Sift

Sift: Digital Trust & Safety and Real-Time Machine Learning

Sift is a pioneer in Digital Trust & Safety, empowering global enterprises to unlock revenue without accepting unnecessary risk. Moving away from static, rules-based systems, Sift relies on a massive global data network and advanced, real-time machine learning to accurately predict and stop payment fraud, account takeovers, and content abuse before they impact your bottom line.

Capabilities:

• Sift Global Data Network: Ingests and analyzes billions of events per month across thousands of global merchants. If a fraudster attacks one merchant in the network, Sift's machine learning models instantly adapt to protect all other merchants from the same vector.
• Real-Time Dynamic Friction: Allows you to tailor the checkout experience instantly based on user behavior. Trusted users get a frictionless fast-track, while suspicious actors are met with verification hurdles.
• Account Defense (ATO Prevention): Deeply analyzes login behaviors, device fingerprints, and credential stuffing patterns to stop fraudsters from hijacking legitimate customer accounts and exploiting stored payment methods.
• Dispute Management: Offers comprehensive chargeback management capabilities to automatically compile compelling evidence and fight friendly fraud or first-party misuse.

How It Works With Our Orchestration Engine: Sift evaluates transactions and returns a detailed fraud probability score (from 0 to 100) alongside configurable automated decisions (such as Accept, Watch, or Block). When integrated into our platform, we instantly translate these outputs into our unified risk tiers. A low probability score or "Accept" maps to Low Risk for an immediate, frictionless checkout; a mid-range score or "Watch" maps to Medium Risk to trigger a 3D Secure (3DS) challenge; and a high score or "Block" maps to High Risk to instantly deny the transaction and prevent authorization fees.

Risk Tiers

The table below outlines how Sift's probability scores and decision outputs are mapped to our three standardized risk tiers. (Note: Sift scores represent the probability of fraud, meaning a higher score indicates a higher risk. You can fully customize the exact score thresholds in our platform based on your business logic).

Sift returns a Fraud Score (0-100, representing the percentage likelihood of fraud) and a corresponding Decision based on your configured policies..