Driving Retailer Performance by Optimizing Promotions (with Daniel Vogler)

Abstract

Promotions are a popular tactic used by retailers across different industries to attract consumers for various reasons, such as increasing foot traffic, retaining or acquiring customers, or gaining market share. However, with the growing demand for seamless omni-channel shopping experiences, managing and optimizing promotions has become more challenging, making it difficult to measure their impact on overall business success. Two approaches discussed here are (1) calculating the “net” effect of in-store promotions and (2) leveraging algorithms to optimize in-store promotion use and frequency. The first approach involves decomposing gross revenue into key revenue drivers using statistical modeling and machine learning, considering factors such as substitutive relationships between products, additional purchases due to halo effects, and stockpiling effects. We have implemented a modular and comprehensible measurement framework for our business partners, starting with developing simple statistical models and gradually moving towards more advanced modeling of complex topics, such as the degree of product substitution. This approach has allowed us to build trust and firmly establish our data products within the company while also maintaining a fast iterative development cycle. The second approach involves optimizing the use and frequency of promotions, which is particularly complex in highly decentralized organizations like Migros, with thousands of products across various verticals. We demonstrate how Migros transitioned from strictly adhering to business rules for optimizing promotion usage and frequency to adopting an algorithmic approach that considers business guidelines but doesn’t follow them blindly. Our focus is on the difficulties of translating business rules into optimization constraints, and the significance of change management in moving towards data-driven retail. We explore critical aspects of business rules and their integration into an analytical model and identify key success factors and potential obstacles for implementing data-driven performance management in practice. This presentation emphasizes the significance of comprehending the difficulties involved in managing and optimizing in-store promotions. It showcases the potential of advanced analytics in assisting promotions and the requirement for co-creation with business stakeholders to enhance performance management in retail. Overall, the presentation stresses the need for a collaborative approach towards addressing the challenges associated with managing and optimizing in-store promotions.