Velocity Checks and Fraud Rules: Settings That Stop Bad Orders
Set velocity checks in your gateway, pick block or review for each rule, and stop card testing without turning away the good customers you want to keep.

Photo: Pixabay
On this page 6 sections
- Velocity Checks Are Your Settings, but Visa Now Counts the Results
- The Core Velocity Checks and What Each One Catches
- Worked Example: Setting a Cards-Per-IP Velocity Limit
- Block, Review or Allow: A Test for Each Rule
- Pair Velocity With Address, Location and Card Rules
- Tune the Rules Monthly and Know When They Stop Working
Key takeaways
- If your integration doesn't send the customer's IP address, an IP velocity filter has nothing to count, even though it shows as turned on.
- Stripe's velocity counts leave out the payment being screened, so "> 3" blocks the fifth attempt, not the fourth. Read how each field is defined before you pick a number.
- Give each velocity rule two limits. The low one sends orders to review and the high one blocks them. A review queue someone clears every day often costs less than declining good customers.
- Visa sets no velocity number. Under VAMP, though, acquirers have to take proactive steps against card testing, so your processor will care how well your rules work.
Velocity checks are fraud rules that count how often something happens in a set time window and decline or hold the payment once the count passes your limit. Common examples are charges on one card number in an hour and attempts from one IP address in an hour. Card testing and scripted stolen-card orders only work at volume, which makes velocity checks the first line of defense against both. You choose the numbers, so set them from your own order history, launch new rules in review mode before switching to block, and confirm exactly what your gateway counts before you trust any threshold.
Velocity Checks Are Your Settings, but Visa Now Counts the Results
Velocity limits are tools your gateway gives you. Authorize.net's fraud tools include a Daily Velocity Filter and an Hourly Velocity Filter, which cap total transactions on the account. A third, the Transaction IP Velocity Filter, caps transactions from one IP address per hour. Authorize.net's help pages say the Daily Velocity Filter is free to every merchant, including those who don't subscribe to its Advanced Fraud Detection Suite (AFDS). Stripe's Radar Rules 101 guide points to attributes such as authorized_charges_per_card_number_hourly and authorized_charges_per_ip_address_hourly for high-velocity fraud. Its card-testing example blocks a payment when total_charges_per_ip_address_hourly is greater than 1. Neither Visa nor Mastercard tells you to allow three attempts or five, and no federal law sets a velocity number.
The card networks control what happens when your screening falls short. According to Checkout.com, Visa retired its separate fraud and dispute monitoring programs on April 1, 2025, and replaced them with the Visa Acquirer Monitoring Program (VAMP). NMI's timeline says the VAMP ratio applied from June 1, 2025, and the advisory period ended October 1, 2025. VAMP tracks two numbers:
- The VAMP ratio. Visa's fact sheet defines it as fraud reports (TC40s) plus disputes (TC15s), divided by settled transactions (TC05s). Relyon's reading of the fact sheet adds that only card-not-present sales count. In the U.S., a merchant is measured only after it reaches 1,500 or more fraud reports and disputes combined in a month. That minimum is a count of fraud reports and disputes, not of sales. Visa's 2025 fact sheet put the merchant "Excessive" line at 220 basis points (2.2%). Beast Insights reports that the line dropped to 1.5% on April 1, 2026, for U.S. merchants.
- The enumeration ratio. This is card-testing authorizations, approved and declined, as a share of all authorizations. A merchant is flagged when that share reaches 20% and the month includes 300,000 or more enumerated authorizations.
Few small stores will ever see 300,000 card-testing attempts. For them the sentence that matters in Visa's fact sheet is a different one: VAMP requires acquirers to take proactive steps to keep merchants from exceeding the enumeration thresholds. After an attack, expect your processor to ask what controls you have in place. Velocity checks are usually the first answer.
The networks also decide who pays for fraud. On card-not-present sales, the merchant generally absorbs fraud chargebacks unless the sale was authenticated in a way that shifts liability to the issuer, as our explainer on 3-D Secure and the liability shift covers. Visa has already lowered the merchant threshold once, so confirm current program terms and fees with your processor.
