The Development of Brand Loyalty Author(s): W. T. Tucker Source: Journal of Marketing Research, Vol. 1, No. 3 (Aug., 1964), pp. 32-35 Published by: American Marketing Association Stable URL: http://www.jstor.org/stable/3150053 Accessed: 27/03/2009 10:19 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ama. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing Research. http://www.jstor.org 32 The of Brand Development Loyalty W. T. TUCKER* > The experiment reported here examines the growth of brand loyalty in a setting where there is no prior consumer knowledge about any of the available brands. Search behavior precedes the development of brand loyalty, which grows to measurable strength despite the virtual identity of the available brands, suggesting that some consumersare prone to brand-loyal behavior. Most studies of brand loyalty have involved the measurementand descriptionof loyalties to existing brands of merchandise.From these studies it can be safely concludedthat there are rather wide variations in loyalty among individualsand that brand loyalty is at least in part a function of the frequencyand regularitywith which a brandhas been selectedin the past [5] and in part a function of the type of product involved [6]. Rathersophisticatedanalyseshave suggested that some sort of Markov process best describes the growth of brand loyalty [3, 4]. One of the primary characteristicsof the relevantresearchis its relianceon maximumrealism and the minimal distortion of the context of consumerbehavior[2]. Even effortssuch as Pessemier'sstudy measuringthe strengthof brandloyalty are ingeniousin their effortsto maintainthe semblanceof realismunderdifficultcircumstances[7]. Valuableas such methodologicaltechniquesare, they cannot be broughtto bear on some critical aspects of brandloyalty.In orderto discussthese meaningfully,it is necessary to define brand loyalty, despite the apparent simplicity of the terms. For purposes of this paper, brand loyalty is conceived to be simply biased choice behaviorwith respect to brandedmerchandise. If there are two cola drinksofferedto a person a number of times, his degree of brandloyalty can be stated in termsof the relativefrequencywith whichhe chooses one brand rather than the other. If he selects Pepsi Cola rather than Coca Cola (and both are equally available) enough of the time to persuadethe statistically sophisticatedobserverthat the differencein frequency is not due to chance, he may be said to be brandloyal. No considerationshouldbe given to what the subjectthinks or what goes on in his central nervous system;his behavioris the full statementof what brandloyaltyis. It is always dangerousto fragmentmolar behavior into theoretical sub-systems,but the obviously complex characterof brand loyalty demandssome further analysisif it is to be fully useful eithertheoreticallyor to practicalmarketers.The loyaltymay be to some sub* W. T. Tucker is professor of marketing at the University of Texas. set of characteristics:the shapeof the bottle, the sweetness of the drink,the colorson the cap, the brandname, or whatever.Now imagine that the loyalty to Pepsi Cola described above existed during that time when Pepsi was marketedin 12 oz. export beer bottles and the only bottled Coca Cola contained 6 oz. Imagine furtherthat a single change was made: a new 12 oz. bottle of Coca Cola (there was such an experimental size at one time) was availableto him as an alternative to the 12 oz. Pepsi. This might change his choice so that he would subsequentlychoose Coca Cola more frequentlythan Pepsi. It is tempting to suggest that the individual was neverreallybrandloyal to Pepsi at all, but merelypreferred a larger drink. But chaos lies in this direction. Suppose for instance that the Pepsi Cola formula or drink is next placed in the Coca Cola bottle capped with the red, white and blue Pepsi Cola cap and the Coca Cola is placed in the old exportbeer bottle with a Coca Cola cap-and that the individualshiftsback to the Pepsi. What becomes of brand loyalty? In fact brandloyaltyis alwaysa biasedresponseto some combinationof characteristics,not all of which are critical stimuli.This way of looking at brand loyalty may not be congenial,but somethingmuch like it (the gestalt approach suggests variations) seems required either for researchon the natureof brandloyalty or practical questionsregardingchanges in product, packaging,or advertising. One of the hypothesesthat the experimentreported here undertakesto support is the notion that brand loyaltywill grow in an almostcompletelyinfertilefield, that biased choices will develop even when products are virtually identical and brand names are close to meaningless.Clearly, this hypothesisis an outgrowth of the precedingview of brand loyalty. It also stems from a confirmedbelief