E-Commerce Platform Ranking Based on Target-Based Performance Indicators

Abstract

Managing a broad product assortment on an e-commerce platform requires category performance to be judged jointly across several, frequently conflicting, indicators rather than through a single sales figure, which turns category ranking into a multi-criteria decision-making problem. In such problems the outcome is driven above all by how the criteria weights are obtained, yet applications in the e-commerce domain almost always adopt either a subjective or an objective weighting scheme, rarely both. This study evaluates and ranks thirty-two product categories within an active e-commerce firm using seven target-based key performance indicators (KPIs), where each criterion value is operationalized as the ratio of realized performance to the predetermined target. Furthermore, the paper provides a comparative analysis of the outcomes yielded by subjective versus objective weighting methods applied to the same decision-making problem. Criteria weights are derived from the pairwise judgements of two domain experts through the Analytic Hierarchy Process (AHP) and, independently, from the statistical structure of the decision matrix through the Criteria Importance Through Intercriteria Correlation (CRITIC) method; both weight vectors are then supplied to a common TOPSIS ranking. Three months data are condensed into a single decision matrix using an exponentially weighted moving average (λ = 0.7) so that the most recent period dominates while earlier trends are retained. The two methods produce almost opposite weight structures — AHP concentrates roughly two-thirds of the weight in gross order count and net sales, whereas CRITIC assigns the largest weights to conversion rate and session count — yet the resulting rankings agree strongly (Spearman's ρ = 0.878, mean absolute rank shift of 3.7 positions, four of the top five categories shared). Categories that perform consistently across all criteria are insensitive to the weighting scheme, whereas categories with sharply unbalanced criteria profiles move by up to ten positions. The study delivers a systematic and reproducible decision-support model for category prioritisation and shows that the choice between subjective and objective weighting matters selectively rather than uniformly.