Wallets and EDPs
Money mule detection for e-wallets and payment institutions
· 7 min read
Mobile wallets and electronic payment institutions have brought millions of people into digital payments across North Africa and the Middle East. Fast onboarding, instant transfers and cash-out points everywhere are what make them useful. They are also exactly what a fraudster needs to move stolen money quickly.
The accounts that carry that money are called money mules. Catching them is now one of the main fraud and anti-money-laundering (AML) problems for wallets, and regulators expect it.
What a money mule is
A money mule is an account used to receive and pass on funds that came from fraud: a phishing scam, a fake online sale, a hacked account, a social-engineering call. Some mules are fake identities. Many are real people: students, job seekers, people offered a small fee to “lend” their wallet, or people who sold their account outright. Because the person is real, the KYC check passes.
Mules rarely work alone. A single scam usually feeds several mule accounts, which forward to a few collector accounts, which cash out. That structure is a fraud ring.
Warning signs in wallet transaction data
- Pass-through behaviour. Money comes in and leaves again within minutes or hours, leaving the balance near zero.
- Many unrelated senders. An account that suddenly receives from people it has never dealt with, often in similar amounts.
- Fan-in, fan-out. Several accounts paying into one, which then pays out to a few others or to cash-out agents.
- Dormant then busy. An old, quiet account that becomes very active overnight.
- Shared devices and SIMs. Several “independent” wallets operated from the same phone, or a phone number that keeps moving between accounts.
- Agent concentration. Cash-outs clustered at the same few agents or merchants, sometimes right after the incoming transfers.
- Amounts just under limits. Transfers that sit just below daily limits or reporting thresholds, also known as structuring or smurfing.
Why mules are hard to catch with rules alone
Every rule has a threshold, and mule networks learn the thresholds. They spread money across more accounts, wait a little longer between hops, and vary the amounts. Each account looks like a busy but normal customer. The pattern is only clear when you follow the money across accounts and see the same devices, agents and timing repeating.
A practical approach for wallets
- Map the flows. Build a graph of who pays whom, and which devices, SIMs and agents each account uses.
- Score accounts, not just transfers. Pass-through ratio, number of new counterparties and time between in and out say more than any single amount.
- Investigate networks. When one mule is confirmed, look at everything linked to it. The rest of the ring is usually one or two hops away.
- Act before cash-out. The window between the first incoming transfer and the withdrawal is short. Real-time scoring on transfers and cash-outs matters more than a monthly report.
- Keep the evidence. A clear write-up of each case makes suspicious activity reports to the financial intelligence unit faster and better.
Privacy matters too
Fraud analysis touches phone numbers, national IDs and account numbers. Masking that data before analysis, while keeping the links between records intact, lets a team find networks without spreading personal data to more systems and people than necessary.
That is how Argos works: it reads your wallet data without changing it, masks personal data first, finds mule networks and cash-out rings in your history, explains each one, and scores every new transfer. It has engines tuned for wallets and EDPs, PSPs, acquirers, e-commerce and banks, and reads amounts in any currency, millimes included.