purpose questions interrupt a meaningful share, especially when they name the specific script the victim is hearing.
  • Weak against:
    • Scams where the destination account is actually in the name the victim expects, such as a mule opened in the "advisor's" name or a crypto exchange account in the victim's own name (Investment Scams (Pig Butchering))
    • Victims coached to ignore warnings or to give false purpose answers
    • Payments that leave via crypto purchases or gift cards rather than bank transfers (Gift Card Fraud)
  • Adversary adaptation (conceptual): as name checks spread, scammers shift toward accounts whose names fit the story, toward cards and crypto rails, and toward coaching victims through warnings. Expect GenAI-scaled scripts to make that coaching more convincing (Generative AI and Deepfake Fraud). Payee-side controls, meaning mule detection at the receiving institution, become more important.

Design considerations

  • Make the result unmistakable. A no-match must look different from a match. Interface studies and scheme guidance agree that how the result is presented largely decides whether it works. Never default to "continue".

  • Give friction a purpose. A generic "Are you sure?" is ignored. Warnings tied to the scam type the model suspects (impersonation, investment, romance, purchase, invoice), written in plain language, work better. Purpose questions give you both a signal and a route to the right warning.

  • Score scam risk separately from ATO risk. Scam indicators differ from takeover indicators:

    See Real-Time Transaction Risk Scoring on why one score shouldn't serve both.

  • Make delays work. Cooling-off delays succeed when they break the scammer's control (urgency, isolation). Pair the delay with outreach through a trusted channel: a call from a known number, an in-app message, contact with a trusted contact (Trusted Contact and Senior Investor Holds).

  • Cap first payments to new payees. A low limit on the first payment, rising with payee age, reduces how much a victim can lose before second thoughts or intervention (Velocity and Exposure Limits).

  • Use your receiving side too. As a receiving institution, inbound credits to new accounts that fan out quickly are a mule signal. Under Nacha's 2026 rules RDFIs must monitor ACH credits for fraud, including False Pretenses (Money Mules, Transaction Monitoring).

  • Share payee intelligence. Known scam payees and mule accounts identified by one institution are valuable to others (Consortium and Data Sharing).

  • Industry lens:

    • Brokerage: outbound wires or ACH after liquidating a portfolio are classic investment-scam and elder-exploitation exits. Pair purpose questions with trusted-contact outreach and FINRA 2165 holds.
    • Neobanks / P2P: new-payee P2P and RTP sends. Name display plus dynamic warnings, with the first-send cap.
    • Marketplaces: off-platform payment requests and fake "payment received" messages (Seller-Targeted Scams (Fake Payment and Overpayment)). In-product warnings when a buyer or seller asks to move off-platform.
    • Retail: gift-card purchase interstitials at checkout for large or unusual gift-card baskets (a scam payment rail).
    • Cashback / rewards: lower exposure. Relevant when reward balances can be pushed to third-party P2P handles.

Implementation path

StageWhat it looks likePrerequisites
CrawlShow the recipient's registered name where available; static scam warning on new-payee payments; first-payment capsPayee-add flow; limits framework
WalkFormal name check (CoP/VoP where available; network or vendor name-match elsewhere); match-tier UX; scam score separate from ATO score; dynamic warnings by scam type; purpose questions; cooling-off on high-risk paymentsName-check scheme participation or vendor; scam labels; UX research
RunScam-specific models with session and behavioral features; trained intervention team for outbound calls; trusted-contact workflows; consortium payee intelligence; inbound mule detection on the receiving side; A/B testing warning effectivenessBehavioral Biometrics, Consortium and Data Sharing, Experimentation and Champion-Challenger, Playbook - Scam Intervention Members

Customer friction and trade-offs

  • Payment friction directly affects payment conversion. Close-match and no-match results on legitimate payments (business trading names, joint accounts, nicknames) cause abandonment and support contacts. Scheme data on genuine no-match rates should guide UX.
  • Overused warnings train customers to click through them. Keep heavy friction for the riskiest few percent of payments.
  • Delays on legitimate urgent payments (house deposits, tuition) generate complaints. Have a clear expedited path that requires stronger verification, so persistence alone doesn't get a payment through.
  • Where reimbursement is mandatory (UK) or contested (US P2P), friction design is also a legal and financial decision (Case Study - UK APP Scam Reimbursement Rules (2024), Case Study - Zelle Scam Liability Disputes (2024-2025)).

Measuring effectiveness

  • Scam loss per 1,000 payments to new payees, by rail, segment and friction level, using matured labels from scam reports.
  • Warning efficacy: the abandonment rate after a warning, split by later-confirmed scam versus genuine. You want high abandonment on scams and low on genuine payments.
  • Name-check outcome mix: match, close match, no match and unavailable, and the payment completion rate after each.
  • Intervention outcomes: of outbound calls and delays, the share of scams stopped and the dollars saved.
  • Override analysis: confirmed scams where the customer overrode a warning. Review these cases to improve the warning wording.
  • Receiving-side mule detection: inbound scam funds frozen or returned.

Common failure modes

  • Weak no-match handling: a no-match warning that looks like a match, or defaults to "continue anyway".
  • One warning for everyone: generic warnings on every payment cause warning fatigue.
  • Step-up used against scams: the customer passes it, and the payment looks more authorized.
  • No scam-specific score, so scam payments look like normal customer behavior to an ATO model.
  • Delays with no outreach, which just postpone the loss.
  • Gaps in coverage: some rails (crypto purchases, card-to-card, gift cards) carry no friction while the bank transfer path is heavily protected, so scams move to the unprotected rail.

Regulatory and policy notes

  • EU Instant Payments Regulation (2024/886). Requires PSPs to offer Verification of Payee for euro credit transfers (instant and standard) before the payer authorizes, free of charge, with match / close match / no match / other outcomes. Deadline: 9 October 2025 for euro-area PSPs; 9 July 2027 for PSPs in non-euro member states. A payer who proceeds after being warned bears the risk of a misdirected payment. The forthcoming Payment Services Regulation would extend name-IBAN checks to credit transfers outside the IPR's scope, but it was not yet adopted as of October 2026 (see Risk-Based Authentication and Step-Up).
  • UK Confirmation of Payee. Operated by Pay.UK under PSR directions: SD10 (six largest groups, by end of March 2020) and SD17 (Group 1 by 31 October 2023; about 400 Group 2 firms by 31 October 2024). UK mandatory reimbursement for APP scams on Faster Payments took effect on 7 October 2024, capped at £85,000 per claim (cut from the £415,000 originally set) (Case Study - UK APP Scam Reimbursement Rules (2024)).
  • US:
    • No CoP/VoP mandate.
    • Reg E. Covers unauthorized transfers; whether induced (authorized) scam payments are covered remains contested (Regulation E and Unauthorized Transfers, Case Study - Zelle Scam Liability Disputes (2024-2025)).
    • Nacha 2026. The new definition of False Pretenses covers inducing a payment by misrepresenting identity, authority or ownership of the receiving account. Name-check-style controls address exactly that. ODFIs and RDFIs must have risk-based processes to identify such entries (Nacha Rules and ACH Risk).
  • FFIEC 2021. Flags risks from push-payment capabilities.
  • Brokerage. FINRA Rule 2165 temporary holds, and Rule 4512 trusted contact, for suspected exploitation of specified adults (Trusted Contact and Senior Investor Holds).

Mechanisms addressed

Sources and further reading