🏠Recentcollected in 10m

Meituan optimizes delivery algorithms for rider welfare

Meituan optimizes delivery algorithms for rider welfare
PostLinkedIn
🏠Read original on IT之家
#logistics-ai#gig-economymeituan-delivery-algorithm

💡Learn how a major logistics platform uses algorithmic governance to balance operational efficiency and worker welfare.

⚡ 30-Second TL;DR

What Changed

Implementing dynamic compensation for 'difficult' delivery orders.

Why It Matters

These changes demonstrate a shift toward human-centric algorithmic management in gig economy platforms. It highlights the importance of incorporating real-world constraints into optimization models to improve operational sustainability.

What To Do Next

Analyze your system's feedback loops to identify if your optimization metrics are inadvertently penalizing users or agents, and introduce 'soft' constraints to improve retention.

Who should care:Developers & AI Engineers

🧠 Deep Insight

Web-grounded analysis with 25 cited sources.

🔑 Enhanced Key Takeaways

  • Meituan's algorithm now explicitly selects the longest among four possible estimated delivery times to display to riders, aiming to provide more realistic expectations and reduce pressure.
  • The positive scoring system, replacing negative penalty deductions, has been piloted in over 15 cities since March 2022 and expanded to 80 cities by December 2024, covering all Meituan Express riders, with full implementation for other rider types expected by the end of 2025.
  • Meituan has implemented an 'anti-fatigue mechanism' nationwide since late 2024/early 2025, which includes pop-up reminders after 8 hours of work and mandatory log-offs after 12 hours to ensure riders take adequate rest.
  • The company has expanded its pension insurance subsidy program nationwide, marking the industry's first social security subsidy scheme available to all delivery workers, covering basic benefits like work injury insurance and accident coverage, and advanced benefits such as serious illness support and children's education funds.
  • Meituan has introduced 'after-meal scheduling' and 'active reassignment' functions, with pilot data showing a 51.79% reduction in timeout rates for reassigned riders and a 51% decrease in average waiting time for delivery personnel in pilot stores.
📊 Competitor Analysis▸ Show
Feature/PlatformMeituanEle.me (Alibaba)JD.com
Algorithm TransparencyPublicly explained algorithm logic, shifted to 'people-oriented' model, quarterly roundtable discussions with stakeholders. Explicitly chooses longest ETA from four algorithms.Emphasizes lengthening delivery times, mechanisms for drivers to request more time. Signed agreement in Shanghai for algorithmic fairness/transparency.N/A (New entrant to food delivery)
Penalty SystemPhasing out late-delivery penalties by end of 2025, replacing with positive scoring system; piloted in 80 cities by Dec 2024.Exempts couriers with good records from late penalties; allows customers to grant extra delivery time. 'No penalty for overtime' policy rolled out.Piloting similar no-penalty program in 25 cities.
Social Insurance/WelfareExpanded nationwide pension insurance subsidy program (industry's first for all riders); offers work injury insurance, accident coverage, foul-weather subsidies, serious illness support, children's education funds.Piloted social insurance programs in 7 cities by Feb 2025; expanded work-injury insurance coverage; offers social insurance subsidy (at least 50% for consistent riders).Gradually introducing China's five types of social insurance and housing-fund benefits for full-time riders from March 2025; accident and medical coverage for part-time riders.
Fatigue ManagementImplemented nationwide 'anti-fatigue mechanism' with pop-up reminders after 8 hours and forced log-offs after 12 hours.Launched rider rest measures nationwide (August, likely 2024/2025); recommended short break pop-up reminders.N/A
Smart HardwareDeploys 'Chucanbao' (smart POS system) to optimize merchant prep times, increasing meal delivery speed by 30%+. Offers smart helmets with voice navigation for couriers.Offers smart gear (e.g., helmets) at discounts or redeemable with points.N/A

🛠️ Technical Deep Dive

  • Meituan's real-time intelligent dispatch system, implemented in 2019, decomposes the original multi-period, multi-objective stochastic assignment problem into a series of single-period, single-objective deterministic subproblems, which are solved independently at each dispatch moment through weight optimization.
  • The algorithm for estimated delivery time considers four evaluation algorithms: historical data model estimation, urban traffic characteristics estimation, cumulative estimation time for delivery scenarios (e.g., picking up food), and distribution distance estimation time, ultimately selecting the longest among these to display.
  • The system has achieved a 20.96% decrease in average order delivery time and a 23.77% decrease in average courier traveling distance per order since its implementation.
  • AI is utilized to predict demand, enhance Estimated Time of Arrival (ETA) accuracy, and guide merchant operations.
  • The 'Chucanbao' smart hardware likely refers to or integrates with Meituan's smart POS system, which features an integrated cash drawer, real-time processing of online food delivery orders, and can increase meal delivery speed by over 30%.

🔮 Future ImplicationsAI analysis grounded in cited sources

Meituan's explicit shift to a 'people-oriented' algorithm model will set a new industry benchmark for balancing efficiency with worker welfare.
By publicly committing to and implementing comprehensive welfare measures and algorithm adjustments, Meituan could pressure competitors to adopt similar practices to attract and retain riders, fostering a healthier platform economy.
Increased transparency and integration of rider feedback will lead to more resilient and adaptable delivery algorithms.
Meituan's plans for quarterly algorithm roundtable discussions with diverse stakeholders will provide continuous, real-world feedback, enabling the algorithms to better adapt to complex and unforeseen delivery scenarios.
Widespread adoption of smart hardware like 'Chucanbao' will significantly reduce friction points in the delivery chain, improving overall efficiency and rider experience.
By directly addressing merchant preparation times and order processing, smart hardware can mitigate major causes of rider delays and stress, leading to smoother operations and better service quality.

Timeline

2019
Meituan implemented its real-time intelligent dispatch system, reducing delivery time and courier travel distance.
2020-09
Chinese magazine Renwu published 'Delivery Workers Trapped in the System,' exposing harsh conditions and algorithmic pressures, leading to public outcry and Meituan's initial response to add an 8-minute buffer.
2021-07
Chinese authorities issued guidelines urging platforms to improve compensation and benefits for food delivery riders, including social insurance coverage.
2021-11
Meituan published its algorithm rules for order allocation, introducing 'after-meal scheduling' and 'active reassignment' functions.
2022-03
Meituan piloted a new incentive mechanism in 15 cities, replacing penalties with a scoring system for bad reviews and late deliveries.
2024-12
Meituan announced plans to abolish late-delivery penalties by end of 2025, transitioning to positive incentives and a points-based system, already implemented in 80 cities. Anti-fatigue mechanisms (pop-up reminders and forced log-offs) were also implemented nationwide.
2025-11
Meituan expanded its pension insurance subsidy program nationwide, becoming the industry's first to offer such a scheme to all delivery workers.
2026-06
Meituan announced a shift from an 'efficiency-first' to a 'people-oriented' algorithm model, optimizing order dispatching, pricing, and performance assessment penalties, including penalty exemptions for late deliveries and enhanced support for new riders.
📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: IT之家