View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by School of Hotel Administration, Cornell University

Cornell University School of Hotel Administration The Scholarly Commons

Articles and Chapters School of Hotel Administration Collection

12-1999

Designing and Positioning Food Services for Multicultural Markets

Rohit Verma Cornell University, [email protected]

Madeleine E. Pullman Southern Methodist University

John C. Goodale Ball State University

Follow this and additional works at: https://scholarship.sha.cornell.edu/articles

Part of the Hospitality Administration and Management Commons

Recommended Citation Verma, R., Pullman, M. E., & Goodale, J. C. (1999). Designing and positioning food services for multicultural markets [Electronic version]. Cornell Hotel and Restaurant Administration Quarterly, 40(6), 76-87. Retrieved [insert date], from Cornell University, School of Hospitality Administration site: http://scholarship.sha.cornell.edu/articles/60/

This Article or Chapter is brought to you for free and open access by the School of Hotel Administration Collection at The Scholarly Commons. It has been accepted for inclusion in Articles and Chapters by an authorized administrator of The Scholarly Commons. For more information, please contact [email protected].

If you have a disability and are having trouble accessing information on this website or need materials in an alternate format, contact [email protected] for assistance. Designing and Positioning Food Services for Multicultural Markets

Abstract [Excerpt] Businesses that seek to develop an appropriate operations strategy for serving a multicultural customer market face challenges that are distinct from businesses that serve a relatively homogeneous local market. While the strategic implications of expanding services from a domestic market to international locations have been well documented, the issue of dealing with multinational customers at a single location has largely been neglected by researchers, as far as we can determine.

This paper attempts to fill the esearr ch gap by presenting a method for determining the extent to which restaurant managers should maintain standard menus and food items, as opposed to customizing their operations for different ethnic and cultural groups. To that end, we applied a customer-based approach to help managers at four international-airport food outlets to improve their food- service revenues from their three major passenger groups: English-, Japanese-, and Spanish-speaking customers. In this case, language preference was used as a proxy for cultural identity. We submit that although there are many differences among, say, English-speaking peoples, they are more similar to each other than they are to, say, Spanish speakers. Moreover, the language a person speaks is a substantial factor in trying to communicate in a particular location. One can guess that Japanese speakers in the United States, for example, might experience more language barriers than either English or Spanish speakers.

We present an approach for modeling the preferences of different cultural groups, evaluating the differences among the groups, and determining a strategy to maximize market share for each of the four food-service providers that we studied. Indeed, one food-service vendor implemented our study's recommendations and enjoyed a substantial revenue gain over the previous year's same-period sales. We believe that the method we propose has valuable implications for any service provider who must consider operating strategies for a multicultural or multinational site, although we focus on the distinct concept of a domestic foodservice business that serves a multicultural market.

Keywords food service, restaurant management, menus, multicultural markets, sales

Disciplines Hospitality Administration and Management

Comments Required Publisher Statement © Cornell University. Reprinted with permission. All rights reserved.

This article or chapter is available at The Scholarly Commons: https://scholarship.sha.cornell.edu/articles/60 Designing and Positionin; Food Services for Multicultural Markets

People waiting in an airport are far from being a captive audience, and it behooves

by Rohit Verma, food-service operators to determine what those customers want. Madeleine E. Pullman, and John C. Goodale D I • usinesses that seek to develop food-service revenues from their WK0 an appropriate operations three major passenger groups: strategy for serving a multicultural English-, Japanese-, and Spanish- customer market face challenges speaking customers. In this case, that are distinct from businesses that language preference was used as a serve a relatively homogeneous local proxy for cultural identity. We sub­ market. While the strategic impli­ mit that although there are many cations of expanding services from differences among, say, English- a domestic market to international speaking peoples, they are more locations have been well docu­ similar to each other than they are mented, the issue of dealing with to, say, Spanish speakers. Moreover, multinational customers at a single the language a person speaks is a location has largely been neglected by researchers, as far as we can Rohit Verma, Ph.D., is an assistant determine. professor in the Kellstadt Graduate This paper attempts to fill the School of Business at DePaul University research gap by presenting a method «rverma@u>ppost.depaul.edu». for determining the extent to which Madeleine E. Pullman, Ph.D., is an restaurant managers should maintain assistant professor in the Cox School of standard menus and food items, as Business at Southern Methodist Univer­ opposed to customizing their oper­ sity «[email protected]». ations for different ethnic and cul­ John C. Goodale, Ph.D., is an assistant tural groups. To that end, we applied professor in the College of Business at a customer-based approach to help Ball State University «[email protected]». managers at four international- airport food outlets to improve their ©1999. Cornell University

