MICHAEL COLVIN, ROGER HEELER, & JIM THORPE

Promotion strategies for a major car company are developedfor different international market segments by comparingconsumer preferences measured by tradeoff analysis withtheir perceptions of new and existing products.

DEVELOPING INTERNATIONALADVERTISING STRATEGY

THE marketing of automobiles across national bound-aries is a particularly interesting and complex prob-

lem in the identification and exploitation of the needs ofvarious market segments. Many of the variables used todelineate segments (for example see Wind 1978) differconsiderably across national boundaries but are rela-tively homogeneous within. Because elements of themarketing mix such as price and product positioning, asdefmed by the advertising approach, are amenable tocontrol within countries, it is possible both to specifycountries as segments and to develop an appropriatemarketing strategy.

In practical terms the task of developing a marketingstrategy per country is daunting, even for companieswith large resources. Numerous research programs arerequired in different countries at different times andthese must be repeated each time a new strategy optionis to be tested. Consequently, questions of sample com-patibility are difficult to solve and many companieshave adopted a "prototype standardization" approach(Peebles, Ryan, and Vemon 1977) in which minor mod-ifications are made to some basic strategy. The current

Michael Colvin is Manager, Customer and Merchandis-ing Research, Ford of Europe, Roger Heeler is AssociateProfessor of Marketing, York University, Canada, formerlywith Ford of Britain, and Jim Thorpe is a doctoral stu-dent, Manchester Business School, England. Social Sci-ences and Humanities Research Council of Canadasupport is gratefully acknowledged, and data collectionby Q.E.D. International.

trend is toward "pattem standardization" whereby astrategy is designed from the outset to be susceptible toextensive modification to suit local conditions, whilemaintaining sufficient common elements to minimizethe drain on resources and management time.

We describe how Ford of Europe implemented a"pattem standardization" approach to new car advertis-ing in Europe.

In the case of Ford the objective of an advertisingcampaign within the marketing mix associated with aproduct is twofold.

1. To achieve product awareness.2. To communicate to each market segment the

product attributes the segment perceives as ful-filling its needs.

The first objective is, of course, supranational, and inthis article we consider how the second objective isachieved.

MethodBoth buyers' perceived needs and buyers' perceptionsof product attributes differ among countries so both arestudied. Europe is divided into a number of submarkets—in our example, Germany, United Kingdom, France,and Sweden. Within each country a panel is recruitedfrom persons whose intended next car purchase is withina range compatible with the new Ford product. Thesample size is typically 200-300 per country and is strati-fied to match the relevant consumer profile as estab-lished by past research.

Journal of MarketingVol. 44 (FaU 1980), 73-79. Developing International Advertising Strategy / 73

PerceptionsInformation on respondents' perceptions of a new car(or van or truck) is obtained in response to stimuli attwo levels, which approximate the two extremes ofpossible communication success. At the "low awareness"level, respondents are simply shown a "round the clock"series of photographs on the exterior of the new car andare told it is a new Ford. This level is consideredsimilar in impact to low key advertising plus streetappearances of the car that achieve little more thanbasic product awareness. At the "high awareness" level,the same respondents are given a full photographic andverbal briefing, in the language of their country, on theexterior and interior appearance, features, performance,etc., of the car. This level approximates the maximumproduct knowledge that could reasonably be communi-cated by heavy advertising. At both high and lowawareness levels respondents rate the car on 27 attributesshown to be significant by past research, and indicatepurchase interest. Attribute rating and purchase interestdata are also obtained for the existing cars seen by therespondents as possible altematives to the new car.

PreferencesRespondents' perceived needs are analyzed by conjointanalysis. This method of obtaining consumer utilities isgaining increasing acceptance in consumer behavior re-search (for a comprehensive review, see Green andSrinivasan 1978).

In this application the tradeoff method of data col-lection is employed (Johnson 1974) which requires re-spondents to evaluate matrices based on pairs of productattributes. For example, in Figure 1 for "Workmanship/Finish" versus "Servicing Facilities" each respondentis asked to state a ranked preference for all combina-tions of attribute levels.

In completing this task, the respondent must makedecisions about his or her priorities, for example, whento forego "good servicing" in order to retain "high

FIGURE 1Tradeoff Matrix

Workmanship/

Finish

HIGH

AVERAGE

LOW

Ser

Good

1stchoice

3rS-

vicinq racilities

ftdequate

^ y• flth^^^^^^

Poor

5th

i6th

9th

levels of workmanship and finish" or the reverse. InFigure 1 the respondent prefers to keep a high level ofworkmanship/finish at the cost of a lowered standard ofservicing facilities.