that the stochastic learning model is not wholly satisfactoryas a descriptionof brandchoice. The stochasticlearningmodel is based on an historical sequence of psychological experimentsin which there were alwaysright (rewardedor unpunished)responses and wrong (unrewarded or punished) responses. (Those few experimentsnot of this nature THEDEVELOPMENT OF BRANDLOYALTY seem to have receivedlittle independentattentionuntil recent years.) There seem to be two underlyingproblems in the learningmodel: (1) it assumesthat a reinforced (usually a rewarded) response will have an increased probability of recurring; (2) it seems to imply that reinforcement,or differentialsin level of reinforcement,form the sole basis for choice. It is possible that a rewardedchoice (say the selection of a brand of shaving cream by the first-timeuser at age 15) may in fact decrease the likelihood of a repeat purchase if the individualis interestedin finding out more about differentkinds of creams or soaps. This could be referredto as search behavior.And, further, it seems possible that this boy could become brand loyal even if he could perceiveno advantagewhatsoever in any particular preparation, since such behavior would at least decreasethe effort of decision making. More positive influences may be in the connotative meaningsand associationsthat naturallygrow around objects one uses, the activities he engages in, and people with whom he interacts.It is too easy an assumption to declare that brand loyalties established in this fashionwould be fragileor transient.Psychological theory,if not learningtheory,suggeststhat one may learnto like what he chooses as readilyas he may learn to choose what he likes. Such possibilitiesseem worth 33 In order to determine the strength of any brand loyalties formed during the experiment,once a panel memberchose the same breadthreetimes in succession, a premiumwas placed on another brand. The brand selected for the premiumwas that brand most seldom selectedpreviously.Wheretwo or more brandssatisfied this requirementa random selection among them of the brand for the premiumwas made. The premium used was a new penny fixed to the brandlabel. If the woman did not select the brand with the premiumon the firsttrial, the premiumwas increasedby one penny per trial for each subsequenttrial until the panel member selectedthe desiredbrandor the experimentended. It was anticipatedthat the experimentwould begin with a period of search or exploratorybehavior,during which the selection of any brand would decrease the probabilitythat the same brand would be selected again. Following this period, it was believed that the selection of any brand would increase the likelihood that the same brandwould be selectedon the following trial.It was furtheranticipatedthat brandloyaltywould be establisheddespitehomogeneityof productand that the degree of loyalty indicated by three successive choicesof the samebrandmightbe measurablein terms of the number of cents required to cause a change in selection. investigating. The present experiment differs from most of the previous work in that it suffers the disabilitiesof obvious artificialityat the same time that it gains the capacityto examinesome of the above aspectsof brand loyalty not open to observationin the "real" world. The experimentconsisted simply of twelve successive consumerchoices of breadfrom amongfour previously unknownbrands.Forty-twowomen participatedin the experiment.They were selected by two-stage random samplingfrom a single census tract in order to minimize deliveryproblems.Each womanwas told that the study was designedto find out how women went about purchasingwhen they moved to a new location and were faced with unfamiliarbrands.She was then asked to select one of four loaves of breadmarked"L," "M," "P,"and "H."Packagingwas otherwiseidentical.Letter designationswere chosen for ease of memory.And, althoughall are consonantsfrom the middleof the alphabet and have approximatelythe same frequency of use in the Englishlanguage,it is not assumedthat they are "neutral"symbols. In fact, it seems probablethat no set of symbols which are discriminablecan be neutralor equally pleasing or have common meanings for all individuals. The position of the brands on the tray was varied in Latin squaredesignso that no brandwas in the same position two times in a row and so that each brand occurred in each position with equal frequency. The bread used for all brandswas identical, sandwich-loaf, thin-sliced bread, taken from a single oven on the morningof delivery. THE PERIOD OF SEARCH The evidence for a period of search or exploratory consumer behavior is clear. During the first several choices, the sequence of two choices of the same brandwas far below that expectedby chance. Twentynine of the forty-twowomen