Consumer Grocery Search: Dimensions and Segments

ABSTRACT - This research investigates the dimensions of consumer grocery search and identifies and profiles shopper segments based on distinctive patterns of search behavior. Results from a large random sample of U.S. grocery shoppers reveal the existence of eight distinct search dimensions (coupons, unit prices, number of stores visited, brand comparisons, advertised specials, word of mouth, checking price tags, and published product evaluations) and five segments (low-effort price/brand comparison, high-effort value seeking, search averse, high search, and time-pressured low search). Demographic differences between segments are identified and areas of strategic relevance of the findings are highlighted.


Sanjay Putrevu and Kenneth R. Lord (1999) ,"Consumer Grocery Search: Dimensions and Segments", in E - European Advances in Consumer Research Volume 4, eds. Bernard Dubois, Tina M. Lowrey, and L. J. Shrum, Marc Vanhuele, Provo, UT : Association for Consumer Research, Pages: 112-118.

European Advances in Consumer Research Volume 4, 1999      Pages 112-118


Sanjay Putrevu, University of Western Australia

Kenneth R. Lord, Mercer University, U.S.A.


This research investigates the dimensions of consumer grocery search and identifies and profiles shopper segments based on distinctive patterns of search behavior. Results from a large random sample of U.S. grocery shoppers reveal the existence of eight distinct search dimensions (coupons, unit prices, number of stores visited, brand comparisons, advertised specials, word of mouth, checking price tags, and published product evaluations) and five segments (low-effort price/brand comparison, high-effort value seeking, search averse, high search, and time-pressured low search). Demographic differences between segments are identified and areas of strategic relevance of the findings are highlighted.


Consumers of most cultures, classes and lifestyles have one experience in commonBgrocery shopping. An understaning of this form of consumer behavior is vital to retailers who often depend on high volume with thin margins, to producers whose products must grab the attention and arouse the preference of the time-pressured buyer, and to consumers who would seek to optimize their own shopping outcomes.

Information derived from the search process frequently influences consumer grocery-purchase decisions. No model or description of the grocery-decision process would be complete without a consideration of the search that accompanies it. Search and choice are bi-directional and inseparable: decision objectives motivate search and give it direction, and search outcomes provide the basis for applying decision rules.

Most prior research in this area models search as a general construct that can be explained as a function of variables such as benefits, costs, price dispersion, budget, time constraints, knowledge, and shopping enjoyment. While such studies, discussed in the next section, have successfully predicted search levels, no published research to date has fully explored the dimensions of consumer grocery search or examined whether distinctive patterns of search behaviors can form a basis for segmenting shoppers. Consumer search, in the context of grocery shopping, can take a number of forms, such as clipping coupons, checking list and/or unit prices, comparing brands or stores, reading articles or advertised specials, or soliciting advice from friends. Understanding which search patterns characterize distinct shopper segments could help marketers identify the information strategies best designed to assist consumers in their efforts and aid consumers in evaluating and improving the efficacy of their decision processes. Hence, this study seeks to: (1) explore the dimensions of consumer search; and (2) identify and profile shopper segments based on distinctive patterns of search behavior.


Most prior research in the area of consumer information search has focused on durable goods. Studies by Newman and Staelin (1972) and Punj and Staelin (1983) are characteristic of this body of literature; they reported the extent of consumer search when purchasing new cars and how it varies in relation to several exogenous variables. Beatty and Smith (1987) investigated the extent of consumer search for other durable goods (e.g., appliances) and tested its relationship with some individual-difference variables.

Some scholars have expanded this line of research to include consumer search for frequently purchased non-durable goods such as grocery products. Jacoby, Chestnut and Fisher (1978) and Moore and Lehmann (1980) conducted experimental studies that documented the extent and type of consumer search for a single grocery product based on direct observation. Carlson and Gieseke (1983) studied consumer search for groceries using panel data from 1956, denoting search as the number of store visits reported by the respondent. Hoyer (1984) and Dickson and Sawyer (1990) directly observed in-store search behaviors relevant to the full basket of goods purchased. Urbany, Dickson and Kalapurakal (1996) examined search from the perspective of price comparisons between stores. Putrevu and Ratchford (1997) developed a normative model of search that captures the tradeoff between costs of search and loss due to inadequate information. Murthi and Srinivasan (1999) modeled search for one grocery product as a function of whether it was on feature or display, which day of the week the shopping trip occurred, and the consumer’s store loyalty, purchase frequency, time availability, income and education. These studies explain the role that a variety of consumer characteristics, market conditions, and marketing and retail strategies play in affecting the amount of consumer grocery search.

