Abstract:Clickstream data provides information about the sequence of pages or the path viewed by users as they navigate a web site. We show how path information can be categorized and modeled using a dynamic multinomial probit model of web browsing. We estimate this model using data from a major online bookseller. Our results show that the memory component of the model is crucial in accurately predicting a path. In comparison traditional multinomial probit and firstorder markov models predict paths poorly. These results suggest that paths may reflect a user's goals, which could be helpful in predicting future movements at a web site. One potential application of our model is to predict purchase conversion. We find that after only six viewings purchasers can be predicted with more than 40% accuracy, which is much better than the benchmark 7% purchase conversion prediction rate made without path information. This technique could be used to personalize web designs and product offerings based upon a user's path.

Abstract:Grocery retailers increasingly view other retail formats, particularly mass merchandisers, as a competitive threat. We present an empirical study of household shopping and packaged goods spending across retail formats -grocery stores, mass merchandisers, and drug stores. Our study considers competition between these formats and explores how retailer assortment, pricing and promotional policies, as well as household demographics, affect shopping behavior and expenditures in these different formats. This research is made possible by a new panel dataset collected by Information Resources Inc. (IRI) which captures consumer packaged good purchases made at alternative retail outlets. These purchases have previously been missed by panels that use only purchases at supermarkets.We estimate a hierarchical multivariate tobit model which captures consumer decisions about "where to shop" and "how much to buy." We find that shopping and spending vary much more across than within formats, and that the retailer's marketing mix explains more variation in shopping behavior than travel time. Of the marketing mix variables considered, we find that expenditures respond more to varying levels of assortment (in particular grocery stores) and promotion than price. This is surprising in light of the grocery industry's efforts to reduce retail assortments. Price sensitivity is most evident at grocers. Shoppers at drug stores are more sensitive to travel time than other formats, perhaps due to the convenience orientation of drug stores. We also find that households which shop more at mass merchandisers also shop more in all other formats, suggesting that visits to mass merchandisers do not substitute for trips to the grocery store.

Micro-marketing refers to the customization of marketing mix variables to the store-level. This paper shows how prices can be profitably customized at the store-level, rather than adopting a uniform pricing policy across all stores. Historically, there has been a trend by retailers to consolidate independent stores into large national and regional chains. This move toward consolidation has been driven by the economies of scale associated with these larger operations. However, some of these large chains have lost the adaptability of independent neighborhood stores. Micro-marketing represents an interest on the part of managers to combine the advantages of these large operations with the flexibility of independent neighborhood stores. A basis for these customized pricing strategies is the result of differences in interbrand competition across stores. These changes in interbrand competition are measured using weekly store-level scanner data at the product level. Obviously, this presents a huge estimation problem, since we wish to measure substitution between each product at a store-level. For a chain with 100 stores and 10 products in a category we would need to estimate over 100,000 parameters. To reliably estimate these individual store differences we phrase our problem in a hierarchical Bayesian framework. Essentially, each store-level parameter can be thought of as a combination of chain-level and random store-specific effects. The improvement in estimating this model comes from exploiting the common chain-level component. In addition, we relate these store-specific changes to demographic and competitive characteristics of the store's trading area, which helps explain why these differences are present. These estimated differences in price response are in turn used to set store-level prices. To simplify and focus the problem we limit our attention to everyday price changes (i.e., the prices of products that are not advertised). There are many marketing variables that can be adjusted at a storelevel (e.g., promotions and product assortments); the reason we concentrate upon everyday pricing is driven by its importance in the marketing mix, that most profits are earned on products sold at their everyday price, and the amenability of everyday prices to store-level customizations. A limitation of this approach is that it yields only a partial solution to the retailer's global optimization problem. A challenge for the retailer in implementing micro-marketing pricing strategies is to retain a consistent image while altering prices that adapt to neighborhood differences in demand. Our approach is to search for price changes that leave image unchanged. We argue that a sufficient condition for holding the input to store image constant from everyday pricing is to hold average price and revenues at their current levels. We implement this condition by introducing constraints into the profit maximization problem. Future research into store choice may yield more efficient conditions. A benefit of holding the retailer's image co...

Using weekly scanner data representing 18 product categories, the authors estimated store-specific price elasticities for a chain of 83 supermarkets. They related these price sensitivities to a comprehensive set of demographic and competitor variables that described the trading areas of each of the stores. Despite the inability of previous research to find much of a relationship between consumer characteristics and price sensitivity, 11 demographic and competitive variables explain on average 67% of the variation in price response. Moreover, the authors found that the consumer demographic variables are much more influential than competitive variables. Their findings open the possibility for more effective everyday and promotional pricing strategies that exploit store-level differences in price sensitivity.

No abstract