Paper Type

ERF

Abstract

Cart abandonment remains a persistent challenge in digital commerce, particularly when shoppers encounter shipping costs near checkout. This study proposes a threshold-aware checkout optimization system that reframes cart completion as a multi-objective decision-support problem. Grounded in goal-gradient theory, multi-objective recommender systems, and explainable AI, the system recommends personalized add-on items that help shoppers reach free-shipping thresholds while balancing consumer fit and retailer-side constraints such as inventory and margin priorities. Using real-world retail transaction data, we conduct an offline structural evaluation with reconstructed threshold-proximate baskets. Results show that the proposed multi-objective decision-support system improves structural threshold completion feasibility relative to a static add-on recommender, though this improvement involves trade-offs in historical co-occurrence alignment. This initial study establishes technical logic and structural feasibility rather than behavioral or conversion effects. Future field validation will examine trust, explanation quality, decision quality, and actual cart completion.

Paper Number

1939

Comments

NEXTTRANS

Download

Share

COinS
 
Aug 15th, 12:00 AM

From Abandonment to Completion: A Multi-Objective Decision Support Approach to Checkout Optimization

Cart abandonment remains a persistent challenge in digital commerce, particularly when shoppers encounter shipping costs near checkout. This study proposes a threshold-aware checkout optimization system that reframes cart completion as a multi-objective decision-support problem. Grounded in goal-gradient theory, multi-objective recommender systems, and explainable AI, the system recommends personalized add-on items that help shoppers reach free-shipping thresholds while balancing consumer fit and retailer-side constraints such as inventory and margin priorities. Using real-world retail transaction data, we conduct an offline structural evaluation with reconstructed threshold-proximate baskets. Results show that the proposed multi-objective decision-support system improves structural threshold completion feasibility relative to a static add-on recommender, though this improvement involves trade-offs in historical co-occurrence alignment. This initial study establishes technical logic and structural feasibility rather than behavioral or conversion effects. Future field validation will examine trust, explanation quality, decision quality, and actual cart completion.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.