Accertify's Predictive Yes Platform has applied continuous ML-based fraud decisioning across the full customer lifecycle, from account creation to post-purchase.
The capability expansion reflects what Accertify describes as a structural shift in fraud patterns, one in which attack vectors now span account creation, login, account changes, and post-purchase activity such as refund abuse and promotion exploitation, in addition to checkout.
Connecting signals across the customer journey
The platform aggregates identity, behavioural, browsing, purchase, and outcome signals, including returns, refunds, and chargebacks, into a unified view of each customer. Rather than issuing isolated decisions at discrete transaction points, the system applies risk decisioning continuously from the moment a user arrives on a platform through to post-purchase interactions. In addition, the stated intent is to reduce friction for verified customers while maintaining fraud controls, operating on the principle that a trusted customer confirmed earlier in the journey should face fewer barriers at checkout.
The platform's decisioning layer draws on several model types, including decision trees, supervised machine learning, and graph networks, each applied to specific fraud categories using vertical-specific signals.
Three structural features underpin the approach. First, scale: between 1 May 2025 and 30 April 2026, Accertify processed more than ten billion transactions representing USD 1.25 trillion in commerce, according to the company's own client data. Second, a consortium model in which 98% of clients have opted in to share anonymised fraud intelligence across hundreds of enterprise brands, enabling network-wide pattern recognition. Third, layered AI models purpose-built for specific fraud types and continuously updated.
Recent platform improvements
Accertify has introduced several additions to the platform. An AI-driven capability now applies machine learning to the manual review queue prior to analyst intervention, aiming to automate a portion of reviews and reduce the misclassification of legitimate customers. Graph network functionality has been added to enable real-time detection of fraud rings and coordinated attacks through the process of mapping non-linear relationships between identities, devices, behaviours, and transactions. Updated industry-specific AI models have also been deployed to improve decisioning precision across Accertify's largest client segments.
Furthermore, the launch reflects a broader convergence in the fraud and cybersecurity disciplines, as merchants grapple with account takeover, identity confidence, and card-not-present fraud, threat types that increasingly require cross-functional responses and systems capable of treating the customer journey as a single, continuous attack surface rather than a series of isolated checkpoints.