76 , HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY FOOD AND m MANAGEMENT substantial factor in trying to com­ We suggest an approach for deter­ be observed and tabulated, and municate in a particular location. mining the extent of both service- (3) observed differences have signifi­ One can guess that Japanese speakers product and process-attribute stan­ cant bearing on both consumer be­ in the United States, for example, dardization versus customization in havior and the strategic decision­ might experience more language multicultural settings. Specifically, makers in firms. Research has barriers than either English or we: focused largely on the standardiza­ Spanish speakers. • Present an approach for identify­ tion level of marketing strategies We present an approach for mod­ ing the needs and preferences of during globalization.3 Research has eling the preferences of different customers of different nationali­ also shown that customers from cultural groups, evaluating the dif­ ties or cultures; different cultures and nationalities ferences among the groups, and de­ • Demonstrate how key differences have different expectations from termining a strategy to maximize between the needs and prefer­ services and perceptions of the ac­ market share for each of the four ences of different customer tual service delivered. In 1998 food-service providers that we stud­ groups may be quantified; and Donthu and Yoo studied the effects ied. Indeed, one food-service vendor • Demonstrate the influence of of consumers' cultural orientation implemented our study's recommen­ different operating strategies on on their expectations of bank ser­ dations and enjoyed a substantial market performance in multi­ vices.4 They found relationships revenue gain over the previous year's cultural environments. between groups' cultural orientation same-period sales. We believe that We first discuss the previous re­ and their expectations of banks' the method we propose has valuable search on multicultural customer service-quality dimensions of reli­ implications for any service provider groups' different expectations and ability, responsiveness, empathy, and who must consider operating strate­ perceptions of services. We then assurance. Lee and Ulgado discov­ gies for a multicultural or multina­ outline our method for addressing ered that South Korean and Ameri­ tional site, although we focus on the those issues in the formulation of can consumers held different per­ distinct concept of a domestic food- operations strategy and service de­ ceptions of fast-food services.5 In service business that serves a multi­ sign. Next, we describe how our their study, American consumers cultural market. method was applied to food-service preferred corporate reputations or improvements at a major interna­ brand names, low prices, and consis­ Process and Product tional airport terminal in the United tent quality, while Korean customers, We believe that a major component States (Chicago's O'Hare). We pro­ on the other hand, valued reliability of establishing a strategy for re­ vide the implications and recom­ and empathetic employees. sponding to different cultural groups mendations for managers and discuss Given that different cultural ex­ is to determine both process and the results from the implementation. pectations and perceptions of ser­ product attributes. Process attributes Finally, we discuss the significance of vices can exist, managers need to include waiting time, employee in­ our findings and draw conclusions know whether their service strategy teractions with customers, reliability, based on the study. should, for example, emphasize or and speed of completing transac­ target the needs of one group (in tions.' Product attributes cover tan­ The Matter of Culture terms of revenue potential or other gible items such as employee appear­ Many business researchers have stud­ objectives), all groups simultaneously, ance, building design, cleanliness ied national character or cultural or different groups at different times. levels, branding variety, and printed differences.2 According to Clark, materials. Depending on the service national character is defined as en­ *' See: Clark, pp. 66—79; S.C.Jain, "Standardiza­ concept and operating strategy, one during personality characteristics tion of International Marketing Strategy-: Some would expect that certain service among the populations of particular Research Hypotheses," JoHniuf of Marketing, Vol. attributes should be adjusted for nation states. Cross-national studies 53, No. 1(1989), pp. 70-79; and R. Martcnson, "Is Standardization of Marketing Feasible in different cultural segments, while are valuable in international market Culture-bound Industries? A European Case others have universal appeal. Many settings because (1) national differ­ Study," International Marketing Review, Vol. 4, times, however, restaurants' attributes ences exist, (2) these differences can No. 3 (1987), pp. 7-17. 4N. Donthu and B. Yoo,"Cultural Influence on are standardized, having been pat­ Service Quality Expectations," Journal of Service terned after successful domestic or 2 For an extensive review, sec: T. Clark, Researched l,No.2 (1998),pp. 178-186. international operations. "International Marketing and National Charac­ ' M. Lee and F. Ulgado, "Consumer Evaluations ter: A Review and Proposal for an Integrative of Fast-food Services: A Cross-national Com­ Theory," Journal of Marketing, Vol. 54, No. 4 parison," Journal of Services Marketing, Vol. 11, 1 Often referred to as "throughput." (1990), pp. 66-79. No. 1 (1997), pp. 39-52.

