Payment Fraud Detection
Score every transaction in real time, challenge only what should be challenged, and tune the security/conversion balance with explicit rules — not a black box.
Most fraud is lost at the edges — in false positives that block good customers, in chargebacks that arrive months after the transaction, and in friction-heavy authentication flows that drop conversion. Inyo's anti-fraud layer combines real-time machine-learning risk scoring with a rules engine you control, 3D Secure 2 invocation only where it earns its keep, and chargeback management that closes the loop.
For money movement businesses processing AFTs, the risk model is tuned for funding transactions specifically — not generic e-commerce. Sender-data validation, BIN-level patterns, corridor risk, and historical chargeback behavior all factor into the score.
Real-time ML risk scoring
Every transaction is scored in milliseconds against models trained on Inyo's processing volume and corridor-specific fraud patterns. The score and the top contributing features are returned with the authorization — not as a separate API call — so your decision logic can use them inline.
Rules engine you can read
Risk decisions blend the ML score with explicit rules you own: velocity limits, BIN allow/block lists, country-pair restrictions, amount thresholds. Rules are versioned and auditable — you can always answer "why did we approve this?" after the fact.
3D Secure 2: frictionless first
3DS2 is invoked selectively: full data enrichment to maximize frictionless authentication, challenges only where the risk score or regulatory regime warrants. The result is liability shift on the transactions that need it without burning conversion on the ones that don't. Deep dive in the 3DS2 guide.
Chargeback management built in
Pre-dispute alerts via Verifi Order Insight and Ethoca surface issues before they become chargebacks. When disputes do come in, the platform supports CE 3.0 representment and tracks VDMP/ECP thresholds so a sudden chargeback spike doesn't silently put your processing at risk. See the chargeback management guide.
Frequently Asked Questions
How does Inyo's fraud detection work?
Every transaction is scored in real time using ML models trained on Inyo's processing volume and corridor-specific patterns, combined with an explicit rules engine you control (velocity limits, BIN lists, country pairs, amount thresholds). The score and top contributing features are returned inline with the authorization.
Does Inyo support 3D Secure 2?
Yes. Inyo applies 3DS2 with full data enrichment to maximize frictionless authentication and challenges only where risk or regulation warrants — preserving liability shift without burning conversion.
How are chargebacks handled?
Inyo integrates Verifi Order Insight and Ethoca pre-dispute alerts to resolve issues before they become chargebacks. Disputes that proceed are supported with CE 3.0 representment, and VDMP/ECP thresholds are tracked so you don't silently breach a processor program.
Can I write my own rules?
Yes. Rules are versioned and auditable. You can blend the ML score with explicit rules — velocity, BIN allow/block lists, country pairs, amount thresholds — without giving up the model's contribution.
Is the fraud detection tuned for AFTs?
Yes. The model is specifically tuned for Account Funding Transactions and other money-movement flows — not generic e-commerce. Sender data, corridor risk, and BIN-level patterns relevant to funding transactions all contribute to the score.
Related reading
3D Secure 2 Guide
Frictionless authentication, challenge handling, authorization-rate improvements, and integration modes.
Chargeback Management Guide
Dispute lifecycle, Visa reason codes, CE 3.0, pre-dispute alerts, and representment best practices.
Authorization Rate Optimization
How to recover revenue from false declines without opening the fraud door.
Tune the security/conversion balance — deliberately
Talk to the Inyo team about ML scoring, 3DS2, and chargeback management for your payment volume.
Talk to Sales