Measuring brand perceptions: Testing quantity and quality

attributes) to be chosen as desirable, thusleading to marketing objectives andsubsequent advertising activities to‘reposition’ the brand/firm.1,2 Firms alsotrack changes in the proportion ofrespondents that hold particularperceptions about the firm/brand.‘Scores’ on particular image perceptionsare frequently built into managers’performance targets (key performanceindicators). This practice assumes that

INTRODUCTIONThe belief that brand perceptionsstrongly influence buying behaviour iswidespread, and this belief underpinsmuch of a business’s marketingcommunication strategy. Firms look toimage studies to explain currentmarketplace performance, eg numbers ofunits sold and price gained. It iscommon for particular positions (ie aparticular combination of image

218 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218–229 � Henry Stewart Publications 1479-1862 (2003)

Measuring brand perceptions:Testing quantity and qualityReceived (in revised form): 21st November, 2002

Jenni Romaniukis a Senior Research Fellow with the Marketing Science Centre, a research centre based at the University of South Australia.Her key research areas include brand image, salience and how brand perceptions influence buyer behaviour. She is alsoinvolved in measurement and strategic issues in brand research.

Byron Sharpis Director of the Marketing Science Centre and Worldwide Director of the Centre’s R&D Initiative for Marketing. Participatingcompanies, around the world, include Procter & Gamble, Coca-Cola, General Motors, Commonwealth Bank of Australia,Unilever, AC Nielsen and many others.

Abstract The image of a brand is considered to be important as is evident from thevast sums of money spent by companies on the development and measurement of theircorporate/brand image. Yet very little is known about the relationship between brandperceptions and buyer behaviour. The authors empirically tested three hypotheses aboutthe relationship between brand perceptions and loyalty. They found that (a) there waslittle evidence that any particular attributes are more related to customer loyalty thanany others nor (b) that there were specific brand positions that were uniquelyassociated with higher loyalty. They did, however, find the more attributes associatedwith a brand the more loyal the customer. This is a relatively unexplored effect of brandperceptions, which should be included in brand tracking, and has some profoundimplications for marketing practice. It suggests that while distinctiveness is useful inmaking sure that the brand’s marketing activities are noticed and correctly branded, thesource of that distinctiveness is a less important marketing decision. Finally, theyrecommend that there should be different long and short-term goals for brand building.In the short term a choice may be made to focus on specific attributes. In the longterm, however, marketers should work towards building the number of links between thebrand and attributes in the market place, ie building the brand’s share of mind.

Jenni RomaniukMarketing Science Centre,University of SouthAustralia, PO Box 2471,South Australia 5001.

Tel: �61 2 9380 9146;Fax: �61 2 9475 0203;e-mail: [email protected]

memory and thus become part of thatbrand’s image.9

BRAND IMAGE AND CUSTOMERLOYALTYThe following section presents threetheories regarding how brand perceptionscan influence buyer behaviour. Thesetheories have been developed drawingfrom relevant marketing, psychology andmemory literature. From these aredeveloped three hypotheses, which aresubsequently tested. The ability todevelop and test (potentially) competinghypotheses is a crucial aspect of thisresearch. The use of competinghypotheses reduces the risk ofconfirmation bias.10,11

The first two hypotheses concern the‘quality’ of the perceptions held of thebrand, while the third hypothesis focuseson the quantity of perceptions devotedto the brand.

Single attribute positioning

It has been proposed that not allperceptions are as important as others,that there are certain perceptions that canoperate as ‘triggers’ for purchase. That is,if a customer holds a specificbelief/perception about the brand, theywill buy it. Therefore, from a marketingperspective, there is benefit in gettingmore people to link the brand with thatattribute. For example, if a customerthinks Sprint (brand) offers good valuefor money (perception), then largelyirrespective of what else the customerthinks about Sprint, the customer islikely to purchase from that brand.12,13

This brand belief can be unique to thebrand (ie I perceive Sprint is good valuefor money, but I do not perceive thisabout other brands) or relative to otherbrands (ie I believe that Sprint is bettervalue for money than other brands).

these scores either indicate current orfuture market performance or that theyactually cause current/futureperformance. (The practice also assumesthat these perceptual scores can beinfluenced by the managers in question.)This is because of the belief that theperceptions individual customers holdsomehow affect their buying behaviour.Generally, it is held that the morepositively the brand is perceived by themarketplace, the more customers willbuy.