As valuable as an understanding of the amount of search is, it tells only part of the story. Equally important is an understanding of the dimensions of earch and distinct groups of consumers who follow identifiable search patterns. To obtain that understanding requires more comprehensive identification and measurement of types of search than were undertaken by most of the above studies, and a different analytical approach than the modeling of search as a single general construct. A few scholars have offered conceptualizations of types of consumer search that, while not offered specifically in the context of grocery shopping, provide some guidance for this study.

Early work relating to new car purchases (Kiel and Layton 1981; Westbrook and Fornell 1979) identified information sources such as retail (store visits), neutral (newspaper or magazine articles), and personal sources (word of mouth). Beatty and Smith (1987) measured four types of search: media, retail, interpersonal, and neutral in the context of appliance purchases. Sambandam and Lord (1995) demonstrated that two of Beatty and Smith’s search typesBmedia and retail searchBrelate to different stages of the automobile-purchase decision. Schmidt and Spreng (1996) conceptually posited five categories of search: marketer controlled, reseller information, third-party independent organizations, interpersonal sources and direct inspection.

If consumers can gather information from numerous sources, it follows that some segments might favor a particular source while others might be predisposed to alternatives. Claxton, Fry, and Portis (1974) studied consumer search behavior for furniture and appliances and reported the existence of three search segmentsBnon-thorough (those who do not search much), thorough balanced (those who seek information from multiple sources), and store intense (those who depend primarily on store visits). Similarly, studies relating to new car purchases have found the following search segmentsBlow, moderate, selective and high search (Kiel and Layton 1981; Furse, Punj, and Stewart 1984; Westbrook and Fornell 1979). While consumers in the low-search segment consult few sources and spend minimal time deliberating their purchases, those in the high-search segment consult most of the sources and undertake a thorough evaluation of the various alternatives. The selective-search segment prefers a subset of the information sources and relies on these to guide decisions.

It is clear from the above literature that distinct consumer search segments exist for durable goods. To date, however, almost no published research has studied whether consumers follow similar strategies for the repeated purchase of non-durable goods. Murthi and Srinivasan (1999) offer preliminary evidence that distinct segments exist in this market, based on their finding of three segments characterized by varying levels of evaluation propensity in the ketchup product category. This research attempts to expand knowledge by studying the search patterns of grocery shoppers across multiple dimensions of search and for the full basket of grocery products.


The approach adopted in this study was to develop self-report measures of search through a survey of grocery shoppers. The use of self-reports, while questionable for durable goods where the measurement may occur months after the original purchase (e.g., Punj and Staelin 1983; Srinivasan and Ratchford 1991), is appropriate for grocery shopping. In this market, search and purchase activities are commonly undertaken each week and information regarding these activities should be reasonably accessible in the minds of consumers. In addition, certain aspects of search behavior require self-reporting due to the difficulty of observing them directly in the marketplace (e.g., scanning newspaper/magazine ads and articles, soliciting advice from friends).

As suggested by Churchill (1979), a multi-step process was undertaken to develop valid and reliable measures of search in a supermarket setting. First, depth interviews, lasting between one and two hours, were conducted with fifteen shoppers andtwo managers to develop a definition of search and to identify the domain of search behaviors needed for a test of that construct’s dimensionality. Based on these interviews, grocery search was defined as "the effort expended gathering information related to the selection and purchase of items in the family grocery basket."

