Idea
Real-time hold control system increasing ride-hailing trip completion and reducing cancellations through experience-aware matching.
Research Paper
Core Innovation
This paper introduces EXHOLD, a deployable two-stage framework that decouples experience-aware pair assessment from hold-time execution. It optimizes a unified satisfaction objective across the matching funnel and enforces monotone hold-time schedules with service guardrails, outperforming heuristic thresholding methods under non-stationary traffic conditions.
Why It Matters
Ride-hailing platforms face challenges with cancellations and inefficient driver effort due to suboptimal matching. EXHOLD improves passenger and driver satisfaction by strategically delaying matches to find better opportunities, reducing cancellations and wasted effort. This scalable approach enhances marketplace efficiency and driver income, critical for large-scale operations.
Market Size (TAM)
$10–20B TAM for ride-hailing platform optimization; $2–5B SAM from large-scale ride-hailing operators. Driven by growth in urban mobility demand and need for improved user experience.
Potential Customers & Pain Points
- Ride-hailing platforms – High cancellation rates and inefficient driver utilization
- Transportation marketplaces – Need to optimize matching for better user experience
- Fleet management companies – Desire to increase driver earnings and reduce idle time.
Business Model
Licensing or SaaS model offering EXHOLD as a real-time matching optimization service to ride-hailing platforms and transportation marketplaces.
Competitive Landscape
- Uber Matching Algorithms
- Lyft Dispatch System
- Grab Ride Matching
- Ola Intelligent Dispatch
Implementation Challenges
- Integration complexity with existing ride-hailing platforms
- Real-time data processing and scalability challenges
- Adapting to diverse and dynamic traffic patterns across regions
Validation Strategy
- Conduct randomized A/B testing in multiple geographic markets
- Measure key metrics: trip completion rate
- cancellation rate
- driver income
- Perform behavioral analysis to ensure calibrated decisions under traffic heterogeneity
- Iterate model based on real-world feedback and performance data
Research Paper Overview
EXHOLD: Experience-Aware Real-Time Hold Control for Large-Scale Ride-Hailing Matching at DiDi
Summary
EXHOLD is a two-stage framework that improves ride-hailing matching by selectively delaying driver-order pairs to enhance passenger and driver experience. It optimizes hold times based on experience tiers and enforces service guardrails, leading to higher trip completion, reduced cancellations, and increased driver income. Deployed in DiDi's Brazil operations, it adapts to spatiotemporal traffic variations and improves marketplace efficiency.