Fraud prevention and chargeback tools for card-not-present payments (rules + ML + dispute workflows)

Card-not-present payments need two controls: stop bad transactions before authorization, and make chargeback handling cheap when one still gets through. Stripe does both with Radar machine learning, custom fraud rules, dispute prevention tools from Verifi and Ethoca, and guided dispute workflows in the Dashboard. The goal is straightforward: protect margin without adding checkout friction.

Why card-not-present payments need layered fraud controls

Online card payments fail in predictable ways:

Fraudsters test cards at scale.
Legitimate customers create disputes when descriptors are unclear.
First-party misuse shows up as “I don’t recognize this charge.”
Manual review becomes a bottleneck once volume rises.

That is why a single control is never enough. For card-not-present payments, you need a stack that works at three points in the flow:

Stage	Stripe tool	Job
Before authorization	Radar ML + custom rules	Block, allow, or review risky payments
Before chargeback	Verifi + Ethoca dispute prevention	Surface transaction data so cardholders can recognize legitimate charges
After chargeback	Guided disputes + Smart Disputes	Collect evidence and submit it efficiently

Screen risky transactions with Radar machine learning and rules

Stripe Radar uses machine learning trained on data from the Stripe network to score transactions and surface fraud patterns quickly. For card-not-present payments, that matters because you are not relying on a chip, tap, or signed receipt. You are relying on signals.

Use Radar to do three things:

Block obvious fraud
Review ambiguous payments
Allow trusted buyers through

Then add custom rules in the Dashboard to encode your policy.

Common rule inputs include:

Country or region mismatch
Velocity spikes
Repeated failed attempts
New account + high-value order
Shipping and billing inconsistencies
Risky device or IP patterns
Repeat disputes or prior bad behavior

That combination is the core of effective fraud prevention for online card payments:

ML finds patterns you would miss manually.
Rules let you enforce business-specific controls.
Risk scores help you route transactions instead of treating every payment the same.

For teams with real volume, this is where you protect both conversion and margin. Overly strict screening kills legitimate orders. Overly loose screening creates chargebacks. Radar gives you the middle path: automate the obvious, review the gray area, and keep policy in your control.

Pricing note

Stripe’s fraud tooling follows a transparent pricing model. Some fraud protection features are included with Payments on standard pricing, and some advanced configurations use per-screened-transaction pricing on custom plans. If you need platform-wide controls or non-standard pricing, contact sales.

Reduce disputes before they become chargebacks

Not every chargeback starts with fraud. Many start with confusion.

Cardholders dispute a transaction when they do not recognize the merchant name, do not understand the charge, or did not expect the payment. Stripe’s dispute prevention tools help reduce that failure mode before it turns into a formal dispute.

Stripe supports dispute prevention and resolution workflows powered by:

Verifi from Visa
Ethoca from Mastercard

These tools help you:

Surface transaction data earlier
Show cardholders details that identify the charge
Reduce unnecessary disputes
Deflect chargebacks before they hit your backlog

This is especially useful for:

Subscription businesses with recurring charges
Merchants with multiple brands or descriptors
High-volume consumer businesses
Digital goods and services with low-touch support

The practical effect is simple: fewer “what is this charge?” disputes, less manual support load, and a better dispute rate.

Respond to chargebacks with a guided workflow

Even with good prevention, some disputes will get through. When they do, Stripe gives you a guided process for card transactions so your team is not assembling evidence from scratch.

The workflow is built around the things issuers care about:

Order details
Transaction history
Customer communications
Shipping or fulfillment proof
Refund or cancellation records
Prior account activity

That matters because chargeback handling is operational work. The faster your team can collect the right evidence, the better your chance of winning legitimate disputes.

Manual dispute response

If you counter disputes manually, Stripe charges $15.00 per dispute received.

If you win, the fee is returned.
If you lose, the fee is not returned.
Rare dispute types may incur network fees.

Manual response is still useful when:

Dispute volume is low
You need full control over evidence
Your dispute patterns are unusual
You want humans to review edge cases

Smart Disputes

Stripe’s Smart Disputes automatically compiles and submits evidence for eligible disputes using Stripe’s AI. It is in public preview.

Pricing is simple:

30% of the disputed amount if you win
No fee if you lose

This is useful when dispute volume is high and your team is spending too much time on repetitive evidence assembly. Smart Disputes reduces ops overhead and standardizes response quality.

Build a chargeback workflow that scales

The best card-not-present fraud strategy is not just “block more.” It is a sequence:

Screen every payment with Radar
Apply custom rules for your business policy
Route medium-risk transactions into review
Use dispute prevention tools to deflect avoidable chargebacks
Submit evidence fast when a dispute lands
Automate eligible disputes with Smart Disputes

That flow works because it separates three different jobs:

Fraud prevention
Dispute prevention
Dispute response

Most teams fail when they treat those as the same problem.

What to measure

If you are serious about fraud prevention and chargeback tools for card-not-present payments, watch these metrics closely:

Authorization rate
Fraud rate
Chargeback rate
Win rate
False decline rate
Manual review rate
Time to decision
Time to evidence submission
Net revenue recovered from disputes

Use those numbers to tune your rules. If your fraud rate drops but false declines spike, your controls are too tight. If your conversion stays flat but disputes rise, your fraud thresholds are too loose.

A practical setup for online businesses

A good starting point looks like this:

For early-stage teams

Turn on Radar
Add a few high-signal rules
Review suspicious orders manually
Keep dispute evidence organized from day one

For growth-stage teams

Route risky payments into a review queue
Add dispute prevention tools for recurring or high-volume charges
Standardize your evidence checklist
Review rule performance weekly

For high-volume teams

Use Radar at scale with tighter segmentation by region, product, and customer type
Automate eligible dispute responses with Smart Disputes
Track chargeback reasons by cohort
Tune rules to protect authorization rate while reducing fraud loss

Bottom line

For card-not-present payments, the right stack is composable:

Radar ML finds fraud patterns
Custom rules enforce your policy
Verifi and Ethoca help prevent avoidable disputes
Guided dispute workflows make chargeback response operationally manageable
Smart Disputes reduces manual work for eligible cases

Use them together, and you protect revenue without building a separate fraud and disputes team for every new market.

If you are starting from scratch, begin in the Stripe Dashboard. If you already run at scale, add dispute automation and custom controls where your losses are highest.