This exercise yielded nine recurring types of search behavior: the extent to which consumers (1) compare unit prices of products, (2) check price tags on considered and selected items, (3) compare competing brands on the various ingredients, (4) look for in-store promotions, (5) clip and use coupons, (6) look for advertised specials in newspapers and store flyers, (7) make multiple store visits, (8) exchange information with friends through word of mouth, and (9) read published product evaluations in newspapers/magazines. These search behaviors derive from a number of motivations: price savings (unit-price comparisons, price checks, in-store promotions, coupons, advertised specials); gauging product availability (store visits, advertised specials); seeking optimal features, attributes or benefits or updating information in an area of interest (brand comparisons, published product evaluations, word of mouth); and even social rewards (word of mouth). They are not constrained exclusively to the pre-purchase stage, but also reflect behaviors that may occur after or between purchase occasions (e.g., published product evaluations, word of mouth).

Several items were developed for each of the nine search behaviors based on literature search and interviews with shoppers and managers. The items were assessed for face and content validity by several faculty members and graduate students. A convenience sample of fifteen grocery shoppers (different from those interviewed earlier) was shown this list of items. Given the definition of each construct and its measures, they were asked whether each item constituted an appropriate measure of the construct and how easily they could respond to it. Based on this exercise, ten items were changed to enhance clarity and face validity.

A pilot study was then undertaken to assess the reliability of the scales developed as measures of the nine search behaviors. The response scale ranged from "never" (1) to "always" (7). Questionnaires were distributed to a convenience sample of 180 grocery shoppers recruited from a women’s club, a church group, acquaintances of the researchers, and university associates (faculty, staff, and students). Items detracting from overall scale reliability were deleted. Thirty-three items remained with at least three items for each of the nine types of search behavior identified earlier (each possessing adequate reliabilityBCronbach a>.80 for eight and approximately .70 for the ninth).

Following the pilot study, data were collected on the search measures and other consumer- and decision-relevant variables from a larger and more representative sample. Given the objective of identifying search segments, three additional constructsBprior planning of grocery purchases, ease of information processing, and time pressureBwere included because of the expectation that they would be associated with extreme patterns of search. It was expected that the planning of purchases in advance and finding it easy to process information about grocery products would be associated with multiple search behaviors, and that consumers who undertake their shopping expeditions under high time pressure would engage in minimal search of any type (the latter relationship was predicted by Titus and Everett [1995] and Schmidt and Spreng [1996], and is consistent with Urbany et al.’s [1996] finding with respect to price search). Also included were measures of readership of the local metropolitan newspaper (relevant to exposure to published store flyers and product articles) and several demographic items.

A mail survey addressed to a random sample (stratified on the basis of income) of two thousand households in a two-county metropolitan area in northeastern United States yielded 612 returned questionnaires, for response rate of 30.6 percent. Of these, 588 were complete in most categories and thus used in the final analysis. The demographic characteristics of the sample were quite representative of the broader population: 76 percent were married; all age and income groups were represented, with 35-44 as the median age category and a median income in the range of U.S. $30,000B$39,999; mean and median number of persons in respondent households was three. A comparison of questionnaires received in the first two weeks after mailing with those received in the last two weeks of the acceptance period showed no significant differences. This finding, together with the demographic similarity between the sample and the larger population, suggests that there was no systematic non-response bias.


Dimensions of Search

Exploratory factor analysis was conducted to assess whether the constructs identified earlier comprised empirically distinguishable dimensions of search. An initial maximum-likelihood factor analysis of the 54 non-demographic questionnaire items yielded thirteen factors with eigenvalues greater than one and numerous large cross-loadings between factors. Ten items that had low loadings across all factors were deleted for purposes of further analysis. This left four indicators of the use of unit prices (Cronbach a=.88), three of price tags (a=.80), four of comparing brands (a=.84), one of in-store promotions, seven of coupons (a=.84), four of advertised specials (a=.89), three of shopping at multiple stores (a=.89), three of word-of-mouth advice (a=.82), five of published product evaluations (a=.87), four of planning purchases in advance (a=.66), one of ease of processing, and five of time pressure (a=.90).

The factor analysis was run again. Ten common factors emerged with eigenvalues greater than one, accounting for 61.3 percent of the variance among the 44 items. Given the expectation of correlation among the various dimensions of search, an oblique rotation was used (promax with Kaiser normalization). The 44 items and the rotated factor loadings associated with them are shown in Table 1.