Bach respondent completes many matrices of thistype. To reduce the workload to manageable propor-tions a subset of the complete set of possible attributepairs is used. The reduced number is adequate for meas-uring main effects, and use of a reduced number hasproved robust in tests noted by Johnson (1974). In theexample given hereafter, 27 attributes were used. Theseyield a possible 351 matrices of which 63 were actuallyused. The 63 were chosen on a grid basis whereby eachattribute was used in at least three matrices and was nomore than twice removed from any other attribute. Thusif A were compared with B, B with C, and C with D, Awould be twice removed from D.

The matrices of preference data are scaled by anappropriate program, in this case a commercial deriva-tive of MONANOVA (Kruskal 1965), to yield eachrespondent's "utility" values for each attribute. Utilityis simply a measure of joint relative importance. Thismeasure varies according to the level of the attribute.For example, excellent fuel consumption might have ahigh utility, average fuel consumption a low utility.

Model ConstructionThe perception and preference data are now in a formatsuitable for use in strategy development and a decisionmodel is used as a link. Different models have beenemployed from time to time. For our illustration amodel called SCIMITAR (Westwood, Lunn, and Beazley1974) is used. The first step is to apply the utilities toeach respondent's evoked set of car perceptions andthus predict current purchase intentions. This modelprediction is compared with the intentions data from thepanel as an intemal validity check. Agreement betweenmodel predictions and intentions is usually in the rangeof 80-90%. This in effect means that we have a com-puter model of the sample population's preferences bycountry which, when exposed to combinations of therelevant class of inputs, will duplicate response of theactual sample within the error limits of the model.

The next step is to remove the existing Ford product(if any) from the available model range and add the newFord model with attributes communicated at the "lowawareness" level. The share of purchase interest predic-tions yielded thus sets a base level of how the car mightperform should the advertising have only limited com-munication success.

Finally, the potential of different communicationstrategies is tested. Each strategy emphasizes only oneor two main attributes of the car, because the compa-ny's experience has shown the difficulty of effectively

74 / Journal of Marketing, Fall 1980

FIGURE 2Advertising Strategy Development

COUHTBY SELECTIONSelect N countries necessary to

econon-lcally represent total markets

For country II For country f2,....N

PURCHASE IHTERESTSelect respondents

Obtain purchase intentions

PERCEPTIOH MEASUREMENTObtain car image data on1. Current models2. Hew car 'low awareness '3. New car "high awareness '

PPEFEPENCE MEASUREMENTObtain car attribute

utilities by trade-offanalysis

COMPARISON ICompare perception and

preference results acrosscountries

DECISION MODELLINGCombine image and utility data to1. Validate by reproducing purchase intentions2. Evaluate alternative strategies3. Best strategy chosen for execution

COMPARISON IICompare strategy results across countries,Choose common strategies where beststrategies match. Choose differingstrategies where best strategies don't match

TACTICS RESEARCHCommunication effectiveness

testing of advertising executionsbased on the best strategy

Best execution chosen for use

RESULTSBest execution for each countryor group of countries used for

product introduction

communicating more than one or two points in a launchcampaign.

The strategy is operationalized for model purposesby assuming that the attributes to be promoted areraised to the level of perception corresponding to "highawareness" in the initial research. Because of intercor-relation. several attributes may be associated with agiven main point of communication. For example, "per-formance," "road holding," and "maneuverability"might all be associated with an advertising strategy usinga "sports car" focal point. So the model implementationof one main point could involve adjustment of severalsubsidiary attributes. The model yields a market shareprediction for each strategy on the assumption that thestrategy is successfully communicated. The source ofgains and losses constituting that share and the demo-graphic profile of the predicted buyers are also shown.These results are compared with each other and withthose achieved at the "low awareness" level. Superiorstrategies for one country are also evaluated in othercountries to see what commonality of campaigns ispossible.

The strategy research is complemented by separatesets of tactics research in which advertisement execu-tions in accordance with chosen strategies are evaluated.The executions are tested for their effect on fl) inten-tion to purchase and (2) those perceptions of the carconsidered significant in light of the strategy objectives.

Figure 2 is a diagrammatic representation of theprocess, which we illustrate with data from the 1977launch campaign of the Granada.

Illustration of Strategy DevelopmentThe Ford Granada is positioned at the large car end ofthe European market. It competes with the lower rangesof the speciality manufacturers such as B.M.W. and theupper ranges of the volume manufacturers such asRenault. Research was conducted in Britain, Germany,France, and Sweden.

