CVV Attack Defense: Stop Card Testing and CVV Brute Force

A CVV attack is an automated run of authorization attempts against card numbers using guessed verification codes. The goal is usually one of two things: find which stolen cards are still live, or guess the three or four digit code outright on a card the attacker already holds. Defense comes down to three layers. Make enumeration expensive through rate limits, device and IP velocity checks, and bot detection. Refuse to leak signal by returning identical decline responses. Push the check back to the issuer with 3-D Secure. Blocklists, alerting, and chargeback monitoring are cleanup around those three, not substitutes for them.

What the attack looks like on your side

Most merchants see the traffic before they see the fraud. A few patterns show up again and again in authorization logs.

The tell is not any single attempt. It is volume with a shape, hitting many cards one way and many attempts per card the other way.

Why the CVV field is a weak gate on its own

The verification value is short. Visa, Mastercard, and Discover use three digits, American Express uses four. That is a thousand or ten thousand possible values, which sounds like a lot until you remember an attacker can send requests at machine speed. Issuer behavior matters too. Some issuers will not hard-decline on a CVV mismatch alone, and enforcement varies by market and card product. A merchant who treats a CVV match as proof of anything is overreading it. It is one signal, and a small one.

Layer one: make enumeration expensive

You cannot stop a determined attacker, but you can make the cost per guess high enough that the economics stop working.

Rate limits belong at the edge, before the request reaches the payment orchestration layer, so the attacker never gets issuer feedback at scale.

Layer two: stop leaking signal

This is where most implementations give away the answer. If your checkout returns "incorrect security code" for a CVV mismatch and "card declined" for everything else, you have built an oracle. The attacker only needs the distinction, not the reason.

Return one generic decline to unauthenticated users regardless of cause. Log the actual issuer response code internally, where fraud and support teams can see it, and never surface it. Keep response timing roughly uniform as well. A mismatch that returns in 80 milliseconds while a valid card returns in 900 is a signal too. If you run a call center, the same rule applies to agents: they should not read back the reason for a decline.

Layer three: push verification to the issuer

3-D Secure, in its current EMV version, moves the cardholder authentication step to the issuer and shifts liability for eligible fraud. It also breaks scripted enumeration, because the attacker now needs to pass an authentication challenge on every card. Friction is the tradeoff, so use risk-based rules. Trigger the challenge on new devices, mismatched geolocations, high velocity, and first-time customers above a value threshold. Let low-risk repeat buyers through with an exemption request where your acquirer supports it. Track the challenge rate and the abandonment rate after a challenge; if abandonment climbs past a few points, your trigger rules are too broad.

Combining AVS and CVV into something useful

Address Verification System results and CVV results are more useful together than apart. I look for these combinations in rules.

Feed acquirer and network fraud reports back into these rules. Your processor sees enumeration across its whole portfolio and will often flag a BIN range before you notice it on your own traffic.

Data handling: what you cannot keep

PCI DSS prohibits storing the card verification code after authorization, in any form, including encrypted. It cannot go in a database column, a log line, a support ticket, a webhook payload, or an analytics event. Masking is not enough if the field itself is retained. This matters for defense as well as compliance: a merchant who keeps CVVs becomes a target, and a breach exposes values attackers can replay elsewhere. Audit your logging pipeline for accidental capture of full request bodies at the payment endpoint. That is the most common place it turns up.

A checklist I run through

  1. Rate limits on card number, IP, device, and shipping address, enforced before the payment call.
  2. One generic decline message, no field-level reasons exposed to the client.
  3. A challenge after the first failure, not after a streak.
  4. 3-D Secure triggered by risk rules, with challenge and abandonment rates monitored.
  5. Alerting on decline-rate spikes per BIN and per payment method, not just overall.
  6. CVV values absent from logs, queues, and analytics, confirmed by a grep of real traffic samples.
  7. Rules reviewed monthly against chargeback data, because attacker tooling changes faster than your rule set.

None of this is exotic. The merchants who get hit hard are usually missing the boring parts: a rate limit, a uniform error message, and a check on what their logging actually captures.

More

Read our complete guide: Buy CVV Cheap: Pricing, Risks, and What First-Time Buyers Need to Know