What happens when a gig economy platform replaces gut instinct with data-driven trust scores?
In this episode, we sit down with Kunal Kumar, Principal Product Manager at Rebel Foods—a global cloud kitchen company in the food delivery industry—who led the design and rollout of trust scores for both customers and delivery riders. He shares how Rebel built their models, what signals matter most, and the lessons learned while balancing fraud prevention with customer experience.
Key discussion topics include:
- Why Rebel Foods decided to create trust scores in the first place
- How trust scores balance fraud detection with fairness
- Segmenting users into tiers and tailoring treatment flows
- Cross-team alignment and the implementation process
- Using metrics to monitor effectiveness
- Pitfalls to avoid when designing scoring models
Key takeaways:
- Data over instinct. Trust scores remove bias and subjectivity from fraud decisions, replacing gut calls with a consistent, data-driven framework.
- Segmentation & treatment. Customers and riders are divided into tiers, with high scorers getting perks like fast refunds and low scorers facing stricter checks.
- Impact without compromise. Rebel Foods reduced retention spend and doubled refund automation after launching trust scores—while keeping customer satisfaction steady.
- Built for trust. Trust scores were built to reward genuine behavior, with every user starting with a high score and only dropping if risky patterns appear.
- Location is critical. GPS fidelity is a critical rider signal that's used to improve customer ETAs, ensure fair pay for riders, and catch compensation abuse.
Show notes:
Hosted on Acast. See acast.com/privacy for more information.
