E-commerce and PSP fraud
Card testing attacks: how to spot them in your transaction data
· 6 min read
Stolen card numbers are cheap, but most of them no longer work. Before using a batch, fraudsters find out which cards are still live by pushing small payments through a merchant that will let them. That is card testing, also called carding or a BIN attack when the numbers are generated from a known card prefix.
For the merchant or PSP that gets used as the test bench, the cost is real: authorisation fees, a spike in declines that hurts your standing with the card schemes, chargebacks on the payments that went through, and sometimes a payment page knocked over by the volume.
The patterns card testing leaves behind
- Many small amounts. Payments of one unit of currency, or the minimum your checkout accepts, far more often than real customers make them.
- A high decline rate. Most test payments fail with “invalid card”, “expired” or “do not honour”. A sudden jump in decline rate is often the first visible sign.
- Sequential or same-BIN cards. Card numbers that share the first six or eight digits, tried one after the other.
- Few devices, many cards. Hundreds of cards from a handful of devices, browsers or IPs, often on cloud or hosting networks rather than home or mobile connections.
- Mismatched details. Billing country, IP country and card issuer country that do not line up, or names and emails that look generated.
- Odd hours and steady rhythm. Attempts at a regular pace, often at night local time, with no browsing before checkout.
Why per-payment rules struggle
Each test payment, taken alone, is tiny and unremarkable. Velocity rules per card do not fire because each card is tried once or twice. Velocity rules per IP are easy to dodge with rotating proxies. What stays constant is the operation behind the payments: the same scripts, the same few devices, the same cash-out accounts for the cards that work. That is a fraud ring, and it is easier to see as a group than one payment at a time.
How to stop card testing
- Watch the decline rate in real time, per merchant and overall, and alert on sharp jumps.
- Link payments to devices and networks, not just cards, so that many cards on one device become one case.
- Add friction where the pattern appears: a challenge, 3-D Secure, or a temporary block for the device or network, rather than for every customer.
- Follow the cards that passed. The cards that worked will be used soon, often somewhere else. Flag them and the accounts they get attached to.
- Feed decisions back. Every confirmed attack should make the next one faster to catch.
Check your own history
Card testing often goes unnoticed until a scheme warning or a wave of chargebacks. A quick look at small, declined payments grouped by device, IP and card prefix over the last few months usually tells you whether it has already happened.
Argos runs that analysis for you: it connects read-only to your payment data, groups test payments into rings by device, network and timing, explains each attack, and scores new payments so the next wave is stopped while it is still small.