The factor loadings reveal that eight of the nine search behaviors identified in the earlier stages of the research represent distinct but correlated dimensions of consumer grocery search. Factor 1 primarily captures shoppers’ use of coupons. Six of the seven measures of coupon usage account for the highest loadings on that factor (.73B.91), with the remaining indicator showing the eighth highest loading (.64). Other measures with high loadings (greater than .50) reflect searching for advertised specials. Checking unit prices is the dominant behavior associated with Factor 2. The four indicators of that construct have loadings (.76B.90) that exceed those of all other variables. Factor 3 relates to time pressure. The five indicators of that construct have loadings whose absolute values fall between .72 and .86, while those of all other variables are below .30. The three indicators of the number of store visits have high loadings on Factor 4 (.81B.99). Only one other measure has a loading greater than .50 (checking newspaper ads), and that indicator loads more strongly on the factor it shares with other measures of checking advertised specials. Brand comparisons clearly explain Factor 5, with the only high loadings on that dimension coming from three indicators of that construct (.73B.91). Factor 6 captures shoppers’ tendency to check for advertised specials, with the four measures of that construct receiving the highest loadings (.72B.91) and coupon usage and planning having subordinate influence. The three items measuring word of mouth are the only ones with high loadings on Factor 7 (.71B.80). The three highest loadings for Factor 8 (.79B.82) fall on the price-checking items, with lesser contributions from advertised-special indicators. Factor 9 derives priarily from the prior planning of grocery purchases (the highest two loadingsB.70 and .63), with weaker loadings associated with advertised-special and coupon items. Three published-product-evaluation items yielded the only high loadings for Factor 10 (.73B.84).

The factors representing the various dimensions of search are significantly correlated with each other. This is not surprising given that they denote various aspects of the search process. The strongest correlations are between the coupon-usage, advertised-specials, and planned-purchase factors (all correlation coefficients greater than .60). The two price-checking factors (unit and tag prices) are also correlated with one another (r=.52), and the tag-price factor is correlated with use of store flyers (r=.62). All other correlation values are less than .50.

Search Segments

To define shopper segments on the basis of patterns of search, respondents’ factor scores (derived from the factor analysis described above) were submitted to K-means cluster analysis. Five distinct clusters emerged. Their sample sizes and centers (mean factor scores) are depicted in Table 2.

Consumers in Cluster A (30.6 percent of the sample) are more inclined than average to check unit prices. They also engage in a modest amount of in-store brand comparisons. Feeling some degree of time pressure, however, they perform none of the other search behaviors at an above-average level. Some of the more time-consuming behaviors occur in this cluster at levels substantially below the sample average (e.g., multiple store visits, word of mouth, published product evaluations). This cluster is labeled the low-effort price/brand-comparison segment.

Cluster B (17.8 percent of the sample) is highest in comparative shopping across multiple stores and commonly checks advertised specials. Respondents in this cluster are more likely than average to check price tags, use coupons and word of mouth, and plan their grocery purchases in advance. But despite engaging in fairly extensive and varied search behaviors, they are unlikely to engage in non-price brand comparisons or to read published brand evaluations. They represent a high-effort value-seeking segment.

Cluster C, 12.0 percent of the sample, engages in minimal search despite being no more time pressured than average. Members of this cluster are below average on all search dimensions and the lowest of the five clusters in checking unit prices and reading published product evaluations. They can be considered a search-averse segment.

Cluster D is highest in size (31.9 percent of the sample), in its propensity for prior planning of grocery purchases, and in most search activities (coupons, unit prices, brand comparisons, store flyers, word of mouth, price tags, and published brand evaluations). Consumers in this cluster are low in time pressure. They appear to enjoy all forms of search and are labeled the high-search segment.

The final group, Cluster E, is the smallest in the sample (7.8 percent). It is the highest in time pressure and is low in all dimensions of search (lowest planning purchases, visiting multiple stores, checking price tags, and using coupons, store flyers and word of mouth). Consumers in this cluster are a time-pressured low-search segment.

