Fraud rings, explained

What is a fraud ring? How to detect organised payment fraud

· 7 min read

Most fraud teams are set up to judge one payment at a time. Is this amount unusual? Is this card new? Is this IP on a blocklist? Those questions catch the careless fraudster. They miss the organised one, because organised fraud is designed to make every single payment look almost normal.

The unit of organised fraud is not the payment. It is the fraud ring: a set of accounts, cards, devices and merchants that are controlled by the same person or the same script, and that only look suspicious when you see them together.

What a fraud ring looks like

A ring can be three accounts or three hundred. What they have in common is that something links them which should not link strangers:

  • A shared device or browser fingerprint. Twelve “different” customers who all log in from the same phone.
  • A shared server. Accounts created from the same rented cloud IP range, often outside the country they claim to be in.
  • Shared payment instruments. One card added to many wallets, or many cards cashed out to one bank account.
  • Shared counterparties. Different senders who all pay the same small group of receivers, a classic sign of money mules.
  • Shared timing. Accounts that sign up, top up and cash out in the same few minutes, night after night.

On its own, each of these links has an innocent explanation. Families share a phone. Offices share an IP. Together, and at scale, they describe an operation.

Common kinds of rings in payments

Money mule networks

Stolen or scammed funds are split across many accounts, moved a few times to blur the trail, then withdrawn as cash or crypto. Mule accounts are often real people who were recruited or paid for their account, which is why their KYC looks clean.

Card testing farms

A script tries thousands of stolen card numbers with tiny payments to find the ones that still work. The payments come from many accounts and many cards, but from a handful of devices and servers. See our guide to spotting card testing attacks.

Promo and bonus abuse

One person opens dozens of accounts to claim a welcome bonus or cashback many times, then funnels the value into a single account.

Collusive merchants

On acquirers, PSPs and PayFacs, shell sub-merchants share payout accounts, owners or devices and push fake or laundered transactions through what look like unrelated shops.

Why rule engines miss rings

A rule like “flag more than five payments per hour from one card” is easy for a ring to stay under: it just spreads the activity across more cards and accounts. Each piece stays below every threshold. The organisation only becomes visible when you connect the pieces.

Tightening the thresholds does not fix it. It mostly blocks good customers and floods the team with alerts, while the ring adapts within days.

How to detect a fraud ring

Ring detection starts by treating your data as a graph: accounts, cards, devices, IPs, phones, emails and bank accounts are nodes, and every time two of them appear together you draw an edge. Rings then show up as clusters that are far denser than normal customer behaviour.

  1. Resolve identities. Normalise phones, emails, card fingerprints and device IDs so the same thing is always the same node.
  2. Link entities. Build edges from shared attributes and from money flows between accounts.
  3. Find unusual clusters. Look for groups that share too much, move money in loops, or act in lockstep compared with the rest of your customers.
  4. Explain the cluster. An analyst needs to know who runs it, how the scheme works and what to do first, not just a score.
  5. Stop the next payment. Once a ring is known, any new payment touching it should be scored higher in real time.

What to look for in your own data today

You can start without any new tool. A few queries go a long way:

  • Devices or IPs used by more than a handful of distinct accounts in the last 30 days.
  • Bank accounts or wallets receiving money from many unrelated senders.
  • Cards added to more than one account.
  • Bursts of sign-ups followed by quick top-ups and withdrawals.

If those lists are long, you probably have rings in your history. That is the problem Argos was built for: it connects read-only to your database or a file, masks personal data before analysis, finds the rings already in your history and writes up each one like an investigator would.

More from the blog