Promotional retail pricing strategies through optimisation modelling

Abstract

During trade promotions, vendors provide trade spend-funds allocated to retailers to incentivise the promotion of specific stock, typically through the application of price discounts. However, because common approaches to promotion planning are arduous, overreliant on managerial experience, and prone to human error, retailers often apply price promotions sub-optimally by failing to fully consider the interrelated impacts of consumer response and trade spend. While a systematic literature review (SLR) reveals that systematic approaches for promotion optimisation exist in the literature, none simultaneously accounts for own-promotional effects (the demand response of the promoted items), crosspromotional effects (the demand response of related non-promoted items), and trade spend effects. This study aims to address the problem and gap identified in the literature by proposing a new mixed-integer nonlinear programming (MINLP) optimisation model that determines the discount depths of retail items, maximising retailer profit, while incorporating ownpromotional, cross-promotional, and trade spend effects. The model is verified in two stages. First, the model's adherence to the requirements of the selected MINLP heuristic solver (BONMIN) is verified to ensure correct solver use. Second, the correctness of the model's formulation and coding is verified to ensure it implements the intended conceptual design. Next, the model is validated in three stages, each employing non-trivial simulated data. First, a benchmark analysis evaluates the optimality gap between heuristic and exact solutions across 20 datasets, spanning small to large problem sizes. The results indicate that the heuristic solutions are exact in all cases, validating the reliability of the solving approach's solutions. Second, a sensitivity analysis is conducted in which key model parameters are varied, after which the responses are qualitatively assessed. The results indicate that the responses are internally consistent and theoretically aligned across parameter variations, validating the model's ability to reflect realistic behaviour. Third, a scalability analysis is conducted wherein the model is solved using systematically increasing problem sizes to establish a trend in computation time. The results indicate that computation time grows exponentially with the number of variables, validating that the solving approach remains effective only for relatively small problem instances. This study is limited by the absence of a conclusive benchmark of its profit-enhancing capability. In addition to addressing scalability and benchmarking issues, future research could extend the model by incorporating a stochastic framework to account for demand uncertainty, framing it in a multi-period setting to capture overpromotion effects (e.g.,post-promotion demand dips), and expanding the representation of trade spend, crosspromotional, and inventory effects. Despite these limitations, this study contributes to the literature by presenting the first known retail promotion optimisation model that accounts for own-promotional, crosspromotional, and trade spend effects simultaneously. In doing so, the study demonstrates that the model yields higher retailer profits than isolated or ad hoc discounting strategies. Thus, the study highlights the financial value of holistic decision-support tools in reducing reliance on intuition in retail promotion planning.