December 1999 • 77 Service differentiation. Strate­ Managers often have some ad­ including marketing, operations gies for service-differentiation fall vance cues about the mix of their management, transportation, urban on a continuum from "one size fits customer groups. For example, air­ planning, hospitality, and natural- all" (e.g., McDonald's success in lines and airports can estimate the resource economics.11 Typical ser­ adapting to multinational environ­ mix of nationalities from historic vice attributes are measurable items ments with minimal modifications) records and national airlines' times such as waiting time and price, or to totally personalized experiences of arrivals and departures. This in­ items more challenging to quantify for each customer, or the so-called formation can be used to customize such as facility cleanliness or em­ "mass customization" that has lately services for those customers during ployee empathy. By describing a come into vogue. certain time periods or in certain service in terms of appropriate at­ Ideally, if the service firm has the parts of the airport. An example of tributes, DCA can be used to pre­ ability to customize the service for this concept would be creating a dict market share and profit from each client, this approach could sub-brand for a group with a high any hypothetical service design address any cultural or personal yearly revenue potential. Hilton in a competitive environment. preferences. One approach to per­ International Hotels did this in Discrete-choice experiments in­ sonalized service is known as "in­ 1992 with "Wa No Kurutsogi," a volve careful designs of service dustrialized intimacy," as proposed special amenity and service package profiles (with specific levels of at­ by Kolesar, Van Ryzin, and Cutler.6 aimed at its Japanese clientele.'" tributes) in choice sets in which Those researchers suggest that in­ Some 32 Hiltons offered guest- two or more service profiles (alter­ dustrialized intimacy in a mass service materials written in Japanese, natives) are offered to would-be market is achieved through infor­ Japanese food and beverage offer­ customers. The respondents are mation systems that track an indi­ ings, and special slippers and yukata. asked to evaluate the options and vidual customer's history and pref­ The upshot of the research we choose one (or none). Each subject erences, thus allowing a personal have seen is that some services can in a DCA experiment typically touch in service. Examples of com­ be standardized while others should receives several choice sets to evalu­ panies using information systems in be customized to address individual ate (e.g., 8 to 32 sets) with two or this manner include Ritz-Carlton' customers' needs. To address the more hypothetical services to and British Airways.8 While this issue of how much customization choose from in each of the sets.12 approach works well with a base of is appropriate, managers need to DCA applications based on repeat or loyal customers, industri­ determine (1) the customer's expec­ choice experiments typically involve alized intimacy would be hard to tations of each important service the following steps: (1) identifica­ implement in anonymous service attribute, (2) the attributes that are tion of attributes, (2) specification settings or transient markets, such appropriate for standardization for of attribute levels, (3) experimental as at a fast-food outlet in an airport all groups or customization for a design, (4) presentation of alterna­ terminal. Additionally, a strategy specific group, and (3) the eco­ tives to respondents, and (5) estima­ that attempts to give the best ser­ nomic feasibility of customizing tion of choice model. A number of vice to every customer may be pertinent attributes. past studies have shown that, in gen­ unnecessary and expensive.9 eral, the market-share predictions Determining Customer Preferences generated from multinomial-logit 6 P. Kolesar, G. Van Ryzin, and W. Cutler, One method for determining the "Creating Customer Value through Industrialized 11 Intimacy," Strategy and Business, Vol. 12 (1998), attribute levels of a new service See: Rohit Verma and Gary Thompson. pp. 33-43. involves modeling customer prefer­ "Basing Service Management on Customer 7 Determinants: The Importance of Hot Pizza," See: N. Klein, W E. Sasser, and T.O.Jones, ences in response to experimentally "The Ritz-Carlton: Using Information Systems Cornell Hotel and Restaurant Administration to Better Serve the Customer," Harvard Business designed service profiles. This ap­ Quarterly, Vol. 37, No. 3 (June 1996), pp. 18-23; School Case 9-395-064 (1995); and Dube and proach, commonly known as proba­ and M. Pullman and W. Moore, "Optimal Service Design: Integrating Marketing and Operations Renaghan, "Surprisingly Simple Routes to the bilistic discrete choice analysis Top," in this issue of Cornell Quarterly, pp. 34-41. Perspectives tor Capacity Decisions," International "N.Klein and W.E. Sasser,"British Airways: (DCA) has been used to model Journal of Service Industry Management, in press Using Information Systems to Better Serve the choice processes of decision-makers '(1999)/ 12 For a detailed theoretical and statistical Customer," Harvard Business School Case in a variety of academic disciplines, 9-395-065 (1994). background of DCA, see: M. Ben-Akiva and '' R.E. Kordupleski, R.T. Rust, and A.J. Zahorik, S.R. Lerman, Discrete Choice Analysis (Cam­ "Why Improving Quality Doesn't Improve bridge, MA: MIT Press, 1991); or D. McFadden, Quality (or, Whatever Happened to Marketing?)," '''See: Glenn Withiam,"Wa No Kutsurogi," "The Choice Theory Approach to Market California Management Review, Spring 1993, Cornell Hotel and Restaurant Administration Research," Marketing Science, Vol. 5, No. 4 (1986), pp. 82-95. Quarterly, Vol. 33, No. 1 (February 1992). p. 14. pp. 275-297.

78 [(]R\ ELL HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY FOOD AND xsnxsx MANAGEMENT

(MNL) or more-advanced econo­ Hence, we had the opportunity to metric models (e.g., nested-logit characterize the entire market envi­ model) based on discrete-choice ronment, a situation we believe is analysis are extremely accurate.13 unlike other potential study loca­ Although design of choice experi­ tions. (3) Potential customers were ments and estimation of MNL actually able to see all the food- models require sophisticated train­ service vendors at the same time. One food-service vendor ing and skills, implementing the It is reasonable, therefore, to believe estimated model(s) in spreadsheet- that the issues on which customers implemented our study's based decision-support systems is focus are marketing (brand name, fairly easy. Hence, DCA has become price, promotion) and operations recommendations and enjoyed immensely useful for practicing (waiting time, service variety, qual­ managers. In this study we illustrate ity) . (4) The flight schedule at the use of DCA in service design O'Hare yields a reasonably predict­ a substantial revenue gain for a multinational food-service able arrival pattern of different cus­ market. tomer groups. This was beneficial over the previous year's for data collection and observing Food-service at customers' choices. (5) The few same-period sales. O'Hare's International Terminal food-service vendors were essen­ We conducted a detailed case analy­ tially fast-food operators. Conse­ sis of the food-court operations at quently, the DCA could include Chicago's O'Hare, one of North a majority of service possibilities— America's largest and busiest inter­ again, an uncommon situation. national airport terminals. The in­ About revenues. One of the formation presented here is based reasons that the terminal authority on the operational data collected was interested in the success of its from the senior management team food-service vendors is that they of the food-service operations at the are a source of revenue. Landing terminal, the managers or owners of fees may not be a source of great the individual food-service vendors, revenue increases because the air­ and customer-choice data collected port is already heavily subscribed, from three different cultural groups but commercial operations, particu­ during 1998. larly revenues derived from airport We chose to conduct an in-depth retailing, has provided significant analysis of the food-service opera­ revenue opportunities for operating tions at O'Hare for the following authorities. While most major inter­ reasons. (1) Many people from a national terminals derive the major­ variety of nations and cultures pass ity of their revenues from duty-free through O'Hare or visit Chicago, retail operations, the second biggest so that a relatively large number source is revenues from food and of observations can be generated beverage operations.14 Lease pay­ for each cultural group. (2) Because ments for the food court in ques­ access to international airport ter­ tion are based on a percentage of minals is restricted, customers are the vendors' sales. Thus, if our study limited to choosing a food-service finds ways to increase revenues, it vendor from those at the terminal. not only could benefit the vendors themselves, but also airport-terminal operations. 1' See: Ben-Akiva, op. cit.; RE. Green and A.M. Krieger,"Individualized Hybrid Models for Conjoint Analysis," Management Science, Vol. 42, No. 6 (1996), pp. 850-867; andJJ. Louvieve and "P. Freathy and F. O'Connell,"Supply-chain H. Timmermans,"Stated Preference and Choice Relationships within Airport Retailing." Interna­ Models Applied to Recreation Research: A tional Journal of Physical Distribution and Logisitics Review," Leisure Sciences, Vol. 12 (1990;, pp. 9-32. Management, Vol. 28, No. 6 (1998), pp. 451-462.

