What Are Velocity Checks? Fraud Limits That Stop Card Attacks
Velocity checks for fraud cap how often a card, IP, email or device can try to pay. See which rules to set and starter limits that spare real buyers.
On this page 6 sections
Key takeaways
- Count distinct cards per device, email or IP instead of raw attempts. Real customers retry one card, and testers cycle through many.
- Per-card limits barely touch card testing because testers try each stolen card once or twice.
- A site-wide decline alarm catches spread-out attacks that stay under every per-identifier limit.
- Before you rely on an IP rule, confirm your gateway records the customer's IP address and not your server's.
Velocity checks are fraud rules that count how many payment attempts one card, IP address, email address or device makes within a set time window, then block or challenge the attempt that crosses your limit. They are the main defense against card testing, in which a bot runs hundreds of stolen card numbers through your checkout to learn which ones still work.
Two settings do most of the work. First, limit the identifiers an attacker can't swap cheaply, and count distinct cards instead of raw attempts. Second, add a store-wide decline alarm to catch whatever slips under those limits. Set up that way, velocity rules cut a bot run short and stay out of the way of a real customer retrying a declined card.
How Velocity Checks Stop Fraud
Every velocity rule has four parts: an identifier, a count, a time window and an action. Here is one example: "If one device uses more than three different card numbers in 24 hours, block the next attempt." A store-level rule reads differently: "If checkout sees more than 15 declines in 10 minutes, require a CAPTCHA on every order."
These limits are your settings, not card network rules. Visa and Mastercard don't tell merchants what per-IP limit to use, and no federal law sets one. The rules live in your payment gateway, your e-commerce platform, a fraud tool or a plugin, and someone on your side has to turn them on. A store that never touched its settings is running whatever came with the account. Look at those defaults before you assume you're covered, because a new account has no guarantee of including any velocity rules.
Ignoring an attack costs you in fees and in risk. Every attempt passes through your gateway. If your pricing includes per-transaction gateway or authorization fees, declined tests cost money too, and our article on whether declined transactions cost you money walks through the line items. On the risk side, approved test charges come back weeks later as fraud chargebacks. A sudden wall of declines also draws your processor's risk team, which is how reserves and account reviews start. For the full anatomy of an attack, read what card testing is and how fraudsters use your checkout.
Common Velocity Rules: Card, IP, Email and Device
Each identifier catches a different kind of abuse, and attackers pay a different price to change each one.
Per card. This rule limits attempts on a single card number, such as three tries per hour. It catches someone guessing expiration dates or CVV codes on one card. Many merchants assume it stops card testing, and it barely touches it. Testers try each stolen card once or twice and move on, so a per-card limit of three never fires against an attack that uses 300 different cards one time each.
Per IP address. Here the count is attempts, or distinct cards, from one IP address. Attackers rotate IPs through proxy networks, which weakens the rule, and shared addresses cause false positives. Mobile carriers put large numbers of customers behind shared IPs, and office networks do the same. On a store with heavy mobile traffic, a tight IP limit blocks real buyers who happen to share a carrier address.
Per email. Capping distinct cards or attempts per email address stops lazy attacks and people cycling cards on a real account. A bot that invents a fresh email for every attempt walks right past it.
Per device. A device fingerprint is a profile of browser and hardware traits built by a script on your checkout page. Capping distinct cards per fingerprint is one of the strongest single rules, because spoofing a fingerprint takes more work than rotating an IP or an email. It only works if your fraud tool or gateway script actually runs on the checkout page.
Site-wide. This rule tracks no identifier at all. It watches total declines, or total attempts, across the whole store in a short window, and it catches distributed attacks that stay under every per-identifier limit. Look for it by name in your settings, since it's the rule merchants forget.
Example: Stopping 300 Attempts in 10 Minutes
Every number in this example is illustrative, not drawn from a real attack. Picture a nonprofit's donation form hit with 300 attempts in 10 minutes, each a $1 donation on a different stolen card. The bot rotates through 40 IP addresses and 12 device fingerprints and generates a new email every time. Assume 70% of the test cards decline and that the form normally sees two declines an hour.
In the table, each rule acts alone, and the attempts are spread evenly across IPs and devices:
| Rule (illustrative limit) | How the attack looks to it | Attempts that get through |
|---|---|---|
| Per card: 3 per hour | Each card used once | All 300 |
| Per email: 3 per hour | Each email used once | All 300 |
| Per IP: 5 per hour | 300 ÷ 40 = 7.5 per IP | 40 × 5 = 200 |
| Per device: 5 per hour | 300 ÷ 12 = 25 per device | 12 × 5 = 60 |
| Site-wide: 10 declines in 10 minutes | 15 attempts × 70% = 10.5 declines | 15 on average before the trigger fires |
Rotating cards and emails defeats the per-card and per-email rules completely. Against the IP limit, two-thirds of the run still gets through. Device limits cut the damage to 60 attempts, and stacking the IP rule on top changes nothing, because the device rule is already the tighter of the two. Only the site-wide alarm stops the run early. Once it fires, it forces a CAPTCHA or a pause on every checkout until someone looks.
