Picker Selection Model for Concierge Delivery Acceptance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online concierge systems face challenges in reducing late deliveries and increasing costs due to the inefficiencies in sending service requests to pickers, as the process of escalating requests to further-away pickers increases the time for acceptance and degrades service quality.
Innovation Solution
An online concierge system uses machine learning models to predict the likelihood of picker acceptance and response times, simulating responses based on previous orders to determine the minimum number of pickers to send service requests, optimizing the selection of pickers based on proximity and order attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If service requests are sent to progressively more pickers who are further away from retailer locations, then the likelihood of order acceptance increases, but the delivery time increases and service quality degrades
Solution Approach 1:
The system performs preliminary actions by sending service requests to multiple pickers simultaneously rather than sequentially. The picker selection module identifies a plurality of pickers and sends requests to all of them at once, eliminating the waiting time associated with sequential escalation and immediately establishing the fastest possible acceptance timeline.
Solution Approach 2:
The system segments the picker selection process into distinct functional modules: a picker selection module that identifies candidate pickers based on proximity and availability, and an acceptance prediction module that evaluates their likelihood of accepting requests. This segmentation allows for optimized, parallel processing of multiple picker evaluations.
2Reliability
If service requests are sent to more pickers to increase acceptance likelihood, then order fulfillment reliability improves, but system costs increase
Solution Approach 1:
The system applies partial action by sending service requests to a selectively determined number of pickers rather than all available pickers. The acceptance prediction module calculates the optimal number of pickers needed to achieve a target acceptance probability, avoiding the waste of sending requests to excessive numbers of pickers when fewer would suffice.
Solution Approach 2:
The system changes the parameter of picker selection from simple proximity-based ranking to a probabilistic model that incorporates acceptance likelihood predictions. This parameter change enables the system to optimize the number of pickers contacted based on predicted acceptance rates, reducing unnecessary communications and associated costs while maintaining fulfillment reliability.
3Loss of energy
If service requests are sent sequentially to pickers based on proximity, then cost efficiency improves, but the risk of late delivery increases
Solution Approach 1:
The system performs preliminary actions by simultaneously contacting multiple pickers before any acceptance decisions are made. This parallel preliminary action eliminates the sequential waiting period, ensuring that the fastest possible acceptance is captured without incurring additional costs from prolonged search times or late deliveries.
Solution Approach 2:
The acceptance prediction module provides feedback about the likelihood of each picker accepting a request, allowing the system to optimize the picker selection strategy. This feedback mechanism enables the system to balance cost efficiency with on-time delivery reliability by selecting pickers whose predicted acceptance patterns minimize both cost and delivery time.
Data Source
AI summary
An online concierge system accesses and applies a model to predict likelihoods of acceptance of a service request for an order by pickers. The system accesses timespan distributions for accepted service requests and identifies sets of pickers based on the order. Based on the likelihoods and distributions, the system generates simulated responses of the sets of pickers to the service request and trains an additional model based on attributes of the order, the simulated responses, and information associated with corresponding sets of pickers. The system receives a new order, identifies additional sets of pickers based on the new order, and applies the additional model to predict responses of the additional sets of pickers to an additional service request for the new order. Based on the predicted responses and a delivery time associated with the new order, a minimum number of pickers to send the additional service request is determined.