Pricing, especially the area of discounting, poses many practical problems and continues to generate academic interest. In this exploratory study, we proposed an analysis format based on multi-channel purchasing and a method to calculate the amount of discount. Our principal contribution is using single-source panel data to calculate the discounts for several stores. First, we presented our definition of discounts. Next, as few consumers (so-called cherry-pickers) accounted for a large portion of the discounts, we confirmed that discounts follow the Pareto principle. Further, we showed how consumers used different chains; we identified the discount-sensitive consumers. We find that consumers who tend to buy high-priced brands are of key importance for sales and revenues of some chains. In the context of shopping behavior, shopping trip type is one of the main concerns. In a relatively early study, Kahn and Schmittlein (1989) showed quick trips and regular trips; however, they did not focus on discounts. Since the 1990s, researchers have been focusing on discounts and multi-channel shopping trips. Walters and Jamil (2003) explored shopping trip type and discount; however, their data are restrictive regarding the purchase period and chain. Fox and Hoch (2005) also showed how the difference in prices across two grocery store chains on the same day was distributed and how some consumers could exhibit cherry-picking behavior. Nevertheless, their research data are not comprehensive and are rather limited to two popular grocery stores. The same limitation is applicable to other studies in the literature (e.g., Bell, Ho and Tang 1998). One of the contributions of this study is our data. We used single-source panel data, acquired through the service of Macromill, Inc. The monitors of this service can use portable code scanners to read JICFS(JAN Item Code File Service)codes anytime and anywhere. Since we focus on multi-channel shopping, we used only data on food purchases. We included 6 million purchase transactions that covered all food categories in 2012 for 6,422 individuals who live in the metropolitan area around Tokyo. There was, essentially, no gender and age bias in the data. However, the data do not provide locational information of the store and consumer. Location is a very important factor for determining a shopping trip (e.g., Arentze, Oppewal and Timmermans 2005). This is one of the limitations of our research. The first step in our analysis is defining the discounts on individual items. To mitigate the influence of extremely high unit prices, we calculated the discount as the difference between the third-quartile price and the purchase price. The next step, we identified cherry-pickers from cross table of sales decile and discount decile. We examined demographic feartures, women relatively exhibit higher cherry-picking behavior than men, and there is a distinct relationship between household income and cherry-picking. As the income level rises, the ratio of cherry-pickers decreases. As for the relationship with age, we find that the ratio of cherry-pickers is the highest among consumers who are in their 30s. However, this ratio decreases with age. Examining the ratio of cherry-pickers by chain, we find that the ratio is higher in EDLP type than in Hi-Lo type chains or High-quality type chains. To simplify the purpose of our study, we deal with the milk category. The reasons for selecting the milk category are that it is one of the most popular food categories in Japan and that the Japanese milk category is assumed as a loss leader. We present some results from a k-means cluster analysis and show how the customer segments utilize each channel in the milk category. This allows us to observe each segment is more discount-conscious in each channel. Among Japanese milk brands, there are some popular and high-priced (above 200 yen / 1 liter pack) brands that rarely offers bargain sales. And the customer who buy that high-priced brands frequently is known as a loyal one. We showed the frequency of purchases for high-priced brands and the frequency (in days) of chain visits. We compared a well-known EDLP type and low-priced supermarket chain and perceived as average or slightly expensive chain. Consumers who habitually purchase the high-priced brand are likely to be loyal customers of the supermarket chain, but this does not hold for the low-priced chain. We will present the results from further analysis and details at the conference.

Many measurement methods for understanding consumers’ acceptable price range have been developed. Among these, Price Sensitivity Meter (PSM) is one of the most popular. It has been regarded as a convenient research method because of the ease of data collection and data processing. In particular, PSM requires only four questions to determine the price range. Nevertheless, it also has some problems from a theoretical viewpoint. The purpose of the present research is to develop a new price research method for measuring consumers’ acceptable price range. In particular, applying survival analysis to data prepared for PSM, Japanese consumers’ price acceptance ranges for several categories were estimated.