Why identity intelligence is now critical for AI-powered customer experience. By Zahava Dalin-Kaptzan
As retailers accelerate instant, AI-driven customer service, a new problem is emerging: refund fraud and policy abuse are scaling at the same speed as the automation of customer experience (CX) operations. The main challenge is no longer just trying to stop payment fraud at checkout – it’s understanding who you’re dealing with, in real time, across every customer touchpoint. For example, a customer contacting your support team to claim a missing delivery and request an instant refund may look genuine, but they may actually be linked to similar claims across a range of merchants. Without identity context, merchants remain blind to these larger patterns of abuse.
Fraud has evolved and so must defenses
Online fraud today isn’t a lone hacker in a basement. It’s a fast-growing global industry with supply chains, vendors, subcontractors, customer support, performance metrics, and even reputation systems. If payment fraud were an industry, it would be one of the fastest growing in the world, with ecommerce fraud losses alone forecast by Juniper Research to rise from $56 billion in 2025 to $131 billion by 2030.
Fraudsters are often the first to adopt new technology, because it gives them leverage and time before defenses catch up. Generative AI is making synthetic identities, voice spoofing, and social-engineering-led refund fraud far easier to scale. The most sophisticated fraud uses systems exactly as they were designed, just with stolen identities. Triangulation fraud, for example, uses legitimate platforms and real transactions powered by stolen credentials.
Fraud has moved beyond payments into the policies merchants use to create good customer experiences. Returns, refunds, and policy abuse are growing areas of risk. Nearly one in ten returns today are fraudulent, and nearly one in four refund dollars is now driven by abusive claims, underscoring how return policies have become a key target of ecommerce fraud.
What’s changed the most in recent years is the speed and scale of it. CX teams are being pushed to resolve issues instantly, often without the context needed to assess risk. And as more retailers deploy AI agents on the front lines of customer service, real-time identity risk becomes even more crucial. It helps prevent manipulation and abuse while also reducing false positives that can negatively impact customer satisfaction.
The network effect and identity intelligence
A growing share of refund abuse is driven by small networks of repeat offenders operating across multiple merchants. Riskified’s network data shows that just 13 percent of identities with claims operate across multiple retailers, yet these identities are linked to seven times more accounts on average, driving disproportionate losses. A network view of identity is critical for detecting repeat and coordinated abuse.
The future of retail hinges on delivering fast and exceptional experiences to loyal customers while protecting against increasingly sophisticated abuse patterns. Knowing who you are dealing with in real time, referred to as ‘identity intelligence’, is becoming essential to delivering fast, AI-powered customer experiences. By integrating real-time identity risk scores directly into customer service workflows, retailers can give agents – increasingly, AI agents – immediate visibility into who they’re dealing with. For example, when a customer submits a refund claim or requests a package reroute, identity intelligence can provide immediate visibility into risk signals, helping service teams make faster and more accurate decisions. This allows your teams to fast-track trusted customers for instant resolution, improving retention and repeat purchase behavior, while serial abusers are stopped more effectively.
The trade-off between speed, trust, and security
Retailers have long faced a difficult compromise: offer instant claims resolution to drive loyalty, or implement rigid rules to block the rising tide of fraudulent refund claims and return abuse? The path forward lies in turning data and transparency into a dual advantage to both protect business margins while improving customer trust.
Leading merchants are now rethinking the trade-off between speed, trust, and security. With the right intelligence and tools, brands and retailers can automatically flag problematic returns, adjusting policies to match risk levels associated with specific regions and consumers. This reduces risk on the merchant’s side while keeping the returns experience seamless for honest customers.
To address the evolving risks, retailers should focus on leveraging both behavioral and network data to personalize their policies. This could look like offering greater flexibility to reliable customers with a history of legitimate transactions, while applying increased friction to individuals or accounts flagged for suspicious activity. By tailoring the customer experience, merchants can reward loyalty, while deterring repeat offenders who exploit return and refund policies. Additionally, transparency around return costs should be used as a behavioral lever to provide clear and upfront policy language at checkout. This informs customers whilst also encouraging more thoughtful purchasing and discouraging dishonest returns.
From blanket restrictions to nuanced policy design
As AI-generated return claims become more prevalent, it’s essential for service teams to be trained to recognize these patterns, and for merchants to monitor for this type of activity proactively. Rather than applying blanket and outdated restrictions, retailers can apply friction selectively based on risk, using tiered approaches that match the level of scrutiny to the risk profile of each transaction. Finally, it is important to calibrate return policies by market, taking into account regional differences in return culture and fraud risk. This nuanced approach would ensure that policies remain effective and fair, regardless of geographic or demographic variations, and help retailers maintain both operational efficiency and customer trust in an increasingly complex environment.
The days of fraud being about stolen cards are fading fast. It’s already about stolen identities, automated systems, and exploiting trust at scale. As AI agents begin to shop and interact on behalf of customers, the challenge is distinguishing between a helpful assistant and an automated fraud attack. What’s more, in some new AI-driven checkout flows, up to a third of the data used to detect fraud is missing, making identity intelligence even more essential.
Retailers that invest in real-time, network-scale identity intelligence now will be best positioned to deliver seamless experiences for their best customers going forward while continuing to keep fraud and abuse in check.
For a practical example, see the Rue Gilt Groupe case study at
https://www.riskified.com/resources/video/rue-gilt-groupe
Zahava Dalin-Kaptzan
www.riskified.com
Zahava Dalin-Kaptzan is Senior Product Marketing Manager at Riskified. Riskified empowers businesses to unleash ecommerce growth by outsmarting risk. Many of the world’s biggest brands and publicly traded companies selling online rely on Riskified for guaranteed protection against chargebacks, to fight fraud and policy abuse at scale, and to improve customer retention. Developed and managed by the largest team of ecommerce risk analysts, data scientists, and researchers, Riskified’s AI-powered fraud and risk intelligence platform analyzes the individual behind each interaction to provide real-time decisions and robust identity-based insights.