Additional data analysis was conducted to arrive at a demographic profile of the five segments defined above. No significant gender or occupational-status (full-time, part-time, unemployed, retired, etc.) differences were observed between segments. Marital status was significantly associated with cluster membership (c2=21.24, p<.05). A plurality of singles (42.5 percent) fall within the low-effort, price/brand-comparison segment. The majority of married respondents are divided almost exactly evenly between that segment (30.6 percent) and the high-search cluster. The balance shifts in the direction of the high-search cluster among divorced/separated respondents (35.7 percent versus 26.8 percent in the low-effort prce/brand-comparison segment). The high-search segment captures a stronger plurality of widowed respondents (46.9 percent).

Age is also relevant to cluster membership (c2=32.56, p<.05). Both the low-effort price/brand-comparison and high-search segments claim a higher proportion of respondents in younger than in older age categories. Membership in the low-effort price/brand-comparison segment declines systematically with age (37.5 percent of respondents in the youngest categoryBunder 25, to 15.0 percent of those in the oldestB65 or older). Similarly, the high-search segment has about 13 percent of respondents in each of the two youngest categories and only 5 to 9 percent of those in the four older groups. Other segments do not show systematic age differences.

Educational level is shown to differ significantly between the segments (F=5.84, p<.001). A post-hoc Duncan test (p<.05) reveals that this is driven by a mean educational level in the time-pressured low-search segment (15.38 years) that is higher than that associated with any other cluster (13.39 to 13.93 years).

Household size also varies significantly between clusters (F=2.80, p<.05). Duncan test results show significantly fewer household members in the time-pressured low-search segment (mean 2.65) than in any of the other clusters (3.08 to 3.38).

Finally, there are significant income differences between segments (F=5.79, p<.001). Duncan test results show significantly higher income in the time-pressured low-search segment (about $50,000) than in any of the other segments. Respondents in Clusters A (low-effort/price-brand comparison) and C (search-averse) had a higher mean income (about $40,000) than those in the high-search segment (about $30,000).






Using a large, representative sample from a major metropolitan area, this study established the dimensionality of the consumer-grocery-search construct and defined customer segments characterized by varying patterns across search dimensions. Coupons, unit prices, number of stores visited, brand comparisons, advertised specials, word of mouth, checking price tags, and published product evaluations emerged as distinct search dimensions. This finding offers understanding of the way in which the search types or categories revealed in earlier research for other product categories (e.g., Beatty and Smith 1987; Schmidt and Spring 1996) applies to consumer grocery-shopping behavior. The distribution of the search dimensions in the population was shown to coalesce around five patterns that represent viable market segments: low-effort price/brand comparison, high-effort value seeking, search averse, high search, and time-pressured low search. This result expands upon the three segments identified recently by Murthi and Srinivasan (1999) in the ketchup category. Adding to the value of search-behavior differences as a basis for segmentation is the finding of systematic demographic differences between segments.

By examining the combination of demographic and search characteristics in the different segments, marketers can better understand which of their targeted consumers are likely to search and in what ways, allowing for more efficient educational and promotional efforts. Those interested in helping consumers to optimize their expenditures can study the results for a clearer understanding of the search dimensions that are and are not benefiting those in greatest need (e.g., lower income, elderly, divorced or separated, less educatedBpopulation groups who fortunately do not fall disproportionately in the search-averse segment) and use that insight to formulate more effective educational campaigns. And by becoming overtly aware of their own segment membership, consumers can consciously evaluate whether their search practices (or lack thereof) are yielding the desired results in their own consumption behavior.

Additional research is needed to confrm or repudiate the validity of these results in other markets with different cultural values, household profiles, economic conditions and marketing practices. Furthermore, though a strength of this study is its expansion of the search behaviors commonly investigated in the marketing literature, future research could yield more insights by exploring a still more comprehensive array of potential dimensions of consumer grocery search (e.g., the Internet). Efforts at constructing a psychographic profile of search segments would be useful. Finally, a closer look is needed at the search segments reported here, to investigate the motivations behind their search patterns and the extent to which the outcomes match those motivations.


Beatty, Sharon E. and Scott M. Smith (1987), "External Search Effort: An Investigation Across Several Product Categories," Journal of Consumer Research, 14 (June), 83-95.

Carlson, John A. and Robert J. Gieseke (1983), "Price Search in a Product Market," Journal of Consumer Research, 9 (March), 357-365.

