How can Riskified improve its Chargeback Guarantee service to reduce false declines for legitimate transactions?

Introduction

To improve Riskified's Chargeback Guarantee service and reduce false declines for legitimate transactions, we need to conduct a comprehensive analysis of the current system, user behavior, and market dynamics. I'll outline a strategic approach to address this challenge, focusing on key stakeholders, pain points, and innovative solutions.

Step 1

Clarifying Questions

• Looking at Riskified's position in the e-commerce fraud prevention market, I'm curious about the current false decline rate. Could you share the percentage of transactions that are falsely declined and how this compares to industry standards?

Why it matters: Determines the severity of the problem and potential for improvement Expected answer: False decline rate is around 3-5%, slightly higher than the industry average of 2-3% Impact on approach: Would focus on refining the risk assessment algorithm if significantly higher than industry standards

• Considering the complexity of fraud detection, I'm wondering about the primary data sources and signals used in Riskified's current risk assessment model. Can you elaborate on the key factors considered when making approve/decline decisions?

Why it matters: Identifies potential gaps in data or overlooked signals that could improve accuracy Expected answer: Current model uses transaction details, device information, and historical customer data Impact on approach: Would explore integrating additional data sources or advanced machine learning techniques if current inputs are limited

• Given the critical nature of reducing false declines while maintaining fraud prevention efficacy, I'm interested in understanding the balance between these two objectives. What's the current ratio of fraud prevention to false declines, and how has this evolved over time?

Why it matters: Helps determine if the focus should be on reducing false positives or improving overall accuracy Expected answer: Current ratio is 10:1 (fraud prevented vs. false declines), stable over the past year Impact on approach: Would prioritize precision improvements if the ratio has been consistent, or focus on recalibrating the model if there's been recent fluctuation

• Considering the potential impact on merchant relationships, I'm curious about the feedback mechanism for declined transactions. How do merchants currently report and resolve cases of false declines?

Why it matters: Identifies opportunities for improving communication and dispute resolution processes Expected answer: Merchants can report false declines through a dashboard, with a 48-hour resolution time Impact on approach: Would focus on streamlining the reporting process and reducing resolution time if current system is cumbersome