December 1999 • 79 Exhibit 1 Restaurant attributes and levels

Restaurant 1— Restaurant 2— Restaurant 3— Restaurant 4— Attributes, levels, and "Burger" "Pizza" "Dogs" "Deli" (experimental design code) Brand name Level 1 (-1) Local chain Local chain Generic food items Local chain Level 2 (+1) McDonald's Pizza Hut or Domino's La Prefreda or Goya Subway or Boston Market

Variety Level 1 (-1) Burger, fries, ice cream Pizza Hot dog, fries, nachos Sandwich, soup, ice cream Level 2 (0) + chicken nuggets and + lasagna, pasta + burritos, tacos + Udan noodle soup, (add to Level 1 items) salads salads

Level 3 (+1) + special burgers and + salads, soups + tamales, enchiladas + sushi, simple Asian (add to Level 1 & 2 items) sandwiches dishes

Wait to order Level 1 (-1) 0 to 2 minutes 0 to 2 minutes 0 to 2 minutes 0 to 2 minutes Level 2 (0) 3 to 4 minutes 3 to 4 minutes 3 to 4 minutes 3 to 4 minutes Level 3 (+1) 5 to 6 minutes 5 to 6 minutes 5 to 6 minutes 5 to 6 minutes

Service time Level 1 (-1) 0 to 2 minutes 0 to 2 minutes 0 to 2 minutes 0 to 2 minutes Level 2 (0) 3 to 4 minutes 3 to 4 minutes 3 to 4 minutes 3 to 4 minutes Level 3 (+1) 5 to 6 minutes 5 to 6 minutes 5 to 6 minutes 5 to 6 minutes

Menu language Level 1 (-1) English English English English Level 2 (0) English and Spanish English and Spanish English and Spanish English and Spanish Level 3 (+1) English, Spanish, Japanese English, Spanish, Japanese English, Spanish, Japanese English, Spanish, Japanese

Picture display Level 1 (-1) No No No No Level 2 (+1) Yes Yes Yes Yes

Price (including drinks) Level 1 (-1) $4 $4 $4 $4 Level 2 (0) $7 $7 $7 $7 Level 3 (+1) $10 $10 $10 $10

16 The Food Court The design of food and beverage recognized brand names. Other factors such as waiting time and Four food-service companies, in­ services for a multinational setting is prices for food have been shown to cluding McDonald's, are situated challenging, due to cultural differ­ mean different things to different in a single food court on O'Hare's ences in the way people react to national groups. For example, a international-departure level. uncertainties and ambiguities inher­ 15 short waiting time and low prices This particular terminal handles ent in daily living. Put simply, for food are more generally impor- only international flights on non- international travelers generally domestic airlines. The regional attempt to avoid uncertainty by choosing food from their respective breakdown of the departing airlines '" See: Clark, pp. 66-79; T. Clark. D. at that terminal is 20 percent Latin cultures or from internationally Rajaramam, and T. Smith, "Toward a Theory of International Services: Marketing Intangibles in a America, 10 percent Japan and 15 See: G. Hofstede, Culture's Consequences World of Nations," Journal ofInternational Market­ Asia, 60 percent Europe, and 10 (Beverly Hills. CA: Sage Publications, 1980); ing, Vol. 4. No. 2 (1996), pp. 9-28;Jam, pp. 70-79; percent Middle East. and Lee and Ulgado, pp. 39-52. and Martenson, pp. 7-17.

