Authentication Friction Yield (AFY)

GMI-PAY-002Payments & Transaction ProcessingIntegration-Required

Definition

Net revenue impact of AI-targeted step-up authentication per challenged transaction: measuring whether challenges preserve revenue by stopping fraud or destroy it by driving abandonment of legitimate customers.

The Measurement Gap

The Measurement Gap

No metric tells you whether 3DS and step-up authentication challenges are being applied to the right transactions: specifically, whether the challenge converts revenue by stopping fraud on transactions that would have resulted in chargebacks, or destroys revenue by driving legitimate customers to abandon their purchase. The individual components are tracked in isolation: 3DS challenge rate is a standard operational metric, abandonment rate per challenge is sometimes measured, and fraud outcomes on challenged transactions flow through the chargeback system. But nobody connects the targeting accuracy of the challenge decision to its economic outcome.

This gap matters because authentication friction is currently optimized for risk reduction alone, treating every prevented fraud as equally valuable and every abandoned transaction as acceptable collateral damage. When the fraud scoring model targets a step-up challenge at a $15 transaction from a repeat customer with a 0.1% fraud probability, the expected fraud prevention value is approximately $0.015 while the expected revenue destruction from a 20% abandonment probability is roughly $0.45 in margin. The challenge is a net-negative decision, but no metric captures this because risk and revenue live in different dashboards.

The disconnect is compounded by the fact that challenge completion rates vary dramatically by merchant category (70-85% according to Visa implementation studies), meaning the revenue destruction from poorly targeted challenges is not evenly distributed.

Decision Supported

Are AI-targeted step-up authentication challenges net revenue-positive or net revenue-negative per challenge issued?

Formula

Plain Language

For every step-up challenge issued, calculate the fraud value prevented by the challenge (fraudulent transactions that failed authentication or were deterred by the friction), subtract the revenue destroyed when legitimate customers abandoned at the challenge, and divide by total challenges. The result is the net dollar value generated per challenge event. Completed legitimate transactions are excluded from the numerator because the challenge did not create that revenue: the transaction would have been approved without the friction.

Notation

AFY = [Σ(TxnAmount_fraud_prevented × FraudRecoveryRate) − Σ(TxnAmount_abandoned_legit × Margin)] / Total_challenges

Variables

Variable Definition Unit
TxnAmount_fraud_prevented Amount of fraudulent transactions prevented by the challenge: includes failed authentications (fraud stopped) plus estimated fraud in the abandoned population (fraud deterred) $
FraudRecoveryRate Estimated unrecoverable fraud fraction; typically 0.85–0.95 for card-not-present Ratio (0–1)
TxnAmount_abandoned_legit Amount of transactions where a legitimate customer abandoned the challenge and subsequently transacted elsewhere (proxy: same cardholder completed a purchase at another merchant within 24 hours) $
Margin Revenue margin per transaction (interchange + processing fee) Ratio (0–1)
Total_challenges Total step-up authentication challenges issued in the period Count

Estimating fraud deterred: Fraudsters who abandon at the challenge cannot be directly identified because the fraud never materialized. Estimate using the confirmed fraud rate in the completed-challenge population, applied to the abandoned population: Fraud_deterred_estimate = Abandoned_count × FraudRate_in_completed_challenges. This is a conservative lower bound because fraudsters are more likely to abandon than legitimate customers.

Proposed formula, pending validation pilots.

At a Glance

Unit
Currency per challenge
Trend Direction
Higher
Reporting Frequency
Monthly
Category
Fraud Model Economic Integrity
← Representment Evidence Quality Index (REQI)Authorization Threshold Staleness Index (ATSI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment