A single card paying for four free trials in a week, or one card surfacing across accounts attached to unrelated devices, is one of the cheapest abuse signals you can collect at checkout. Spotting the same card on multiple accounts does not require storing card numbers: a keyed hash is enough to count reuse, compare it across accounts and devices, and hand the result to a decision your own code makes.
Card reuse is worth tracking because it is hard to fake. Emails are free, cookies can be cleared and a proxy can make the next session look new. The payment instrument is the one item in the flow that costs the person something — a real authorization, a real card, or a stolen number they are spending.
Why one card behind many accounts is worth a look
The signal runs in two directions. One card, many accounts is trial farming: creating an account costs nothing, a new email takes seconds, and the card is the expensive part of the loop. Many cards, one place is card testing: a list of stolen numbers pushed through signup or checkout in small attempts to see which ones authorize, often with repeated declines and attempts arriving faster than a person could type.
Both are multi-accounting problems — one extracts value from your trial or bonus, the other validates stolen data using your checkout as a test bench. Read the card alongside your payment fraud detection signals rather than on its own.
The two patterns need different responses

Trial farming: protect the trial, not the sale
A card that already funded two trials does not have to lose you the customer. A step-up — email verification, or a card-or-passkey check — or a paid plan without a trial keeps the sale while closing the loop. What you want to avoid is blocking a paid checkout for someone who is finally willing to pay.
Card testing: slow the flow before it drains
Here the cost is authorization fees, processor penalties and the risk to your merchant account. The useful response is velocity control on the attempt itself: a captcha or proof-of-work step-up, caps per device and per network, and a clean record of which attempts were rejected and why.
Count reuse with a keyed hash instead of a card number
Two practical routes: use the stable payment-method token your processor returns for a given card and customer, or compute a keyed hash over a normalized string — brand, last four, expiry month and year — inside the component that already handles card data, so the raw number never reaches your database.
Why the hash must be keyed
A plain SHA-256 over a sixteen-digit number is a search problem, not a secret: an unsalted digest can be reversed with modest hardware. A keyed hash — HMAC-SHA-256 with a secret held in your key management service — can only be recomputed by someone who holds the key. Store a key version next to the value so you can rotate without losing history.
What to keep
The hash, first-seen and last-seen timestamps, the accounts it has been attached to, and the requestId that produced it. Treat the hash as sensitive anyway: it is pseudonymous, not anonymous, and erasure requests have to reach it. How a hashed card identifier is treated depends on your processor agreement; have counsel review the scope question before you build retention around it.
Wire it into a checkout decision
Once the key is stable, checkout becomes a policy question. Send it with the rest of the request and let your own code act on the answer:

// pseudo-code: check the exact schema in the docs
const cardKey = hmacSha256(CARD_KEY, `${brand}:${last4}:${exp}:${expYear}`);
const decision = await prynt.control.check({
useCase: "checkout",
visitorId,
accountId: user.id,
paymentMethod: { type: "card", fingerprint: cardKey },
});
Control answers allow, step-up or block with a reason code and a suggested step, and it never enforces anything by itself — your server decides. Start with the 14-day mirror period: decisions come back, nothing is blocked, and you can compare them with what your own rules would have done.
A policy that reads well: first appearance of a card key, allow. A second account on the same key without a verified email, step-up. A third account, a new device and a risky network, block the trial and still allow the paid checkout. The numbers are yours; the point is that they live in one place in your code.
Separate the innocent cases before you enforce
Card reuse is a signal, not a verdict. Households share a card, companies issue one card to a team, and resellers buy seats with a card that funds many tenants. Reissued cards keep the number and change the expiry, which changes your key if expiry is part of the string — decide deliberately.
Read the key next to accountsOnDevice, deviceTrust and your VPN and proxy signals. A card reused across five accounts that all pay is a different case from reuse that stops at the end of every trial. Pushing decisions to Slack through Prynt Automate, with {{trigger.decision}} and {{trigger.reasons | json}}, keeps review in front of a human.
FAQ

Does the same card on two accounts always mean abuse?
No. Shared household and corporate cards, resellers and reissued cards all produce reuse. Weigh the count with device and network signals and look at the outcome: paid conversion or trial-only activity.
Do I need to store the card number to detect reuse?
No. A processor token or a keyed hash over brand, last four and expiry is enough to count and compare, and the raw number never has to leave the component that already handles it.
Should I block on day one?
You can, but run the decisions alongside your current rules first and review the disagreements. Card reuse is cheap to detect and easy to over-apply.
Try it free
To see a checkout decision before you enforce anything, Control’s 14-day mirror period is the place to start. Create an account at pryntid.com, send one paymentMethod key on your next checkout, and read the decisions before acting on them. The device and network context that pairs with card reuse is in our device fingerprinting guide, and the request schema is in the docs.
Try it free
Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.