80 HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY FOOD AND xnxxx MANAGEMENT tant for people from the United Decision Attributes the selected attribute levels for each States than for people from some The selected service attributes were of the four types of restaurants. 17 European and Asian cultures. as follows: Note that attribute levels for "brand Data collection. We conducted • brand name (i.e., the restaurant name" and "variety" are different the customer-preferences survey is either part of a branded inter­ for each of the four concepts. with just three groups—those cus­ national chain or it sells branded After identifying the attributes tomers flying on airlines based in food items), and their levels, we employed a Latin America,Japan, and Europe— • menu variety (i.e., the number fractional-factorial design to gener­ because they represented the major­ ate 18 experimental profiles for of different food items served 2 ity of the customers at the interna­ by a particular restaurant), each restaurant concept. " A full tional terminal's food-service opera­ • wait time before ordering, experimental design would involve tions. The survey was administered • service time, and 486 possible profiles for each res­ in the terminal's food court and • price of a standard meal and taurant—and render the study im­ practical. Instead, fractional-factorial waiting areas. All questionnaires and drink. or partial-experimental designs interviews were conducted in the Not all customers spoke, under­ reduce the respondent's task to a respondent's language (English, stood, or read English. To overcome more manageable size by using only Spanish, or Japanese) depending on language barriers, managers wanted a sample group of all potential pro­ the customer's preference. In this us to consider communication- files found in design catalogues.21 study, primary language represents related attributes, and so we added 1 The profiles contained different a proxy for national culture. * menu language and picture display levels of each of the seven attributes The first stage in the DCA study of popular meal items as the re­ described above. Each choice set involved identification of relevant maining two attributes. product and service attributes and contains one profile for each of Attribute levels. Next, we the four types of restaurants. those attributes' possible levels. For identified the relevant levels for this we interviewed 100 randomly each of the seven attributes selected As illustrated with the example selected airline passengers to identify for the study. The final attribute in Exhibit 2 (on the next page), the important attributes they con­ levels reflect realistic values that we each respondent was asked to sidered in choosing a food vendor. selected after detailed discussions choose one out of the five possible Based on their responses, we identi­ with terminal managers. Brand choices (one of the four restaurants fied seven attributes that most poten­ name, for example, was a two-level or none of them) for each of the 18 experimentally generated choice tial customers considered when attribute (local chain versus na­ sets. Another part of the survey selecting a food vendor at this par­ tional chain). On the other hand, dealt with respondents' characteris­ ticular airport terminal. This kind of menu variety had three levels, to tics such as age, gender, family size, qualitative-data-collection approach reflect a low, medium, or large amount spent at food court, and for the identification of relevant number of possible menu choices. items purchased. service attributes has been recom­ The remainder of the attributes 19 mended by other researchers. In likewise had either two or three Both the preliminary and final fact, Griffin and Hauser concluded levels. questionnaires were originally writ­ that it takes only as few as 30 re­ One of the considerations in this ten in English, translated into Japa­ spondents to get a majority oi rel­ study is the limitations of space and nese and Spanish by two bilingual evant product and service attributes. legal and regulatory constraints that speakers, and (as a check of those Considering that recommendation, render it impossible tor more than translations) translated back into English by two different bilingual we chose to collect qualitative data four vendors to operate in this par­ speakers, as per recommended from a total of 100 customers repre­ ticular food court. The airport's senting our three market segments. managers were exploring covering - ].J. Louviere. Analyzing Individual Decision a broad menu spectrum with the Making: Metric Conjoint Analysis (Newbury Park. 17 See: L. Copeland and L. Griggs, Going Inter­ four restaurants—those being C A: Sage Publications, 1988), 21 national (New York: Random House, 1085); and burger, pizza and Italian, hot dogs For example, see: GJ. Halm and S.S. Shapiro. E.T. Hall and M.R. Hall, Understanding Cultural "A Catalog and Computer Program tor the Differences (Yarmouth, ME: Intcrcultural Press. and Mexican, and a deli concept Design and Analysis of Orthogonal Symmetric 1990). (with an Asian adjunct). For our and Asymmmetric Fractional Factorial Experi­ '" Hofstede,;'/). at. survey, these possible restaurants are ments" (Schenectady. NY:Technical Report 1" For example, see: A. Griffin and J.R. Hauser, Number 66-C 165, 19<>6). See also: R. McLean "The Voice of the Customer," Marketing Science, referred to as "burger," "pizza," and V. Anderson, Applied Factorial and Fractional Vol. 12. No. 1 (1993), pp. 1-27. "dogs," and "deli." Exhibit 1 lists Designs (New York: Marcel Deckker, 1984).

December 1999 • 81 Exhibit 2 Choice Set 11 a sample

Restaurant 1 Restaurant 2 Restaurant 3 Restaurant 4 None

Brand name McDonald's Local restaurant La Prefreda or Goya Subway or Boston products Market Variety Burger, fries, ice cream Pizza, lasagna, pasta, Hot dogs, fries, Sandwich, soup, salads, soups nachos, burritos, ice cream, tacos, tamales, Udan noodle enchiladas soup, salads

Wait time (before ordering) 5 to 6 minutes 0 to 2 minutes 3 to 4 minutes 0 to 2 minutes

Service time 0 to 2 minutes 3 to 4 minutes 5 to 6 minutes 3 to 4 minutes

Menu language English English, Spanish, English and Spanish English and and Japanese Spanish

Picture display Yes No No No

Price (meal & drinks) $4 $4 $10 $7

/ would purchase food from...

methods.22 The final survey instru­ Exhibit 3 ment was administered in the three Respondents' profile languages (English, Japanese, and Spanish) to approximately 500 trav­ elers from June through October s £ / 1998. Of those questionnaires, 452 / were usable (90-percent response). i // o° ^ / // // All #4 #-f Customer-choice Models A/=452 48.17 $4.51 2.55 33.88 48% 77% 34% 20% As Exhibit 3 indicates, the respon­ English dents' demographic profiles show a n = 253 51.56 $5.25 2.41 35.05 46% 100% 15% 3% number of differences among the groups. The English group spent the Spanish most money per person, while the 17=117 33.42 $3.74 3.00 29.45 50% 66% 100% 0 Japanese spent the least. (Remem­ Japanese ber, these are language groups and n = 82 70.87 $3.02 2.41 36.80 49% 24% 0 100% not necessarily nationalities.) The Spanish group spent below average per person, but bought food for more people than the other two

— See: R.W. Brislin,"Back Translation for Cross-cultural Research," journal of Cross-cultural Psychology, Vol. I (1970).pp. 185-216;and H.C. Triandis. "Methodological Problems of Com­ parative Research," International Journal oj Psychol­ ogy, Voh 1 I. No. 3 (1976).

