You wake up to a pile of failed orders for your cheapest item. A day or two later your decline ratio has climbed and your processor's risk team has emailed. Whether the next message is a document request or a reserve notice depends on what you can show them. Card testing is how a research-use-only store can lose a merchant account over traffic it never wanted.
This is not only a big-retailer problem. Signifyd's 2026 State of Fraud report (August 2026) found card-testing attacks up 175% year over year across its commerce network of thousands of online merchants. Bots do not care what you sell or how much volume you do. They care whether your checkout answers "yes" or "no" when they try a stolen card.
At Peptide Marketing HQ, we hold one view about this. A store can fix the checkout page and still lose the account if the processor never sees a remediation packet. Closing the hole stops the attack. Proving it to your acquirer protects the account. The job has four parts.
Spot the attack in your own order list. Contain it in the first hour. Close the WooCommerce endpoints that stay open. Then hand your processor a documented remediation packet.
Throughout, your product claims, labels and processor application have to stay honest and consistent, because the fix must not create the next underwriting problem.
What card testing looks like in your store
Card testing is the automated use of stolen card details to find out which ones still authorize. Attackers send many small or authorization-only attempts, then reuse or sell the cards that work. Your store is just the test bench.
In the WooCommerce order list
Open WooCommerce > Orders first. Then check your gateway's transaction log and the server access log. You are looking for patterns, not one odd order:
- Bursts of failed or pending orders within minutes
- Tiny order totals, or orders for the cheapest item repeated
- Throwaway email addresses and names that do not match the billing details
- The same few IPs or user agents repeating across many attempts
In the processor dashboard
Look for a sudden jump in authorization attempts and a decline ratio that moves sharply. Watch for many different cards from one device or IP, and issuer decline codes clustered together. Rising authorization volume, more false declines and spikes clustered around a small set of IPs, devices or browser fingerprints are the classic signs.
You need your own baseline to read these signals. Pull the last 30 days of gateway logs and count failed attempts per hour. Find your hourly p95, the count you only exceed in 5% of hours, and note your usual decline ratio. Then set two alerts:
- Failed attempts in any 10-minute window exceed 5 times your trailing 30-day hourly p95.
- More than 3 distinct cards hit one IP, device or session within an hour. Start at 3 and tune it against your real customers.
Either alert firing is your cue to treat it as an attack and move to containment.
How it differs from normal declines
A real shopper retries the same card a few times after a typo or an insufficient-funds decline. A tester cycles through many card numbers quickly and shows no browsing behavior before checkout. Account takeover targets existing customer logins, and BIN attacks guess numbers inside a card range. Velocity alone can mislead. It means much more when it comes with shared infrastructure and many different BINs.
So why are you getting hundreds of failed card payments? Check for these five tells:
- Many different cards from one IP, device or session
- Very small amounts, often the same one repeated
- Disposable emails and mismatched names
- A decline ratio far above your normal
- No product browsing or cart behavior before the attempts
Why does card testing hurt high-risk merchants most?
A high-risk merchant already sits under closer watch. Decline ratios, fraud monitoring and account reviews all have a thinner margin for error. A flood of failed authorizations that a low-risk store absorbs can become a processor conversation for you.
Can it freeze funds or trigger a review?
Yes, and for a high-risk merchant it is a realistic outcome. The network programs that matter are Visa's VAMP (Visa Acquirer Monitoring Program) and Mastercard's Excessive Chargeback Program, which flags merchants as Excessive Chargeback Merchants.
VAMP counts fraud reports and disputes against your settled transactions. As of 2026, it rates a US merchant Excessive at a 1.5% ratio with at least 1,500 fraud reports and disputes a month (lowered from 2.2% on April 1, 2026). Mastercard's entry tier, the Excessive Chargeback Merchant, starts at 100 chargebacks and a 1.5% ratio in a month. Our guide to chargeback prevention for peptide stores covers both programs in more detail.
Both networks revise these figures, so confirm the current ones and the tier you sit in with your acquirer.