This paper investigates the relationshipbetween brand image and customerloyalty and broadly asks the question: docustomers who hold different perceptionsexhibit different levels of loyalty? Theconcern here is with one type of loyalty,differentiation loyalty, which is thesensitivity of a customer to competingbrands. This is operationalised as thecustomer’s self-reported likelihood ofdefecting to a competing brand.3 Theinvestigation reported here concerns asubscription market where sole brandloyalty is the norm (eg home loans,electricity supply) and so where buyinganother brand is an act of defectionrather than being a part of a process ofcycling through a repertoire of acceptablebrands (such as occurs in grocerygoods).4

Brand image: brand perceptions cancome from a variety of sources,including consumer experiences,marketing communications and/or wordof mouth.5 They can consist ofdescriptive information (eg comes in ablue carton), benefits (eg will not raisecholesterol), evaluations of specificaspects of the brand (eg providesexcellent service) and/orpurchase/consumption situations (eg atreat for the kids).6–8 Essentially, anyinformation that is encountered with thebrand name can, if sufficiently processed,become linked to the brand name in

� Henry Stewart Publications 1479-1862 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 219

Measuring brand perceptions: Testing quantity and quality

such as perceptual mapping, clusteranalysis or factor analysis to identifyattributes that are desirable for the brandto be associated with more than (a) otherbrands are and (b) other attributes.Similar to the first theory there is theidea that some positions are ‘better’ thanothers, in that they will lead to greaterbenefits for having that position thanothers. In contrast to the first theory,however, here positioning is based onclusters of attributes, which may becentred on a single theme (eg a brandseeking to build a relationship positionmay seek to be associated with attributesabout likeability of staff and willingnessof staff to listen and perceived interest inthe customer’s business and a businesspartner), or draw from a combination ofdifferent themes (eg being associatedwith attributes of excellent service, goodvalue for money and a business partner).It has been argued that a unique mixtureof strong associations is essential for abrand’s success.17 There has been indirectempirical support for this whereby aunique set of attributes that were relatedto a brand being the primary storechoice have been found.18

This theory is tested by ‘playingmarketing manager’ and using atechnique that, in the authors’experience, marketing managers generallyuse to determine the best position fortheir brand. This is through choosingbundles of desirable attributes. If aperceptual map displaying brands andattributes was developed from imagesurvey data, managers would hope to seetheir brand positioned close to the clusterof these attributes. This was tested acrosscustomers to see if those who perceivedthe brand to have a particular position(ie would have those attributes clusteredaround the brand in their memory) had ahigher loyalty level than those who didnot perceive the brand to be in thatposition. A number of what appear at

This theory implies that marketingmanagers should search for these‘triggers’ and then develop campaignsfocusing solely on those attributes.Traditionally, the scope for positioningwas limited to different product categorycues (ie offers home loans). More recentresearch, however, has shown that brandscan be ‘brought to (the consumer’s)mind’ via any number of imageattributes. This can include situationattributes (eg at the beach), benefitattributes (eg is low in fat), country oforigin attributes (eg is Italian) or qualityattributes (eg is the best). All of theseattributes have the potential to betriggers for someone buying thebrand.14–16 So potentially any attributelinked to the brand could be a ‘trigger’and therefore could be the mostimportant message to communicate tothe marketplace.

One test for the presence of thesetriggers is to see if there are anyattributes where there is a greater linkbetween perceiving that the brand hasthe quality represented by that attributeand future buying behaviour/loyaltytowards the brand. If some attributeshave a greater relationship withloyalty/future purchase than otherattributes this would identify theattributes that would be suitable ‘triggers’for the marketing manager to focus onin marketing communications. This leadsto the first hypothesis:

H1: There are attributes (situations,benefits, quality etc) that are morestrongly related to brand loyalty thanother attributes.

Multiattribute positioning

To understand the relationships betweenbrands and attributes in the marketplace,marketers often draw on multivariateanalysis. This involves using techniques


Romaniuk and Sharp

links to attributes that could becomebrand retrieval cues (ie the brand’s shareof consumer mind) increases theprobability of retrieval (ie greater brandsalience). In contrast to the prior twotheories, this places less emphasis on thenature/quality of the specificattributes/beliefs and more emphasis onthe customer’s propensity to mention thebrand, of the quantity of information inmemory. The attributes provided in anyresearch are not meant to beall-inclusive, but rather sufficient toprovide an indication of the brand’sability to be retrieved by the consumer.This leads to the final hypothesis:

H3: There will be a positiverelationship between the number ofimage attributes the brand is associatedwith and loyalty to that brand.

METHODThe market: the research was conductedin a subscription market (such as bankingor insurance), where consumers’ typicallyused only one brand for the service. Inthis instance the corporate brand and thebrand under which customers buy theservice are the same. In this context theuse of an alternative provider constitutes‘defection’ from the original brand, ratherthan cycling through a repertoire.25 As thesample was drawn from the totalpopulation, the sample sizes vary acrossbrands according to their size in themarketplace. The specific sample sizes foreach of the brands were 4,000 for Brand1, 900 for Brand 2 and 350 for Brand 3.These three brands represent over 95 percent of the total share of the market. Thethree brands are in direct competitionwith each other in the marketplace.

The respondents: the participants inthe research were business users of aservice who were screened to check theyhad influence over the selection of the

face value to be desirable positions weretested. The hypothesis is as follows:

H2: There will be specific clusters ofattributes that will be related to higherloyalty to the brand.