82 CORNELLHOTELAND RESTAURANT ADMINISTRATION QUARTERLY FOOD AND BEVERAGE MANAGEMENT groups. The Japanese group spent the most time waiting in the food Exhibit 4 court, while the Spanish group spent Estimated multinomial logit-choice models the least. Next, the MNL models devel­ Variables English Spanish Japanese oped for each of the three customer Intercepts: groups (English, Japanese, and Span­ Burger 0.7009 * 2.1472* 1.0843 * ish) are shown in Exhibit 4. One can Pizza 0.8630 * 1.5594* 0.2872 * look at the relative intercept values Dogs 0.2180* 1.2289* -0.3996 * to get a general idea of how the Deli 1.2732 * 1.4811 * 1.1170* group felt about each of the four Brand name restaurant concepts. To interpret the Burger -0.0923 -0.1104 -0.1201 attribute information, we first look Pizza 0.1210* -0.1112 0.0683 Dogs at the possible values that an at­ -0.2081 * -0.3951 * 0.0987 Deli -0.1353* 0.1740* -0.0280 tribute can have and multiply a pos­ sible value by its corresponding Variety weight. We then sum these products Burger 0.2699 * 0.1059 0.2296 * Pizza 0.1424* 0.0996 0.1152 over all the attributes. The resulting Dogs 0.0677 0.3382 * -0.4051 * sum is a utility or overall preference Deli 0.0435 -0.0576 0.4295 * that the group of customers or a market has for a given restaurant Wait to order Burger -0.1758* -0.7420 * -0.1496* concept. Higher utilities mean a Pizza -0.2215* -0.1090 -0.0854 higher probability that customers Dogs -0.3152* 0.1902 0.0126 will prefer a particular service Deli 0.0653 0.1038 0.0050 alternative. Service wait Looking at the goodness-of-fit Burger -0.2471 * -0.0800 -0.1182 statistics for the three MNL models, Pizza -0.0306 0.0268 -0.0914 the high r2 and adjusted r2 values Dogs 0.1347* -0.2026 * -0.1590* demonstrate a strong fit between Deli -0.1110* -0.0081 -0.1108 the estimated model and observed Menu language empirical data.23 Burger -0.4804 * -0.8713* -1.0102* Pizza -0.5590 * -0.9151 * -0.8342 * English-speaking Segment Dogs -0.8267 * -1.1485* -0.9787 * Deli -0.4810* -0.7424 * -0.7849 * Interpreting the model for the English-speaking group, we see that Price (meal + drinks) the deli and pizza concepts are most Burger 0.0995 0.3242 * 0.2128* Pizza -0.0362 0.1468 -0.0405 popular. Brand names have relatively Dogs 0.0975 0.3180* 0.2735 * poor strength; La Prefreda and Goya Deli -0.0697 0.2852 * 0.1812* (dogs) has the most negative percep­ tion followed by Subway and Boston Picture display of items Burger -0.1091 * -0.1078 0.1709* Market (deli), and McDonald's Pizza -0.0151 -0.0560 0.0306 (burger). Pizza Hut and Domino's Dogs 0.0314 -0.2240 * 0.2415* (pizza) were the only brands with Deli 0.0680 0.1739* 0.0609 Goodness-of-fit statistics 23 See: Ben-Akiva and Lerman, op. cit. The 2 r values are: 0.67, 0.71, and 0.62. Please note 2 McFadden's r 0.7154 0.7509 0.6855 that although it is possible to identify the general

preference trends for MNL choice models, it is 2 not appropriate to directly compare the (3 coeffi­ Adj. McFadden's r 0.6738 0.7090 0.6271 cients for two models. The recommended statistic' cal test for equality of MNL model parameters * Statistically significant at the 5-percent level. is based on a X2 statistic and was developed by: J. Swait and J.J. Louviere, "The Role of the Scale Note: The intercepts category shows the utility of each restaurant concept with respect Parameter in the Estimation and Comparison of to the alternatives—provided that everything else remains constant. For example, if Multinomial Logit Models," Journal of Marketing everything else remained equal, the English-speaking customers are most likely to Research, Vol. 30 (1993), pp. 305-314. choose deli followed by pizza, burger, and dogs.

December 1999 • 83 Spanish group, increased waits are Exhibit 5 viewed negatively except waiting to Suggested strategy changes order for dogs and deli. Although multilingual menus are undesirable Pizza Dogs Deli for the Japanese speakers, they ap­ preciate pictures of food. We found Local branding Local branding Local branding them to be price sensitive for the Increase to high variety Increase to high variety Increase to high variety pizza concept only. Improve order waits to Improve order waits to Maintain or increase order 0 to 2 minutes 0 to 2 minutes waits to 5 to 6 minutes Preference Differences by Concept Reduce preparation time Maintain existing Reduce preparation time The burger concept was most to 0 to 2 minutes preparation time to 0 to 2 minutes popular with the Spanish group English-only menu English-only menu English-only menu and least popular with the English No pictures Keep picture displays Add picture displays group. All groups shared negative Increase average price Increase average price Increase average price feelings about a burger brand name, long waits (in all concepts), and Share change: +9.72% Share change: +9.53% Share change: +13.70% multilingual menus. They shared positive feelings about increasing Note: Share changes assume that competitors do not change their concepts. menu variety. The three groups' preferences diverged on price, food favorable scores. Variety is important names, Subway and Boston Market, items, and food pictures. While the to these travelers, and a customer's have a positive effect on preference Japanese and Spanish customers preference for the restaurant in­ for the vendor. Second, increasing were not price sensitive, the English creases as the menu variety increases. variety is viewed favorably in all group generally preferred lower Adding variety to the burger con­ cases except at the deli (which in­ prices. On the other hand, the cept (special burgers and sand­ volved adding more Asian food English and Spanish groups did wiches) and to the pizza menu items). Not surprisingly, adding not want pictures of the food, while (pasta, salad, and soups) created the Mexican items to the dogs menu the Japanese did. most favorable preferences. Third, was extremely popular with this Sampling pizza. The groups shorter wait times for burgers and group. Third, longer waiting time differed on their perception of the pizza generate an increase in the hurt all vendors except in two in­ pizza concept. The Japanese and customer's preference for those ven- stances. Respondents viewed in- English groups preferred a brand- dors. The English speakers were not creased order waiting time favorably name operator to the local vendor, impressed by a lengthy wait to order at dogs and deli. Fourth, the addi­ but the Spanish group felt nega­ at the dogs counter but viewed a tion of Spanish- or Japanese- tively about a brand-name product. longer service time favorably. The language menus is viewed nega­ Regarding long waits, the English deli needs a brief wait for service, tively. Pictures of the food only and Spanish groups felt more nega­ and people were comfortable with benefits the deli positively. Finally, tive about the wait to order than that wait to order. Fourth, this seg­ this group was not price sensitive. about waiting for service. The Japa­ ment was happy with an English nese group did not differentiate menu, but their preference for a Japanese-speaking Segment between the two waits. None of the vendor decreases greatly as more From the Japanese customers' per­ groups wanted multilingual menus languages are added to the menu. spective, the burger and deli are the for pizza. For this concept, both the Pictures on the menu returned most popular concepts. This group English and Japanese groups were mixed results: this group doesn't views brand names positively for price sensitive, while the Spanish need pictures of burgers or pizza, pizza (Pizza Hut or Domino's) and group was not. but did like pictures of dogs and for dogs (La Prefreda or Goya Prod­ In the doghouse. The dog con­ deli items. Finally, reducing prices ucts), but not so for the other ven­ cept was the least-popular concept has negligible impact on the group. dors. Increasing variety is favorable for all groups. Incorporating brand- for all vendors except dogs (which name food items was viewed nega­ Spanish-speaking Segment involves an increase in Latino food tively by the English and Spanish Spanish speakers by far preferred the items) and is extremely favorable groups, but positively by the Japa­ burger concept over the other three. for deli (which involves increasing nese group. Increasing variety by However, only the deli-related brand Asian food items). Similar to the adding Mexican food items was

