Why a coached payment passes every check
In a coached payment the real customer signs in on their own phone and approves the payment themselves. Each check does its job, and the money still leaves. What changes the outcome is reading the session as a whole.
In this article
The customer is real, and so is every factor
Most controls in a banking app answer one question: is this the customer? In a coached payment the answer is yes. The customer opens the app on the phone bound to their account, passes the biometric, receives the one-time code on their own number and types it in. A caller on the line tells them what to do at each step.
This is the fraud that is growing. In the UK, losses to scams in which customers are tricked into paying rose 19% in 2025 to £576.4 million, while unauthorised fraud fell 5%1. In India, reported cyber-fraud losses rose from ₹7,465 crore in 2023 to ₹22,846 crore in 20242.
What each check sees
- Device recognition sees the customer’s own phone.
- Biometric login sees the customer’s own face or fingerprint.
- A one-time code reaches the customer’s own number, and the customer types it.
- An amount rule sees one payment, and either lets it through or blocks many genuine ones.
Singapore’s central bank has said how hard the problem is: telling deceived, self-authorised transfers apart from legitimate ones “is not an easy task even for sophisticated fraud detection systems”3.
What the session shows
A coached payment leaves its traces around the payment, not in it. The caller has to stay on the line. Many scams ask the customer to share their screen: in June 2025 Singapore Police warned about impersonation scams using WhatsApp’s screen-sharing function, citing at least 46 cases and losses of at least S$3 million4. The payee’s details are read out and typed a few characters at a time. And the money has to land somewhere, often in an account linked to others.
Each of these alone is weak evidence: people take calls while they bank, and some type slowly. Read together in one session, they are a pattern a bank can act on.
From four alerts to one decision
DeepID reads the call, the screen, the typing and the payee in the same session and returns one decision with its reasons: allow, warn, cool off, or hold and call back. The customer reads words written for them, and the bank’s fraud system receives the same decision as an event. The guide to coached payments follows one session step by step.
Sources
Secondary sources are news or trade reporting of a rule or figure. Check the regulator’s own text before relying on them. Nothing here is legal advice.
- UK Finance (opens in a new tab), Jun 2026Annual Fraud Report 2026
- MHA (opens in a new tab), Dec 2025Ministry of Home Affairs, Lok Sabha reply on cyber-fraud losses reported to I4C
- MAS (opens in a new tab), Feb 2026Written reply to parliamentary questions on the SRF and fraud surveillance
- Singapore Police Force (opens in a new tab), Jun 2025Police advisory on impersonation scams involving the WhatsApp screen-sharing function