systematicallytried each of the brandsin orderduringtheir first four selections. Of course, it should be rememberedthat the nature of the experimentseems to suggest that a complete exploration is the sensible thing to do. Several of the panel members indicated in advance that they had already made the decision to try each of the four brands. There is no clear indication when the period of search ends for an individual.Some of the housewives completed 12 successivechoices without ever selecting the samebrandtwice in succession.Suchbehaviorcould equallywell be describedas extendedsearchor indifference. At the same time, patternsfor the entire group suggest that the first four choices were qualitatively differentfrom the remainingselections.Table 1 shows the relative frequencyof repeatinga choice on each of the 12 trials. Included in the data are only those persons who have not become brand loyal and who have a brandrun lengthof one. Clearlysomethinghappensafter the fourthchoice in termsof willingnessto repeatthe selectionof any given brand. The indication is not, however, that search activityhas ended for all participants.The relativefrequency of repeating the fourth choice on the fifth trial (given no earlierrepeatof the third choice on the 34 JOURNALOF MARKETING AUGUST1964 RESEARCH, Table I RELATIVE OF SELECTING X BRAND,GIVEN FREQUENCY THEPRECEDING SELECTION OF X, BUTNOT XX Trial number Relative frequency 2 3 4 5 6 7 8 9 10 11 12 .048 .050 .026 .210 .300 .346 .288 .240 .136 .261 .105 fourth trial) remains below the expected value of .25, although not significantly so. (The relative frequency of a repeated choice of this sort up to the fourth trial is significant far above the .10 level, is in fact about 31 standard errors below the expected.) The tailing off in the relative frequency of a repeated selection during the later trials is in part the consequence of growth in brand loyalty. Women who had become brand loyal no longer have brand run lengths of one except under the influence of premiums. Therefore, persons who had become brand loyal (had selected the same brand three times in a row) were eliminated from the data shown in Table 1 for all trials subsequent to the establishment of brand loyalty. out the experiment was .396, significantly above chance at the .05 level. While it is impossible to state conclusively the stage of brand loyalty reached by those women who had not selected any brand on three consecutive trials, there were suggestions that some women had established a relative loyalty to two of the brands (ending trials of M, H, M, H, M and P, H, P, H) and that some had essentially eliminated one of the brands from further consideration (six of the non-loyal women selected one of the brands only once in the twelve trials). The extremes in non-loyalty to brand were provided by four women who developed position loyalties of considerable duration. Position loyalties were not simple to judge, since participants could perform exploratory behavior among brands by the simple expedient of selecting from the same tray position for consecutive trials. Where continuous selection from a single posi- tion continuedfor a long period or where it emerged Exactly half of the women engaged in the experiment reached the criterion of brand loyalty by the end of twelve trials. These loyalties emerged at every possible point. Table 2 shows the frequency with which brand loyalty emerged on the various trials. as an apparent solution to the decision process during the later trials, there seems rather conclusive evidence for stating that position loyalties did emerge. (It should be pointed out that all position loyalties were for the tray position at the participant's extreme left.) A conclusion that seems almost inescapable is that women vary greatly in their susceptibility to brand loyalty. In some, the sort of behavior referred to as brand loyalty seems to have become functionally autonomous and may be a preferred form of behavior in any applicable situation.1 At the other extreme there appear to be those (suggested here by the positionloyal participants) to whom brand differences unaccompanied by product differentiation are inconsequential. This variation among women in susceptibility to brand loyalty may be one of the major consumer variables which face the marketer of certain kinds of products. Table 2 THE STRENGTH OF LOYALTIES THE EMERGENCE OF BRAND LOYALTY THETRIALSON WHICHWOMENCOMPLETED THEIR OF ANY BRAND THIRDCONSECUTIVE SELECTION Trialnumber Numberof women 3 4 5 6 7 8 9 10 12 12 I 2 I 3 1 3 3 2 4 4 It can undoubtedly be presumed that additional women would have reached the criterion of brand loyalty with further trials. Two of the women not previously brand loyal selected the same brand on trials eleven and twelve. The probability