84 CORN ELL HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY FOOD AND JJJJJJj MANAGEMENT greatly preferred by the Spanish burger concept receives most of the group and preferred slightly by Japanese and Spanish market share. the English speakers, but that was For the following analysis, we have viewed negatively by the Japanese left the McDonald's attributes at group. The English group did not their existing levels and changed like long waits to order dogs, while each vendor in isolation of the the Japanese and Spanish were will­ other vendors. That is, we tested the All groups shared negative ing to wait. On the other hand, the model as though one vendor could English group felt more positive change its features without any feelings about long waits and about the service wait than their changes by the others. We then counterparts. Once again, multilin­ set each vendor at its maximum multilingual menus. They gual menus were universally dis­ market-share configuration and liked, while food pictures were pre­ looked at the overall impact. ferred by only the Japanese and Exhibit 5 shows market share shared positive feelings about English groups. All the groups that results from maximizing showed price insensitivity towards changes for each of the three ven­ increasing menu variety. the dogs concept. dors other than McDonald's. For Next, please. The deli concept example, if the pizza outlet changed was most popular with the English without competitors retaliating, it respondents. Neither the Japanese would gain 9.7-percent market nor English preferred a brand-name share. On the other hand, if com­ deli, although the Spanish group petitors all change then the pizza did. The Japanese group showed a place gains only 1.8 percent in strong preference for increasing overall share. The biggest positive menu variety by adding Asian items. changes to pizza come from increas­ The English group preferred some ing the variety to include pastas and variety of this type, but the Spanish salads and reducing the overall wait group disliked that increased variety. time (to order and deliver food). In None of the groups was sensitive that case, pizza gains share from the about waiting to order, but all dis­ English and Spanish groups but liked increased service wait. Again, loses share from the Japanese group. none of the groups wanted multi­ Similarly, the biggest positive lingual menus, but did want food changes to the dogs concept come pictures. Finally, of the three groups, from increasing the variety to in­ only the English group was price clude more Mexican food items sensitive to this concept. and reducing the wait time to order. With those changes, dogs gains Strategy Implications 9.53-percent share (in the absence Given the above models for each of other vendors' changes) and gains group and food-service concept, 2.57 percent, if all change. These we next look at what each vendor changes are reflected positively with could do to improve its market a large shift from the Spanish group, share. We determined the composi­ followed by the English group. As tion of the food-court market by we indicated above, those changes converting customer-traffic esti­ are negative for the Japanese group. mates from O'Hare's schedules into Adding Asian foods to the deli's forecasts for food-court customers. menu, reducing waiting time tor Based on that analysis we posit that service, and adding pictures of food the international terminal's market contribute to the largest market- comprises 60 percent persons from share improvements (6.25 percent English-speaking destinations, 25 overall, even when all vendors percent from Spanish destinations, change). For all groups, increasing and 15 percent from Japanese loca­ the average price is not viewed det­ tions. According to our models, the rimentally by the customers. By

December 1999 • 85 adding specialty Asian foods and Exhibit 6 picture displays, the deli gains a large Impact of different operations strategies (hypothetical) portion of the Japanese group, and those changes are viewed favorably Companies and (primary Burger Pizza Dogs Deli by the other two groups also. operations strategy) (fast service) (high quality) (low cost) (high variety) A scenario. Exhibit 6 presents a Intercepts: Burger 1 hypothetical scenario of the impact Pizza 1 of different operations' strategies in Dogs 1 a given market environment. This Deli 1 example is presented for the sake of Brand name Burger -1 illustration only and does not corre­ Pizza 1 spond to any actual company oper­ Dogs -1 ating at the international terminal. Deli -1 The purpose of this example is to Variety Burger -1 demonstrate how MNL choice Pizza 0 Dogs -1 models can be used to position a Deli 1 service firm's operations according to the preferences of one or more Wait to order Burger -1 Pizza 0 market segments. The table shows Dogs 1 the expected market share for four Deli 0 competitors given the following Service wait Burger -1 strategic changes: burger gives fast Pizza 0 service, pizza adds high perceived Dogs 0 quality, dogs goes for the low-cost Deli 0 market, and the deli promotes high Menu language Burger -1 menu variety. The calculations show Pizza 1 Dogs -1 that each firm can expect different Deli 1 market shares in the three markets. Burger's fast-service strategy is fa­ Price (meal and drinks) Burger 0 vored by each of the three customer Pizza 1 groups, but that outlet can expect Dogs -1 only a 33-percent share of the Deli 0 English-speaking market, while it Picture display of items receives approximately 72-percent Burger 1 shares of Spanish and Japanese trav­ Pizza 1 Dogs -1 elers. On the other hand, both pizza Deli 1 (with its high-quality approach) and the deli (offering high variety) are Estimated English 32.74 15.32 22.13 29.81 Market Spanish 71.91 3.20 21.51 3.38 expected to gain a much higher Share (%) Japanese 71.85 3.91 9.34 14.90 share of the English-speaking mar­ ket compared to the other two The table shows the expected market share for four competitors given the following groups. Dogs expect to gain ap­ strategic changes: burger gives fast service, pizza adds high perceived quality, dogs proximately equal share in English goes for the low-cost market, and the deli promotes high menu variety. and Spanish markets and a much lower share in the Japanese markets. Note: The rows at the bottom of the table give the expected share of each restaurant concept for each market for the various attribute levels that were tested. The design codes relating to -1, 0, and 1 are given in Exhibit 1. For example "-1" for brand means "local Differences in Perceptions brand," while "+1" means "national brand." Similarly, "-1" for picture display means no pictures are displayed, and "+1" means they are. For further explanation of this methodol­ Our study shows significant differ­ ogy. see: Rohit Verma and Gary Thompson, "Basing Service Management on Customer ences in the perceptions of the three Determinants: The Importance of Hot Pizza," Cornell Hotel and Restaurant Administration cultural customer segments that we Quarterly, Vol. 37, No. 3 (June 1996), pp. 18-23. studied. Given the current mix of the three groups, we proposed changes to each service design and demonstrated the likely impact of