Churchill, Gilbert A. (1979), "A Paradigm for Developing Better Measures of Marketing Constructs," Journal of Marketing Research, 16 (February), 64-73.

Claxton, John D., Joseph N. Fry, and Bernard Portis (1974), "A Taxonomy of Perpurchase Information Gathering Patterns," Journal of Consumer Research, 1 (December), 35-47.

Dickson, Peter R. and Alan G. Sawyer (1990), "The Price Knowledge and Search of Supermarket Shoppers," Journal of Marketing, 54 (July), 42-53.

Furse, David H., Girish N. Punj, and David W. Stewart (1984), "A Typology of Individual Strategies Among Purchasers of New Automobiles," Journal of Consumer Research, 10 (March), 417-431.

Hoyer, Wayne D. (1984), "An Examination of Consumer Decision Making for a Common Repeat Purchase Product," Journal of Consumer Research, 11 (December), 822-829.

Jacoby, Jacob, Robert W. Chestnut, and William A. Fisher (1978), "A Behavioral Process Approach to Information Acquisition in Nondurable Purchasing," Journal of Marketing Research, 15 (November), 532-544.

Kiel, Geoffrey C. and Roger A. Layton (1981), "Dimensions of Consumer Information Seeking," Journal of Marketing Research, 18 (May), 233-239.

Moore, William L. and Donald R. Lehmann (1980), "Individual Differences in Search Behavior for a Nondurable," Journal of Consumer Research, 7 (December), 296-307.

Murthi, B. P. S. and Kanan Srinivasan (1999), "Consumers’ Extent of Evaluation in Brand Choice," Journal of Business, 72 (April), 229-256.

Newman, Joseph W. and Richard Staelin (1972), "Prepurchase Information Seeking for New Cars and Major Household Appliances," Journal of Marketing Research, 9 (August), 249-257.

Punj, Girish N. and Richard Staelin (1983), "A Model of Consumer Information Search for New Automobiles," Journal of Consumer Research, 9 (March), 366-380.

Putrevu, Sanjay and Brian Ratchford (1997), "A Model of Search Behavior with an Application to Grocery Shopping," Journal of Retailing, 73 (4), 463-486.

Sambandam, Rajan and Kenneth R. Lord (1995), "Switching Behavior in Automobile Markets: A Consideration Sets Model," Journal of the Academy of Markieting Science, 23 (Winter), 57-65.

Schmidt, Jeffrey B. and Richard A. Spreng (1996), "A Proposed Model of External Consumer Information Search," Journal of the Academy of Marketing Science, 24 (Summer), 246-256.

Srinivasan, Narasimhan and Brian T. Ratchford (1991), "An Empirical Test of a Model of External Search for Automobiles," Journal of Consumer Research, 18 (September), 233-242.

Titus, Philip A. and Peter B. Everett (1995), "The Consumer Retail Search Process: A Conceptual Model and Research Agenda," Journal of the Academy of Marketing Science, 23 (Spring), 106-119.

Urbany, Joel E., Peter R. Dickson, and Rosemary Kalapurakal (1996), "Price Search in the Retail Grocery Market," Journal of Marketing, 60 (April), 91-104.

Westbrook, Robert A. and Claes Fornell (1979), "Patterns of Information Sources Usage Among Durable Goods Buyers," Journal of Marketing Research, 16 (August), 303-312.



Sanjay Putrevu, University of Western Australia
Kenneth R. Lord, Mercer University, U.S.A.


E - European Advances in Consumer Research Volume 4 | 1999

Share Proceeding

Featured papers

See More


J1. The Effect of Identity Abstractness on Information Processing Styles

Woojin Choi, University of Seoul
Min Jung Kim, Manhattan College
HyukJin Kwon, University of Seoul
Jiyun Kang, Texas State University

Read More


I12. The Effect of Susceptibility-Induced Threat in the Preventative Communication

Moon-Yong Kim, Hankuk University of Foreign Studies

Read More


L13. The Recipient Effect on Consumers’ Preference for Products Displayed in Different Horizontal Locations

Sheng Bi, Washington State University, USA
Nik Nikolov, Washington State University, USA
Julio Sevilla, University of Georgia, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.