The Core Velocity Checks and What Each One Catches
Card testing is the clearest target. A bot runs a flood of small authorizations to find card numbers that still work. Depending on your processor's pricing, each attempt can carry its own fee, as our piece on whether declined transactions cost you money explains. If an attack is happening now, go straight to our guide to spotting and shutting down card testing.
| Velocity rule | What it counts | What it catches | Sensible first action |
|---|---|---|---|
| Attempts per card number | Authorizations on one card in an hour | A stolen card retried at different amounts | Block above your limit |
| Transactions or cards per IP | Attempts or distinct cards from one IP address in an hour | One bot cycling through a card list | Review at a low limit, block at a high one |
| Cards per email or customer | Distinct cards tied to one account | Fraudsters swapping cards on a single profile | Review |
| Account-wide hourly or daily count | All transactions on your account | Attacks spread across many IP addresses | Alert or hold for review |
Merchants often skip the account-wide count. An attacker who spreads attempts across many IP addresses can stay under every per-IP limit, but the total still shows. Watch for an overnight spike far above your normal hourly volume.
Four Setup Details That Quietly Break Velocity Checks
The IP rule needs the IP. Authorize.net's help page for the Transaction IP Velocity Filter says merchants on its Advanced Integration Method (AIM) must send the customer IP address (x_customer_ip) with each transaction to use the filter. If your integration leaves that field blank, the filter has no address to count, even though the settings page shows it as enabled. The same page says Authorize.net leaves AOL IP address ranges out of the filter because many customers can arrive from the same AOL addresses. Shared IPs are a known problem, and the gateway designed around them.
The count may not include the current payment. Stripe's guide says Radar's frequency rules leave out the payment being processed. So a rule that blocks when authorized_charges_per_ip_address_hourly is greater than 3 lets the fourth attempt through and stops the fifth. Portreeve's analysis of Radar adds two quirks. The time windows are bucketed, so "hourly" can stretch to 3,900 seconds. Some counters also stop at 25, which means a "> 25" rule on those counters never fires. Authorize.net's filter takes a different input: the maximum number of transactions allowed per IP per hour. Also check whether an attribute counts every attempt or only approved charges. Card testing produces a lot of declines, and a count of approvals alone can miss them.
The action may come after authorization. Authorize.net's IP velocity filter offers three actions: authorize and hold for review, do not authorize but hold for review, or decline. With the first option, the issuer has already approved the charge. If you then reject a real customer's order, that customer sees money tied up until you void the charge. When one transaction trips several filters set to different actions, Authorize.net applies the most severe one.
Recurring billing can trip it. Zoho was updating its Authorize.Net integration for Zoho Checkout on November 20, 2024, and warned users that the IP velocity filter could hold or decline legitimate recurring payments. Its advice was to disable the filter. If a subscription platform sends renewals through your gateway account, test how renewals interact with the filter before you tighten it.
Worked Example: Setting a Cards-Per-IP Velocity Limit
Every figure below is illustrative. The numbers were chosen to show the method and the arithmetic and are not benchmarks.
An online store adds a rule that flags any IP address that submits more than two different cards in an hour. Before switching the rule to block, the owner runs it in review mode for a month. It flags 100 orders. A manual check finds 30 fraud attempts and 70 good orders. The good orders come from an office network shared by dozens of employees, a parent buying for several kids, and customers who tried a second card after a decline.
Assumptions:
- Each good order is worth $32 in gross margin.
- Each fraud order that gets through costs $100, counting lost goods, shipping and the processor's chargeback fee.
- Reviewing one flagged order takes 3 minutes of staff time at $20 an hour.
If the rule blocks: it prevents 30 × $100 = $3,000 in fraud losses and gives up 70 × $32 = $2,240 in margin. That nets $760 a month, before counting the repeat purchases those 70 customers would have made.
If the rule sends orders to review: 100 orders × 3 minutes = 300 minutes, or 5 hours, which costs 5 × $20 = $100. If reviewers stop the fraud and approve the good orders, the rule is worth about $3,000 − $100 = $2,900 a month. Changing only the action makes the same rule worth almost four times as much.
That points to a two-tier limit. A low threshold sends orders to review. A much higher one blocks outright, and it should sit above anything a real customer reached in your order history. An IP that tries a dozen cards in an hour is a script. An IP that tries three is often a family.