of selecting a brand the third time given two consecutive selections through- There is a strong temptation to believe that when a woman encounters a preferred brand and another brand at a reduced price (simulated in this experiment by the addition of one or more pennies to a competing brand) her choice is somehow limited to those alternatives. Of course, behavior in the marketplace constantly demonstrates that this is not the case. In the present experiment, six of the women who switched from the brand to which they had become loyal, switched first to a brand that did not contain the premium of one or more pennies. Four of these switched to a non-premium brand when the first penny was placed on the premium brand. One switched after two cents was applied, one after three cents. Undoubtedly a gestalt psychologist would suggest that the addition of 1 Cunningham'sposition that no important number of shoppers is prone to brand loyalty is based on quite different data [1] and does not really contradict this position. THEDEVELOPMENT OF BRANDLOYALTY 35 a premiumto any brand restructuredthe entire situation. As one woman who changed,but not to the premium brand,said, "No wonderyou put the special on brand'P.' It's the worstone of all."To her the premium apparentlywas a signal to do something,but anything ratherthanthe encouragedaction. Six of the brand-loyalwomen switched to the premium brandfor premiumsvaryingfrom 2-7 cents. The average value of these accepted premiums was 3/2 cents. Eight women were still selecting their favorite brandwhen the experimentended. Premiumsthat they refused on the last opportunityvaried from 1-7 cents. Theiraveragewas 3/2 cents. Even the women who had establishedpositionloyalties showedresistanceto change.Three of these women were offered premiums. Two who were offered premiums on a different position after five consecutive choices from the same position switched to the premium position for premiums of one cent and two cents. The third woman,for whom premiumswere begun aftereight successiveselectionsfrom the same position, shifted to a choice from another location when the premiumreachedthree cents. She did not select a loaf from the premiumposition at any time, although a total of five cents was placed on the brand in that positionat the time of her last choice. Obviously because of the variation in response to premiumsand the small numberof women involved,it is impossibleto drawany generalconclusionsaboutthe averagestrengthof brandor position loyalties that developed duringthe experiment.At the same time, it is clear that such loyalties are more than trivial, even thoughthey are based on what may seem trivialdistinctions. Table3 ORDEROF TRIALOF BRANDSTO WHICH PARTICIPANTS BECAMEBRAND-LOYAL Rank order in which brand was chosen Number of women who became loyal First Second Third Fourth 8 6 2 5 OTHER FINDINGS Whilethere is no satisfactoryevidencethat the order of choice affectsthe likelihoodthat a particularbrand will be the one to which a woman becomes loyal, Table 3 showsthe relevantdistributions. CONCLUSIONS An exploratoryexperimentsuch as the one reported here is more often suggestivethan conclusive.The followingconclusionsare, therefore,tentative: 1. Some consumers will become brand loyal even when there is no discriminabledifference between brandsother than the branditself. 2. The brand loyalty established under such conditions is not trivial, although it may be based on what are apparently trivial and superficial differences. 3. Consumers vary greatly in their susceptibility to brandloyalty. 4. Brand loyalty and preference for particularproduct characteristics are quite different considerations that together make up what is normally referred to as brand loyalty. 5. While it is difficult to identify exploratory consumer behavior,it seems clear that some consumer selections are largely exploratory in nature and may indicate that a repeat purchase is highly unlikely. REFERENCES 1. Ross M. Cunningham, "BrandLoyalty-What, Where, How Much?" Harvard Business Review, 34 (1956), 116. 2. Ronald E. Frank, "Brand Choice as a Probability Process," Journal of Business, 35 (1962), 43. 3. Frank Harary and Benjamin Lipstein, "The Dynamics of Brand Loyalty: A Markovian Approach," Operations Research, 10 (1962), 17. 4. Jerome D. Hertiner and John F. Magee, "Customer Behavior as a Markov Process," Operations Research, 9 (1961), 105. 5. Alfred A. Kuehn, "Consumer Brand Choice-A Learning Process" in Ronald E. Frank, Alfred A. Kuehn and William F. Massy, Quantitative Techniques in Marketing Analysis, Homewood, Ill.: Richard D. Irwin, 1962. 6. James 0. Peckham, "The Consumer Speaks," Journal of Marketing, 27 (1963), 21. 7. Edgar A. Pessemier, "A New Way to Determine Buying Decisions," Journal of Marketing, 23 (1959), 41.