Build your rules around the identifier the attacker can't rotate cheaply, and keep an aggregate alarm behind them. A store with nothing but per-card limits has close to zero protection against card testing.
Setting Velocity Limits Without Blocking Good Customers
Neither the card networks nor any regulator sets these thresholds, so treat the table below as a starting point and tune it against your own logs. Your most important choice is to count distinct cards per identifier wherever your tool allows it. A real customer whose card declines retries it once and then reaches for a second card. A tester cycles through dozens.
| Identifier | Illustrative starting limit | Suggested action |
|---|---|---|
| Card number | 3 attempts per hour, 5 per 24 hours | Block |
| Device | 3 distinct cards per 24 hours | Block |
| 3 distinct cards per 24 hours | Hold for review | |
| IP address | 5 distinct cards per hour | CAPTCHA or 3-D Secure challenge |
| Site-wide declines | 3 to 5 times your normal hourly declines, within 10 minutes | CAPTCHA on all checkouts and alert staff |
IP gets a challenge instead of a block because of shared addresses. A real buyer behind a carrier IP solves a CAPTCHA in seconds, while a bot running at speed stalls on it.
Legitimate buying looks like velocity in every kind of business, just in different shapes:
- Digital products and services. Team buyers purchase several licenses or gift codes in one session. Delivery is instant and there's no shipping address to check, which is exactly why testers target these stores. Keep the device and distinct-card limits tight, and exempt logged-in accounts with past successful orders from the attempt-count rules.
- Nonprofits. Donation forms take small amounts and require no login, so they attract testers. Giving days bring real spikes. For that day only, raise the site-wide trigger and leave the device rule in place. Watch for low-dollar gifts from cards with no history.
- WooCommerce stores. Checkout is one part of a WordPress site, so rules end up split between plugins and your gateway's filters. Decide which layer owns each rule so two plugins don't double-block the same customer. MannyPay supports WooCommerce payment integrations if you're rebuilding that stack.
- B2B buyers. Purchasing teams place several orders from one office IP with company cards. Put their accounts or email domains on an allow list rather than loosening the IP rule for everyone.
If your tool has a review or log-only mode, run new rules in it for a week. Count how many real orders would have been stopped, then turn on blocking.
Where to Configure Transaction Velocity Rules
Velocity rules live in four places, and most stores use two or three of them.
Your payment gateway. Gateways ship with transaction filters whose names and options differ by gateway and plan. If you use Authorize.Net for online payments or the NMI gateway, check which velocity and IP filters your account includes and what action each one takes. Our overview of payment gateways shows where the gateway sits between your site and the processor.
Your e-commerce platform. Hosted platforms run their own bot protection on checkout, and you control little of it. Self-hosted platforms such as WooCommerce lean on plugins and server-level rate limiting.
A dedicated fraud tool. Paid tools add device fingerprinting and signals pooled across their merchant base. Our guide to choosing fraud prevention tools for an online store compares the categories.
Your web server or CDN. Rate limits on the checkout URL stop the crudest bots before they reach the payment step.
Check one setting before you trust any IP rule. If your site sends transactions to the gateway from its own server, the gateway records your server's IP on every order unless the integration passes the customer's IP in a separate field. In that setup, a per-IP rule either blocks your whole store at once or never fires. Place a test order from your phone and compare the IP in your gateway report with the phone's address.
Velocity Checks Alongside Other Ecommerce Fraud Filters
Velocity rules count behavior. Whether a card belongs to the person using it is a separate question, so pair them with filters that inspect the transaction itself:
- CVV and AVS checks compare the security code and billing address with the issuer's records. Our guide to AVS and CVV settings that stop fraud without lost sales covers which response codes to decline.
- 3-D Secure sends risky transactions to the card issuer for authentication. Velocity rules make a good trigger for it: challenge only the sessions that trip a soft limit. MannyPay offers 3-D Secure authentication for online merchants.
- CAPTCHA on checkout or donation forms slows bots, and it costs real customers least when it switches on only after the site-wide alarm fires.
- BIN checks read a card's first digits to identify who issued it and in which country. They help when testers cluster on prepaid cards or on issuers outside your market. See what a BIN number tells you.
If an attack is already running, follow the response steps in how to spot a card testing attack and shut it down. For rule settings beyond velocity, see our walkthrough of fraud rule settings that stop bad orders.
This week, export the last 30 days of transactions from your gateway. Find the highest number of distinct cards any single identifier used, along with your busiest hour for declines. Set each limit just above those peaks and run the rules in log-only mode for a week before you switch on blocking. If your current gateway can't count distinct cards or record customer IPs correctly, request a free quote. A MannyPay specialist will respond within one business day to discuss your gateway options.
This article is for general educational purposes and is not legal, tax or financial advice. Card network rules and state laws change; confirm current requirements for your business.