Block, Review or Allow: A Test for Each Rule
The most common mistake is setting every rule to block. A better test is to block a pattern only if a person looking at the order would almost never ship it.
Block what a reviewer would reject on sight: velocity spikes far above your history, card security code (CVV) mismatches and cards on your own blocklist.
Review the judgment calls: moderate velocity hits, first orders above your usual amount, location mismatches and reshipper addresses.
Allow and log returning customers with clean histories, plus any rule you're still testing.
Review only works if someone clears the queue every day, because an order left sitting is often a lost sale. Digital goods need tighter limits. Delivery is instant, so you have no window to catch a bad order, and a fraudster loses nothing by trying again. Stores that already use 3-D Secure authentication at checkout have a middle option: send risky orders through an authentication challenge and let low-risk ones skip the extra step.
Pair Velocity With Address, Location and Card Rules
Velocity rules catch volume. The next layer catches single orders that look wrong. Authorize.net's transaction filters include an IP-Shipping Address Mismatch Filter, which compares the IP address an order came from with its shipping address. There is also a Regional IP Address Filter that flags orders from regions or countries you choose.
Set amount limits from the range of your own past orders rather than a round number. Send IP-location mismatches to review, because travelers, VPN users and gift buyers trip them all the time. The Address Verification Service (AVS) gives noisier results than CVV. A partial AVS match often just means an apartment number was typed differently from the bank's record.
The bank identification number (BIN) is the first several digits of a card number. It shows the issuing country and whether the card is prepaid, so a prepaid card on a high-value digital order is a reasonable trigger for review.
Tune the Rules Monthly and Know When They Stop Working
A short routine keeps rules useful:
- Launch every new rule in log-only or review mode for two weeks.
- Tag each chargeback with the rule that should have caught it, or note that no rule would have.
- Tag each false decline with the rule that caused it. Buyers who retried with another card or emailed to ask why an order failed were probably real.
- Revisit thresholds monthly and after any promotion or new market.
- Delete rules that never fire.
Track your chargeback ratio alongside the rules. Our breakdown of chargeback ratio thresholds explains how close is too close. Extra friction from fraud screens also shows up as abandoned carts, which our guide to payment fixes for checkout abandonment covers.
WooCommerce stores should watch for overlap. Screening can run in the gateway and in a plugin at the same time. Both can then decline the same order, with no clear record of which tool did it. Pick one tool to do your main screening. Our WooCommerce integration page shows how the gateway connects.
Fixed rules struggle once fraudsters space out their attempts or rotate through clean home IP addresses. These are signs you've reached that point:
- No one can explain why half the rules on your list exist.
- The review queue gets longer every month.
- Chargebacks keep rising even as you add rules.
- Fraud comes in through account takeover of customer logins, which velocity rules can't see.
Machine-learning screening has blind spots of its own, which our look at what AI fraud detection catches and misses lays out. Our guide to choosing fraud prevention tools for an online store compares the options.
This week, open one recent transaction in your gateway and check that the customer IP address was recorded. If the field is blank, your IP velocity checks have nothing to count, and that is the first thing to fix. Next, export last quarter's declines and chargebacks and make sure you can tell which rule acted on each one. MannyPay supports Authorize.net and NMI among its payment gateways for online payment processing. Request a free quote and a MannyPay specialist will respond within one business day.
Sources
- Authorize.net Help: Transaction IP Velocity Filter
- Authorize.net Help: Transaction Filters
- Authorize.net Support: What Is the Advanced Fraud Detection Suite (AFDS) and How to Use It
- Zoho: Authorize.Net Update: Disable Transaction IP Velocity Filter to Avoid Payment Failures
- Stripe: Radar Rules 101
- Portreeve: Stripe Radar: What It Catches, What It Misses, What to Add
- Visa Acquirer Monitoring Program Fact Sheet (2025)
- Checkout.com: Visa Acquirer Monitoring Program (VAMP) Explained
- NMI: VAMP: What You Need to Know About Visa's Acquirer Monitoring Program
- Relyon: Visa Acquirer Monitoring Program (VAMP)
- Beast Insights: VAMP Thresholds
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.


