|
Article Excerpt
The main objective of this report is to describe a decision-support system for dynamic retail pricing and promotion planning. Our weekly demand model incorporates price, reference price effects, seasonality, article availability information, features, and discounts. Building on previous research, we quantify demand interdependencies and integrate the resulting profit-lifting effects into the optimal pricing model. The methodology was developed and implemented at bauMax, an Austrian do-it-yourself retailer. Along with the practical requirements, an objective function was employed that can be used as a vehicle for implementing a retailer's strategy. Eight pricing rounds with thousands of different stock-keeping units have each served as a testing ground for our approach. Based on various benchmarking methods, a positive impact on profit was reported. The currently implemented marketing decision-support system increased gross profit on average by 8.1 and sales by 2.1%.
Key words: reference price; demand interdependency; revenue management; retail strategy; pricing research; dynamic pricing
History: This paper was received August 1, 2005, and was with the authors 3 months for 2 revisions; processed by Gary Lilien.
**********
1. Company Background and Management Problem
bauMax was founded in 1976 and today operates 121 retail stores in six Central and Eastern European countries. The company currently realizes yearly revenues of approximately 1 billion euro, which makes it the largest competitor in the Austrian do-it-yourself (DIY) retail industry. The company has a market share of 25%.
Over the last few years, the Austrian DIY industry has experienced stagnation in sales. Most retailers have extended their assortment to compensate for the sales decrease in their traditional range and now--besides tools and building materials--also sell articles such as furniture and gardening products, bauMax currently has an assortment of around 60,000 stock-keeping units (SKUs). Facing fierce price competition and low profit margins in its home market, bauMax management has successfully initiated an expansion strategy into new Eastern European markets.
To successfully handle the managerial and operational complexity provoked by the excessively large assortment, only a subset of articles was subject to extensive pricing and promotion-planning activities. Based on its own experience, bauMax has practiced tactics such as high-low-pulsing, multiple placements of products in stores, and out-of-store advertisements via flyers or trade magazines. For the large majority of articles, no individual pricing process took place. Instead, prices were determined by simple markup rules and a static price grid. Hence, reengineering of the bauMax pricing process via implementation of an assortmentwide decision-support system based on state-of-the-art marketing decision models was considered. The prospective system was intended to help purchasing, merchandising, and marketing managers in their decision making regarding dynamic price adjustments and promotion-planning activities, as well as to facilitate coordination of interrelated decision tasks such as inventory replenishment.
The companywide data warehouse solution provided fast access to article-specific demand data; it also offered the possibility to condition prices on demand histories (cf. Acquisti and Varian 2005). In brief, the following major functional requirements were expected from the fully implemented pricing decision-support model: (i) it should automatically process article-specific pricing and promotion recommendations, (ii) it should respect inventory considerations to avoid out-of-stock situations, (iii) it should account for indirect profit contributions due to observed purchase interdependencies (lift effects), and (iv) it should be applicable in several countries with different currencies and sales taxes.
2. Model Development
Personalized transaction histories and a representative survey conducted among shoppers in the Austrian DIY retail industry have shown evidence of a relatively high level of store visit frequency. This supports the assumption that typical DIY customers may be characterized by a considerable degree of price consciousness. In addition, it justifies the adoption of reference-price models for deriving pricing and promotional recommendations. Reference price models (Winer 1986) have experienced broad empirical support (Kalyanaram and Winer 1995), and researchers have proposed many methods to infer the unobserved reference price (r). The operationalization applied at bauMax is based on an exponentially smoothed function of the item's own price history (cf. Lattin and Bucklin 1989, Greenleaf 1995, Briesch et al. 1997, Kopalle et al. 1996, Seetharaman 2004). A generalization of our model implemented at bauMax is that customers are permitted to faster adapt their reference prices ([micro] [less than or equal to] [eta]) to price reductions than to price increases:
[r.sup.t] = [r.sup.t-1] + [micro]max(0, [p.sup.t-1] - [r.sup.t-1]) + [eta] min (0, [p.sup.t-1] - [r.sup.t-1]), [less than or equal to] [micro], [eta] [less than or equal to] 1. (1)
For the case of [micro] = [eta], this model reduces to the basic adaptive reference price model. This generalization...
|