Perceptions

Table 1 shows the German ratings for current modelsand the new model with awareness at both low and highlevels. It contrasts these with the ratings of the car thatrespondents stated as their intended next purchase. Foreach attribute the Granada rating increases from "lowawareness" to "high awareness." The new car is seenas equal or better than the old car on all attributesexcept luggage capacity. In relation to competition thenew car's rating varies across the attributes although itremains lower than those of the competitor. It is superiorto "next car" on some features such as styling andinferior on others.

Table 2 and the two right columns of Table 1 showthe variation in some of these ratings among countries.

Developing International Advertising Strategy / 75

TABLE 1Car Perception: Ratings of Selected Attributes*

Germany

New NewModel Model-Low -High

Current Aware- Aware- "NextModel ness ness Car"^Attribute

Luggage capacityTechnically

advancedJourney comfortStylingPrestigeSafetyRoadholdingReliability

Purchase interest(%}'^

Sample size

7.37.0

7.25.76.05.06.36.7

7

6.87.4

7.36.87.15.06.66.8

n ^

6.91.1

7.57.27.35.77.27.1

5.1

312

6.37.6

7.56.97.26.37.47.6

^Ratings out of 10.'"'Next car" rating excludes respondents choosing the

current Granada.•^Purchase interest is the percentage of the sample who

indicate that the Ford car is likely to be their next carpurchase. It includes equal rankings, i.e. Granada equallylikely a purchase as previously preferred car. These "equalrankings" form about half of the overall "new model"preferences.

The differences among the car perceptions in the fourcountries indicate the importance of tuning marketingstrategy to each market. In relation to competition,Granada styling scores well in Germany and Sweden,but poorly in Britain and France. On safety the Granada

scores better than competition in Britain but worse else-where. On reliability it is seen as worse than competi-tion in all countries, but particularly so in Sweden.

PreferencesThe preference stage of the research yields utility valuesin each country for each level of each attribute. Theresults for three of the attributes are shown in Figure 3 toillustrate the range of results obtained. The absolutelevels of the scores are arbitrary.

The results for the attribute "durability" are nearlyidentical among countries. On this variable little wouldbe lost in relation to preference by treating all fourcountries similarly. Robinson and Wind (1977), in astudy of transatlantic air fares, found little difference inattribute ratings between countries and proceeded tocreate crossnational segments. The utilities are a posi-tive monotonic function of degree of durability andindicate a substantial desire for durability. In contrastthe attribute "technically advanced." though alsomonotonic, shows considerable differences among coun-tries. For the step "not advanced" to "technically ad-vanced" the increase in utility in Germany is doublethat in Sweden. The increase for Britain and France isintermediate. Treatment of this variable by the marketdiffers considerably among countries. The third varia-ble, "exterior size," shows both differences amongcountries and nonmonotonic relationships. In all coun-tries respondents prefer an "average size" car to a"larger" car. In Germany "smaller" is preferred to"average," yielding an inverse monotonic result. Forthe other countries "average" is preferred to "smaller"as well as "larger," yielding an inverted V effect.

TABLE 2Attribute Ratings

Attribute

Luggage capacityTechnically

advancedJourney comfortStylingPrestigeSafetyRoadholdingReliabilityPurchase

Interest (%}

Sample size

by CountryU.K.

New Model-High

Awareness

7.4

7.27.56.66.56.87.06.9

33n = 286

"NextCar"

6.2

6.77.37.06.96.77.47.5

France

New Model-High

Awareness

6.8

6.67.56.97.06.36.76.8

33

n = 193

"NextCar"

6.2

6.67.27.37.26.5117.3

Sweden

New Model-High

Awareness

6.8

7.47.47.16.65.87.06.5

37

n = 223

"NextCar"

6.4

6.67.46.56.46.47.51.1


FIGURE 3Utility Values for Four Countries for ThreeAttributes

DURABILITY TECHNICALADVANCEMENT EXTERNAL SIZE

Utility

300

200

100

0

9X

-

- •1

•X

Less As Longer Not Advanced Small Average LargeLong Long Advanced

. Sweden * United Kingdom * France and United Kingdom

O France X Germany • All Countries

Model Validation

Thus, both perception and preference data yield resultsthat vary by country. When the two are related to oneanother, markedly different strategies might be neededto suit the different countries. The decision model isused to combine the two elements of preference andperception.

The model was first validated by using preferenceand brand rating data to predict respondents" currentpurchase intentions. These predictions correctly matchedrespondents' stated first choice 82% of the time.