Failed test attempts do not count as chargebacks. The damage comes later. Any stolen card that does authorize can come back as a fraud report or dispute, and that is what pushes your ratio toward a threshold. Your acquirer answers for your ratios, so the pressure gets passed down to you as a risk review, a document request, a fee increase or a reserve.
A rolling reserve means the processor holds a set percentage of each day's sales for a set number of days, then releases it on a schedule. Your agreement sets the percentage and the window. Stripe and PayPal each run their own risk reviews and can limit an account or hold funds under their terms, so read those terms before an attack, not after.
The processor's decision usually turns on what you can put in front of it. A store with logs, a timeline and closed endpoints gives the risk team something to approve. A store with none of that gives them only the ratio. That is why the packet near the end of this article matters as much as the fixes.
Switching gateways will not fix it
If your checkout stays open to automated retries, the same traffic follows you to the new route. A backup route protects you when one processor pauses you. It does not stop bots. Our peptide payment processing work pairs a backup route with a hardened checkout for exactly this reason.
If card testing has already cost you a processor, our guide to the best high risk payment processors for peptide brands covers what to check before you sign with the next one.
Keep the store honest
Your product claims, labels, checkout copy and processor application must all say the same thing. We never help anyone hide product types from processors, and we never help dress up human-use sales as research. If a remediation depends on concealment, it is not a remediation. Card testing is a payment-security problem, and it gives no one permission to make a prohibited business look acceptable.
Use the first 60 minutes to contain, then clean up
Contain first. Investigate second. Every extra minute of open checkout means more attempts, more fees and a worse ratio.
Contain
- Turn on a challenge for all guest checkouts.
- Tighten rate limits at the edge and in the application.
- Pause any open card-add or payment-method endpoints while you investigate.
Preserve evidence before it rotates
Logs roll over, and dashboards change. Export these now:
- Gateway transaction logs
- WooCommerce order and customer records
- Web server and WAF logs
- Timestamped screenshots of the processor dashboard
Check fulfillment and refunds
Hold anything that looks suspicious. Review any successful test authorizations. Do not refund a test payment and assume the incident is closed. A refund does not guarantee a favorable chargeback outcome.
Call the processor early
Bring facts, not guesses. Say what you saw, when it started and what you have already changed. Then ask what evidence they want. A merchant who calls first with a timeline reads very differently from one the processor has to chase.
The layers that stop card testing at checkout
No single control carries this. On a WooCommerce store, each layer maps to a real tool or setting:

- Edge or WAF: a Cloudflare rate-limiting rule on POST requests to
/?wc-ajax=checkout, with a managed challenge or block once one IP passes the limit you set from your baseline. - Bot challenge: a Cloudflare Turnstile plugin on the checkout form, set to fire when a session looks risky.
- Application rules: limits on attempts by IP, device, email and billing details.
- Gateway fraud tools: AVS and CVV rules in your gateway settings, plus any rule engine it offers.
- Issuer authentication: EMV 3-D Secure where it fits.
- Processor monitoring: alerts on attempt volume and decline ratio, sent to a real person.
For a store under $100k a month, two layers matter most. The first is the edge rate limit on the checkout endpoints, because it stops the volume before WordPress or your gateway sees it. The second is the bot challenge, especially on card-add. Get those two live this week, then work down the list.
Challenge by risk, hard on card-add
Should CAPTCHA run on every checkout? We say no. Use a risk-based challenge on checkout, with hCaptcha or Cloudflare Turnstile firing when a session looks off. Put a hard challenge on card-add, payment-method update and any validation endpoint. Those endpoints matter as much as the checkout page. CAPTCHA alone does not stop card testing. Current guidance treats it as one layer.
Limits and quiet failures
Set rate limits and velocity rules by IP, device, email and billing details. Pair them with less revealing decline messages. A bot that learns "wrong CVV" can tune one field at a time. Return one generic message instead.
Add a minimum order value, but pick the number with your own order data. Testers favor the cheapest item. If your typical order sits well above your cheapest item, a sensible minimum or a bundled shipping line removes the cheap test target without hurting real buyers. It does not close authorization-only endpoints, so close those first.