HOTEL AND RESTAURANT ADMINISTRATION QUARTERLY FOOD AND BEVERAGE MANAGEMENT

operating-strategy changes. In all of the day only, both of which senger departures at the terminal cases, the groups want more variety, are off-peak times for most other remained essentially constant). Our and it is no surprise that they par­ travelers. study looked only at the possibilities ticularly want more foods with From an operations-management of shifting shares among the existing which they are familiar. Context- perspective, the challenge is to four vendors, and we did not esti­ specific attributes such as waiting for modify process attributes, such as mate the effects of drawing new a whole pizza (an item that requires time spent waiting to order and to customers from the population that more preparation time than the receive the order (usually prepara­ were previously non-purchasers. Yet other categories) or desiring pic­ tion time). In most cases, the differ­ that is exactly what must have oc­ tures of unfamiliar foods depended ent groups had similar preferences curred, for sales greatly exceeded on the service concept and the cus­ for waiting times, but there were our projections. In this case, it ap­ tomer group's familiarity with that some exceptions, particularly for pears that research aimed at maxi­ concept. Using the DCA method, the dog and deli concepts. For those mizing market share is a worthy wTe were able to model the prefer­ concepts the groups had different objective for each food-court ven­ ence trade-off among groups and expectations regarding how long dor given its substantial effect of determine improvements that would they should have to wait to order attracting new customers. improve the overall market share for and wait for preparation. We can all the vendors in the food court. only speculate that some customers Looking Ahead From the descriptive results (Ex­ expect to wait longer when menu This study looked at the effects on hibit 3), it appears that the Japanese variety increases or the concept is market share of catering to different group spent the most time in the not expressly labeled quick service. cultural groups in a specific service food court but spent the least per For this relatively straightforward setting. Future research is needed to capita. The Spanish group spent the service setting, none of the groups determine the cost and benefits of least time in the food court and wanted multilingual menus, but in customizing certain portions of a spent below-average amounts on some cases the groups wanted food service versus making a standard each person, but they had the largest pictures. The Japanese group wanted service offering. It has long been party size. Thus the vendors had an pictures for all menus, the Spanish established that certain market seg­ opportunity to increase sales to each group wanted pictures only for the ments generate more revenue than of these groups. In this competitive deli (particularly as variety increased others (regardless of how the seg­ environment it made sense for some with the addition of Asian foods), ments are determined). Moreover, vendors to customize their food and the English group wanted pic­ seeking to please all market seg­ offerings for different customer tures for both the dogs (again as ments at once will probably add too 24 groups. For example, the dogs ven­ variety increased with the addition much complexity and cost. dor could add the Mexican food of Mexican foods) and deli. We While this study examined food items we had proposed to appeal conclude, then, that each group service, other service industries to Spanish-speaking customers, wants pictures of foods with which could benefit from market analysis and the deli vendor could add the they may be unfamiliar as variety using the DCA approach. To sum­ Asian food items to appeal to the expands. marize, we have presented an ap­ Japanese-speaking customers. Those proach for positioning a service and variety additions generally had at Phased Implementation formulating operations strategy in least some cross-cultural appeal. The O'Hare Authority planned to a multicultural environment. The While that variety increase may implement the results of this study case analysis at the international appear to add to the complexity in several phases. The first phase was terminal's food-service operations of service delivery, those two ven­ the modification of dogs, by adding demonstrates the value of DCA dors could take advantage of the Mexican food items to the menu at and market-utility models in service departure-time windows for Japa­ night from 8:00 to midnight while environments. We hope that other nese and Spanish travelers. Owing at the same time reducing waiting research teams will undertake simi­ to airline schedules, the majority of times by improving process and lar projects to analyze interdiscipli­ the Japanese travelers are in the food labor-scheduling efficiencies. At the nary issues related to service opera­ court between 8 AM and 12 noon, time of this report, the modifica­ tions strategy formulation. CQ while Spanish groups tend to pa­ tions had been in effect for three 24 tronize the food court from 8 PM to months. During this period, the For a discussion of the cost and revenue vendor increased sales by 50 percent outcomes of adding complexity to a service 12 midnight. Thus, vendors could operation, see Enz, Potter, and Siguaw, on modify their menus for those hours from the previous year (while pas­ pp. 54—62 of this Cornell Quarterly.

December 1999 • 87