Strategy Testing

In the strategy testing phase single-theme strategies weretested in each country. The results are shown in Table3. In Germany, Britain, and Sweden the most success-fill single strategy is reliability. In France the mostsuccessful single strategies are comfort and safety.

Several dual-theme strategies also were developed;three are shown in Table 3. For Germany and Britainthe best dual-theme strategy is reliability-safety. ForFrance and Sweden it is safety-comfort. It is interestingthat in Britain, Germany, and France the most effectivedual-theme strategies are combinations of the two mosteffective single-theme strategies, but the same is nottrue for Sweden.

Sensitivity Considerations

The percentage shares in Table 3 should not be taken asa direct indicator of success because they are based onthe assumption that the simulated strategies have beenfully communicated and believed. If these tasks are notachieved then advertising effect may fall disastrouslyshort of expectations. In Britain. Ford as a company hasa particularly favorable image, probably because ofthe relative weakness of the domestic competition. Arecent campaign for another model (Cortina) success-fully emphasized reliability so success with this appealfor the Granada appears reasonable. In Germany thedecision model predicts a reasonable level of purchaseinterest even if reliability is not fully communicated, sothe "downside" risk with this element in the strategy isdeemed small. In France and Sweden the model predic-tions are for very low sales if a reliability appeal is triedand fails. Similar sensitivity testing on other strategythemes leads to the conclusion that the reliability-safetystrategy would be best for Britain and Germany, and thesafety-comfort strategy would be best for France andSweden.

Tactics ResearchIn the tactics phase of research, several advertisingexecutions based on these strategies were exposed tonew respondent samples. The results are given in Table4 for the execution strategies actually used. Execution

TABLE 3Predicted Percentage Share:Campaign

Single StrategyReliabilitySafetyLow Running CostsComfortTechnically Advanced

Dual StrategyReliability-safetyReliability-comfortSafety-comfort

> of Purchase Interest in

Germany

34^33272223

43NA38

^Predicted share of preferences including equal rankings.NA = not available.

Strategy andBritain

3627262424

464230

CountryFrance

2831273123

363239

Sweden

3129222518

282839


TABLE 4Percentage of Respondents Giving Ford asProbable 1

Country

BritainGermanyFranceSweden

Mext Car PurchaseInitial

SampleAfter

Exposureto High

AwarenessAdvertising

33413337

ModelSimulation

Effect ofStrategy

50433939

ExecutionTest of

Strategy53454430

test results match or exceed the preference predictionsbased on the strategies in Germany, Britain, and France.In Sweden execution results are less successful, the bestexecution variant achieving about three-quarters of thepredicted level of purchase interest.

Relative CostsThe need for an intemational marketing approach thatrecognizes the differences in product perceptions andpreferences among countries is illustrated in the preced-ing results. But what is the relationship between thecost and effort involved in the research and the value ofthe results?

The Granada in Europe is worth more than a billiondollars of business per year so a substantial researcheffort per se is justified. The question is actually one ofchoosing among alternative methods. A simpler approachwould be to conduct a series of tests of ad hoc advertis-ing executions. With the Granada seven small-sample(IOO) tests could be run for about the same cost as one300 sample (per country) of the kind we describe. Butfor minimal incremental computer time costs the buyingintention model can test the effectiveness of numerousstrategies. Also the model yields diagnostic behavioralinformation. Through an understanding of consumerperceptions and interests, more effective strategy devel-opment is possible (Boyd, Ray, and Strong 1972). Speedand sample stability are also advantages, the only limitingtime factor being computer job turnaround speed. Withsmall-sample tests, sampling error may lead to differentresults between executions. With our modeling approachthe same larger sample is used throughout so there is nocomparability problem between tests.

System ValidityThe validity of the system is obviously an importantquestion. No questionnaire system can be totally relied

upon to predict eventual consumer marketplace reac-tions; there are just too many intervening variables.

Nevertheless, some consistent set of decision rulesis preferable to strategy by guesswork. The model yieldstwo validity checks of its own. One is the use of theattribute rating and utility data to predict car purchaseintentions. The 80% agreement obtained is at least indic-ative of basic system congruence. To infer more isunwise; though a low percentage of agreement wouldcertainly indicate failure, a high percentage may indi-cate validity or it may indicate halo effects betweenmeasures obtained in the same questioning situation(Beckwith and Lehmann 1975).

