Fraud Scoring

What fraud scoring is

Fraud scoring assigns a risk rating to a payment or order. A scoring system analyzes available information and estimates whether the transaction looks legitimate, suspicious, or likely to be fraudulent. A score is an input to a decision, not proof that a customer is committing fraud.

Signals can include transaction history, amount, device, location, billing details, and the pace of repeated attempts. Some systems use a 0-to-99 scale, with 0 indicating very low risk and 99 very high risk. That scale is not universal, and there is no industry-wide score cutoff for approving or declining a sale.

A fraud score is a risk signal, not a verdict. Your rules decide what happens next.

How fraud scoring works

A model or set of rules evaluates signals in milliseconds. A system might flag a burst of payment attempts, an unfamiliar device, or a mismatch between billing and shipping details. One signal alone does not establish fraud. The system weighs the combination, and the same order can receive different scores from different tools because they use different data and models.

Your business, an issuer, or a payment provider can use a score at different points. A merchant can screen an online order before authorization, an issuer can use risk signals while deciding whether to authorize the card, and your team can review an order before fulfillment. The score provider or your own rules determine how the result is used.

What it means for your business

Scoring helps you manage the tradeoff between fraud losses and rejected legitimate orders. A low-risk order can proceed with less friction; a suspicious order can be declined, sent for manual review, or routed to additional authentication. Blocking too many borderline orders creates false declines and lost sales. Letting too many high-risk orders through can lead to fraud losses and disputes.

For example, an illustrative $100 online order could be legitimate even if an unusual device or shipping change raises its score. A hard block protects against one possible fraud loss but also loses the sale if the customer is genuine. Reviewing the order or requesting another verification step gives you a middle option.

What it means for your costs

Fraud scoring is not one of the three basic card-processing cost layers: interchange, network fees, and processor markup. The research does not establish a standard per-score fee or a standard fraud-scoring line on a merchant statement. A fraud tool could have separate provider pricing, so treat a listed charge as a service fee and compare its description and billing terms.

The more direct financial effect comes from decisions. Fraud losses and related disputes cost you money, while false declines sacrifice legitimate revenue. Review fraud outcomes alongside approval and conversion rates, rather than judging a scoring tool only by how many orders it blocks.

What you can actually change

1. 01
   Set graduated actions. Allow low-risk orders, send uncertain cases to review or extra verification, and reserve automatic declines for stronger risk signals. No universal score cutoff fits every business.
2. 02
   Track false declines. Review customer complaints, abandoned checkouts, and later-confirmed legitimate orders. Compare those results with fraud losses to see whether your rules reject too many good customers.
3. 03
   Review signal patterns. Look at which signals repeatedly precede confirmed fraud, such as rapid attempts or device changes. Adjust rules when the pattern changes instead of keeping a stale block rule.
4. 04
   Check the bill. Look for a separate fraud-tool or risk-service charge and compare its description with the service and pricing terms. Do not assume a risk score itself is a card-network fee.

Frequently asked questions