Rue Gilt Groupe and Riskified have partnered to embed real-time identity risk scoring into RGG's customer service workflows.
The integration aims to allow service agents to differentiate between verified loyal members and suspected fraudulent actors at the point of interaction, without requiring additional manual review steps.
Customer experience teams in retail have long faced a trade-off between applying friction to deter abuse and offering service to genuine customers. That balance has become harder to strike as fraud tactics have grown more sophisticated. At the same time, the rise of generative AI has expanded the methods available to bad actors, including synthetic identities, manipulated imagery, voice spoofing, and social engineering, all of which increase the volume and plausibility of fraudulent refund and return claims.
With this in mind, RGG is addressing this by embedding Riskified's real-time identity risk scores into its Zendesk service console. When a customer contacts the service centre, whether to request a refund, report a missing package, or reroute a delivery, agents will receive an immediate risk signal tied to the identity behind the request. This process aims to enable faster resolution for trusted members while introducing targeted friction for high-risk interactions.
Network-scale intelligence
The underlying capability draws on Riskified's identity engine, which processes data from historical transactions across its merchant network. The system groups signals across emails, orders, accounts, addresses, phone numbers, and claims to build a high-confidence view of each individual. According to Riskified, 13% of identities associated with multiple claims show activity across more than one merchant, and those identities are linked to seven times as many accounts on average as identities with no claims, a pattern consistent with coordinated or repeat abuse.
Furthermore, merchants using Riskified's Policy Protect solution have reported up to a 30% reduction in complaint rates, as well as reductions in refund and return costs running into seven figures in some cases. A previously documented deployment with Ring, Amazon's smart home security brand, found that 600 individuals accounted for more than USD 4 million in annual abuse, with some individuals responsible for as much as USD 150.000 per year.
As enterprises increasingly deploy AI-powered agents at the front end of customer service, embedding identity risk signals at that layer becomes operationally significant. Automated workflows that can distinguish low-risk from high-risk interactions in real time reduce both fraud exposure and the rate of false positives that can erode customer satisfaction.
The partnership reflects a broader shift in how fraud prevention is being repositioned, not only as a back-end control mechanism, but as a capability integrated into customer-facing service operations. In addition, for off-price ecommerce platforms managing high volumes of returns and claims, real-time identity intelligence offers a path to tightening loss controls without degrading the experience for the majority of legitimate customers.
