fraud · stripe · payments · checkout

Card Testing Attacks: How Small Sellers Stop Them

Learn how to spot a card testing attack, contain suspicious payments, and protect your checkout while keeping the path open for your legitimate buyers.

By cartt.io Team · November 10, 2026 · 11 min read

Card Testing Attacks: How Small Sellers Stop Them

You open your payment dashboard and find a burst of tiny failed charges with nonsense names. A card testing attack may be using your checkout to check stolen cards. Identify the affected payment route, refund confirmed fraudulent successes, and add protections in layers while watching for legitimate buyers caught in the filters. This is also called a carding attack.

What is a card testing attack?

A card tester submits stolen card details to see which ones work. A successful small payment or card-saving request can tell the attacker that a card is worth using elsewhere. Your checkout is the testing tool; the attacker doesn't need to want your product.

A declined payment is not proof of an attack by itself. A genuine buyer can mistype a card number or try another card. Look for a sudden pattern across attempts and compare it with your normal traffic.

How can you tell card testing from real sales?

Look in your processor's payment and request logs, not only your store's order list. Card-saving requests can be tested without creating an order. A spike in small failed or blocked payments, repeated fake addresses or emails, nonsense names, or a run of HTTP 402 errors with generic_decline can point to an attack. Check the route, timestamps, outcomes and any available IP or customer details together.

Fake orders at checkout can also appear as abandoned carts with junk data. Compare checkout starts with completed payments and investigate a sudden gap. Keep suspect sessions out of conversion and advertising analysis where your analytics tool allows it; don't delete legitimate orders to make the chart look better.

Check your abandonment recovery emails, too. Fake addresses can enter a flow and bounce. Pause or filter suspect entries, suppress bounced addresses, and use confirmed addresses and appropriate consent for marketing. A delay gives you time to filter suspicious entries, though a delay alone doesn't verify an address. Google recommends verifying subscriptions and removing repeatedly bounced addresses.

Why does card testing cost small sellers money?

The successful payments may need refunds. If cardholders report them first, you can face disputes and dispute fees. Stripe also warns that early fraud warnings, a high decline rate, and card-network monitoring programs can follow card testing. A high decline rate can hurt your standing with networks and make legitimate payments more likely to be declined. Continued problems can lead to account review or restrictions; closure isn't automatic. See why Stripe holds funds for the wider account risk.

Even failed attempts can have a cost. Stripe lists authorization fees under some custom pricing arrangements, and paid Radar plans can charge for screening. Your contract and the way an attempt is processed determine the bill. Check the fee detail rather than assuming every decline is free or that every request incurs the same charge.

What should you do in the first hour?

  • Find the route and pattern. Inspect failed, blocked and successful payments, API logs, and card-saving flows. Record the start time, affected endpoints and examples without copying card details.
  • Contain and refund. Check suspicious successful payments promptly, refund those identified as fraudulent before disputes arise, and report them as fraud. Don't fulfill suspicious orders while you investigate. Refunds don't guarantee recovery of processing fees or prevent every dispute.
  • Check protections and credentials. Confirm which checkout integration and controls are active. If there is evidence a secret key was exposed, rotate it and investigate its use. A public payment URL alone isn't evidence of a key leak.
  • Contact your processor and keep monitoring. Give support the timeline, representative payment IDs, affected routes and steps taken. Watch both the attack volume and legitimate payment failures after each change.

How do you prevent card testing? Layers that work together

Start with a current Stripe-recommended integration. Stripe says its latest Checkout and Payment Element integrations have automated card-testing controls, including flow limits, risk models and CAPTCHA triggers. Some blocked payments appear as “Blocked by Stripe.” These controls reduce exposure; they don't make a public payment flow immune. Compare Stripe checkout page options before changing your integration.

LayerWhat it stops or limitsCost to real buyersWho can set it
Recommended Checkout or Payment ElementAutomated controls flag and limit suspicious attemptsSome buyers may face a challengePlatform or developer
CAPTCHA on checkout and card-saving routesSome automated requests before card validationChallenges and accessibility friction; bots can still get throughDeveloper on a controlled form; processor on its hosted page
Rate limits and session checksExcessive requests from a source or sessionShared networks and repeat attempts can be caughtDeveloper or platform, where supported
Radar rules and 3D SecurePayments matching risk signals; authentication adds a hurdleFalse declines or an authentication stepStripe account administrator, subject to plan and integration
Minimum purchase or delayed captureRemoves some low-value targets or lets you cancel an authorization before captureA minimum excludes small orders; a hold affects available fundsSeller or developer, where the platform supports it

For your own form, validate CAPTCHA on the server and cover every request that can validate a card, including account pages that add a payment method. A client-side checkbox alone can be bypassed. Add rate limits around payment and setup endpoints as well as account creation. IP limits help, but shared networks and rotating addresses make IP a poor sole signal. Development time and any provider charges vary.

For access control, choose between friction on every attempt and friction triggered by unusual velocity. Requiring login or a validated session all the time is easier to reason about, but makes guest buyers do more work. An adaptive check preserves a shorter normal checkout, but needs a platform or developer able to detect and enforce the trigger. Test both the protection and the customer path. Follow the rest of your checkout page optimization work with the same attention to lost sales.