Gateway tools and BIN blocking
Use AVS and CVV rules intelligently, and add 3-D Secure where the extra step will not cost you real orders. Be careful with BIN blocking. A blunt block can decline legitimate buyers who share that range. The same goes for blocking a whole country because one attack used foreign infrastructure. Block narrowly, review often, and watch your false-decline rate after every rule change.
Close the doors attackers actually use
Many stores protect the checkout page and leave the AJAX route open. Bots do not need your page. They call the endpoint behind it. Walk through this WooCommerce checklist:
- Checkout and
wc-ajaxcalls - REST and Store API routes
- Saved cards and payment-method updates
- Subscriptions and renewal endpoints
- Gift-card or balance checks and coupon validation
- Guest checkout and account creation
- Webhooks and any custom API route
Tokenize and check server-side
The card form should hand data straight to the gateway. Your gateway's hosted fields keep card numbers off your server. Use unique, server-generated transaction references, and protect your API keys. Add server-side limits on payment attempts, tokenization requests and session-key creation too. Before installing any payment plugin, check who maintains it, how it handles tokens, and whether it exposes endpoints without limits.
Keep it maintained
A WooCommerce or gateway plugin update can add a new route that your rate-limit rule does not match. A Turnstile plugin update can reset its settings or change when the challenge fires. Either one quietly re-opens the door you just closed.
The fix is a routine: after every update, rerun the staging test below and confirm your edge rule still matches the live endpoints. That upkeep is the work our website care plans are built around for WooCommerce stores.
Test your own setup safely
Never run live attack loops against your own gateway. Use a staging copy and your gateway's test mode instead:
- Confirm your challenge fires on a fresh session.
- Send repeated requests to a protected endpoint and confirm you get throttled.
- Check that decline messages stay generic.
- Confirm that alerts reach a real person.
Build a remediation packet for your processor
Processors and acquiring banks want proof, not reassurance. Put these items in one document:
- An incident timeline with start, detection and containment times
- Attempt and approval counts
- Decline ratio before and after
- IP and device clusters
- Affected BINs, if you can see them
- A list of controls added, with dates
- Evidence that successful transactions were reviewed
Show that it stays fixed
Add your monitoring: alert rules, who reviews them and how often. A processor wants to know the problem stays fixed, not that it stopped once. Name an owner and a review schedule.
Disclose exactly what you sell
Describe your products, intended use and policies exactly as they appear in your application and on your site. Concealment is never part of a remediation, and we do not help with it. A packet that matches your claims, labels, refund policy and checkout copy builds trust. One that contradicts them ends the conversation.
If you want a second pair of eyes on your logs before you send this to your processor, we'll review it in the free store audit.
Frequently asked questions
Why am I getting hundreds of failed card payments on my WooCommerce store?
Most likely, bots are testing stolen card numbers against your checkout or an open AJAX or REST endpoint. Look for many cards from one IP or device, tiny amounts, throwaway emails and no browsing before checkout. Contain it first, then preserve your logs.
Can card testing make my processor freeze funds or end my account?
It can lead to reviews, reserves, fee increases or tighter monitoring, and processors differ. We cannot promise any outcome. Check your processor agreement for exact terms, and contact them early with a timeline of what you saw and changed.
How do I tell card testing from a customer retrying a declined card?
A real customer retries the same card a few times, usually after browsing. A tester cycles many card numbers quickly with no shopping behavior. Combine signals such as shared IP or device, varied BINs and fast velocity, rather than judging one attempt.
Should CAPTCHA run on every checkout or only suspicious sessions?
We recommend a risk-based challenge on checkout so real buyers face less friction. Use a hard challenge on card-add, payment-method update and validation endpoints. CAPTCHA is one layer and should sit beside rate limits, device signals and gateway rules.
Can I block a BIN, IP or country without causing false declines?
Not without some risk. Block narrowly and for a limited time, then check your false-decline rate after each change. A whole-country block because one attack used foreign infrastructure usually costs more real orders than it saves.