The second validity check is intrinsically strongerbecause it is external in that it uses the system predic-tions from one sample to project behavior by another.The strategies designed from the strategy phase are testedin the tactics phase. As noted heretofore, execution re-sults, in terms of advertising communication goals, arein line with strategy phase predictions for three of fourcountries for Granada. The results for other car andtruck models have yielded even more agreement be-tween the strategy and tactics phases.

A third desirable level of testing is field experimen-tation (Ray 1978). A full test is precluded by the costsand impracticability of running different car advertisingprograms within different sectors of a European coun-try. But a practical field test of the whole program isgiven by actual Granada sales. Executions based on thebest mode I-predicted strategies were used in the actuallaunch campaign. In 1978 sales reached 211,000, thelimit of production capacity. This outcome does notprove that the best possible strategies were used, butcertainly shows that they were far from failures.

Further indirect validation is provided by the conjointmeasurement studies noted by Green and Srinivasan(1978).

SummaryThe system described provides an effective means ofsegmenting promotional strategies for intemationalmarkets by allowing for cross-country differences inproduct perceptions and product attribute preferences.The results appear to be of reasonable validity, at leastfor this class of product, and the costs in proportion tothe benefits derived.

REFERENCESBeckwith, N, E. and D. R. Lehmann (1975), "The Importance

of Halo Effects in Multi-Attribute Attitude Models," Journal


of Marketing Research. 12 (August), 265-75.Boyd, H. W., M. L. Ray, and E. Strong (1972). "An Attitudinal

Framework for Advertising Strategy," Journal of Marketing,36 (April). 27-33.

Green, P. E., and V. Srinivasan (1978), "Conjoint Analysis inConsumer Researcti: Issues and Outlook," Journal of ConsumerResearch. 5 (September). 103-23.

Johnson, R. M. (1974). "Trade-off Analysis of Consumer \d\-ne&,'^ Journal of Marketing Research, 11 (May), 121-27.

Kruskal, J.B. (1965), "Analysis of Factorial Experiments byEstimating Monotone Transformations of the Data." Journalof the Royal Statistical Society. Series B, 27, 251-63.

Peebles, D. M., J. K. Ryan, and I. R. Vemon (1977), "A NewPerspective on Advertising Standardisation," European Jour-nal of Marketing, 11, 569-76.

Ray, M. L. (1978), "The Present and Potential Linkages BetweenMicrotheoretical Notions of Behavioral Science and the Prob-lems of Advertising—A Proposal for a Research System." inThe Behavioral Sciences in Marketing, H. L. Davies and A. J.Silk, eds.. New York: John Wiley & Sons, inc.

Robinson, P. J. and Yoram Wind (1977), "Multinational Trade-off Segmentation," in Moving A Head with Attitude Research,Y. Wind and M. G. Greenberg. eds., Chicago: AmericanMarketing Association.

Westwood, R.. A. Lunn, and D. Beazley (1974). "TheTrade-offModel and its Extensions," Journal of the Market ResearchSociety, 16 (July), 227-4i.

Wind, Y. (1978), "Issues and Advances in Segmentation Re-search," Journal of Marketing Research, 15 (August), 317-37.

Marketing: Helpful to Health?Marketing and Preventive Health Care: Interdisciplinary and Interorganizational Per-spectives—edited by Philip D. Cooper, Memphis State University, William J. Kehoe,University of Virginia and Patrick E. Murphy, Marquette University

In March 1977 a multi-disciplinary workshop explor-ing marketing and preventive health care was co-spon-sored by Blue Cross/Blue Shield of Virginia and theAmerican Marketing Association.

Interaction between marketing and preventive healthcare is just beginning. At this workshop participantsattempted to seek out mutuality of interests and thepotential contributions that each discipline can make.All agreed that the role of marketing in general healthcare delivery, and preventive health care in particular,is destined to increase.

Some of the topics covered were:

• Communications and Marketing in Health and

Family Planning Programs

• A Consumer Perspective on Preventive HealthCare Service Usage

• Advertising E*reventive Health Care: Informational,Persuasive, Ethical?

• Is Marketing Too Stigmatized to be EffectivelyAccepted in Preventive Health Care?

• From the Viewpoint of the Government

In developing these proceedings AMA contributes toa new and fast-growing frontier for marketing as itjoins with other disciplines to face the problems ofour society. 132 pages

$5/member $ 7/nonmember

Write to: Order DepartmentAmerican Marketing Association222 S. Riverside PlazaChicago, IL 60606

^ ^ or phone: (312) 648-0536

All orders for $10.00 or less must be prepaid.


Copyright of Journal of Marketing is the property of American Marketing Association and its content may not

be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.

advertisement

Documents