Use 3D Secure when the risk justifies asking the buyer to authenticate. A Radar rule can request it for elevated-risk payments; Stripe's adaptive 3D Secure is listed with Radar Plus and Pro. During an intense attack, requesting it more broadly may be appropriate while you contain the source. Authentication can add friction, isn't supported in every situation, and doesn't necessarily eliminate screening or authentication fees. Check your pricing and monitor outcomes.

If you ship physical goods later, manual capture can give you time to review an authorized order and cancel a suspicious one before capturing it. It does not stop attempts or remove all possible screening or authorization costs. Capture before the authorization expires: Stripe lists seven days for standard customer-initiated online Visa, Mastercard, American Express and Discover transactions, with different windows for other cases. Check the actual expiration and payment method. This is a poor fit for instant digital delivery, where access is usually granted as soon as payment succeeds.

What can you control on a hosted page?

On a hosted link or store builder, the provider controls its payment page and built-in challenges. You can check the processor dashboard, available rules, monitoring and catalog settings, and ask the platform which endpoints its protections cover. Your own site's firewall may not protect a separate provider-hosted URL.

A minimum order can make very small tests less attractive if your customers can accept it, but it won't stop an attacker willing to try a larger amount. Stripe supports a minimum for a customer-chosen price on a Payment Link; that isn't a universal minimum cart setting. Stripe's own minimum charge amount is a payment-processing floor, not your fraud policy. Check what your store builder can enforce.

Which common fixes fail?

Blocking one country may reduce a particular burst, but addresses can change and buyers travel. Block a region only when it fits your actual selling policy, and keep other controls in place.

A common tip is to rotate API keys whenever a public payment link is attacked. The link is designed to be shared, so rotating an uncompromised secret key won't make the link private. Investigate the route first; rotate promptly when exposure is suspected.

Deleting the cheapest item can interrupt a script aimed at that item, but another product can become the target. Similarly, a generic name is a signal to investigate, not proof of fraud. Check whether matching traffic belongs to a legitimate advertising or compliance crawler before blocking it. Finally, don't leave an unprotected card form at an alternate public URL or overlook an add-card endpoint when securing checkout.

How do you set up Stripe Radar rules without blocking real buyers?

Check existing default rules and the data your integration collects. CVC and postal checks are useful when available, but an issuer may return an unavailable result. A rule that treats every non-pass as fraud can reject legitimate buyers. Look at repeated card, IP and email activity together, and choose thresholds against your own normal traffic rather than copying a number from another business.

Backtest a proposed rule in the Dashboard. Start with a review rule when you're unsure, inspect matched payments and false positives, and narrow it before moving to a block rule. Review is not a silent test: matching payments can still be processed and successful ones enter a queue. During an active attack, use other containment measures while you evaluate. Stripe also offers traffic allocation when creating rules; start small and monitor the result.

Stripe lists custom rules and backtesting on Radar Plus and Pro. Radar Lite includes card-payment and card-testing protection at no extra charge for businesses on standard Payments pricing. Stripe's Radar pricing page lists the paid tiers as a fee per screened transaction, which is a payment that Radar evaluates; the amount depends on your country and tier. Ask Stripe how failed and blocked attempts are billed under your plan; a decline alone does not establish that screening is free. The exact fee, eligibility and treatment depend on your plan, country and contract, so inspect your Dashboard and agreement before upgrading. Screening is separate from any applicable payment or 3D Secure charges.

What if your API key leaked?

Treat an exposed secret key as a security incident. Rotate it in Stripe, replace it in the systems that need it, and investigate the request logs for unauthorized activity. Coordinate the replacement so your legitimate integration continues to work. Contact Stripe support with the timeline and what you contained.

Keep secret keys out of browser code and public repositories. Use restricted keys with only the permissions an integration needs, and store credentials in a secret manager or sensitive hosting environment variables with limited access. If a hosting provider reports a breach that may have exposed credentials, investigate and rotate affected keys. Publishable keys are intended for client-side use; secret keys are not.

If the incident produced unexpected fees, provide support with the timeline, affected payments and remediation and ask whether relief is available. A waiver is a discretionary request, not a guarantee. For the next incident, keep the response alongside your plan for how to prevent chargebacks.

Frequently asked questions

Does a failed payment cost me money?

It can. Stripe documents authorization fees for some custom pricing and paid Radar screening fees. Whether a particular failed attempt is billed depends on your agreement and how it was processed. Check your fee report and ask support about entries you can't reconcile.

Will card testing get my Stripe account closed?

Not automatically. Stripe warns that card testing can damage network standing and lead to monitoring. Persistent fraud or related risk can lead to review or restrictions, so document the incident, contain it and communicate with Stripe.

Should I turn on 3D Secure for every payment?

A broad rule can be a temporary response to an active attack if the extra friction is acceptable. For ongoing use, requesting authentication on higher-risk payments can protect more normal checkouts from unnecessary challenges. Neither approach replaces endpoint protection or guarantees that attempts are free.

Can I add CAPTCHA to a Stripe Payment Link?

You can't insert your own CAPTCHA into Stripe's hosted payment page. Stripe controls that page and its protections; check with Stripe about the controls available for your link. You can place a verification step on a page you control before revealing a link, but anyone with the direct link may still reach it. Monitor the link and use the dashboard controls available to your account.