Brand salience/share of mind

The third theory draws on a relativelysmall stream of research that hasexamined the role of the number ofattributes associated with the brand andbrand choice. This theory draws on theAssociative Network Theories ofMemory (ANT).19 ANT suggests thatinformation in memory consist ofconcepts that are linked together in anetwork. Retrieval of information relieson the stimulus of a concept, which thenactivates connected nodes in a ‘spreadingactivation’ phenomenon.20 These links allhave a chance of being retrieved at anypoint in time, but only a certain amountof information will be.21 In a buyingcontext, these attributes have thepotential to act and interact as retrievalcues or evaluation influencers in choicesituations.22 Thus the more attributes towhich the brand is linked, the greaterthe likelihood the brand will be retrievedand fit the criteria of interest to thecustomer, therefore chosen.

Some empirical support has emergedfor this theory. In a series of experiments,a positive relationship has been foundbetween the number of attributesassociated with the brand, andsubsequent choice of that brand.23 Otherresearchers who have examined quantity,rather than quality, of information havealso found encouraging results.24 In someinstances it has been suggested as anindicator of the strength of an overallattitude. The authors, however, take amore retrieval/memory theory approachand refer to this as a salience/share ofmind measure. That is, the increasing



than leaving, ie being loyal wasconsidered the inverse of the probabilityof defecting from the brand.

Respondents gave loyalty scores thatranged from nought to ten, ie, from nochance of staying with the brand throughto 100 per cent chance of staying. Thusany of the image attributes perceptionshad the potential to distinguish betweenhighly loyal and very disloyal customers,however it turned out that no individualimage attribute or even bundle ofattributes has this much discriminatorypower.

The questions regarding the imageattributes and loyalty to the brand wereseparated by approximately five minutesof questions on other topics relevant tothe market.

Results for hypothesis 1: Singleattributes

Each image attribute for each brand wasanalysed by dividing respondents intotwo groups, (1) those who associated thebrand with that attribute and (2) thosewho did not.

The difference in loyalty scores foreach group by individual attribute for thethree brands is shown in the Table 1.ANOVA tests were used to determine ifthe differences in loyalty means betweenthese two groups were statisticallysignificant.

Examining Table 1, it is apparent thatthe majority of attributes/brandrelationships coincide with a higherloyalty to the brand, though not muchhigher. None of the image attributesseem to act as a ‘trigger’ or majordeterminant of defection, so that thoserespondents who perceived the brand inthis way are much less likely to defect.The mean loyalty difference betweenrespondents who did and did notassociate the brand with the attribute isonly 0.6, less than a single scale point

supplier for this service. Respondentswere contacted, recruited andinterviewed via telephone by professionalmarket research interviewers.

The attributes: both the ad agencyand market research department ofBrand 1 selected the brand imageattributes. These were the attributesconsidered to be important in customerdecision making. Details of the specificattributes are contained in the resultssection. They covered a variety ofdifferent aspects of brands that wereperceived to be relevant to buyers inthis market. This included facets suchas pricing, service, expertise andrelationship quality. The attributes weredrawn from the market and effort wasmade to ensure that attributes relevantto all three brands (and not just onebrand) were included.

The measures: image attributeresponses were collected using a freechoice, picking any format where bothbrands and perceptions are provided torespondents.26 That is, respondents werepresented with an image attribute (eg‘tastes good’) and asked which, if any, ofthe listed brands they associated with thisattribute. This has been found to be avalid and reliable method of collectingperceptual data;27 it is also commonlyused in the market research industry. Allattributes were positive and evaluative innature.28

Brand loyalty was captured using aderivative of the verbal probability scale,an 11-point probabilistic measure ofbrand switching. This is derived from theJuster scale for administration viatelephone29,30 and has been tested interms of its reliability and validity andhas also been shown to have a strongability to predict future loyalty/defectionbehaviour.31 To make it a ‘loyalty’indicator, the probability of switchingwas subtracted from the maximum of tengiving the probability of staying rather


Romaniuk and Sharp

the study. Given, however, that peopleknowledgeable about the market and thebrands chose the attributes it seemsunlikely that any other attributes wouldprovide dramatically different results.

Results for hypothesis 2: Multiattributepositioning

To test this hypothesis a series of ‘multi-attribute positions’ (combinations of upto four attributes) were examined. Ifthere is a relationship between having aparticular ‘position’ and loyalty a greaterpositive difference in loyalty would beexpected if a customer perceives a brandin this ‘position’ (ie has all of theseattributes) than if it is not. The followingeight positions were tested. The first fiveare centred around a single theme, whilethe final three were combinations ofdifferent themes:32

— service based (three attributes)— expertise based (three attributes)— information/solution based (three

attributes)— relationship based (three attributes)— pricing based (two attributes)

(and 6 per cent difference in defectionprobability). There is more variationbetween attributes (0.4 to 1.0) thanbetween brands (0.8 to 0.6) but neithervaries dramatically from the overallmean. There are some attributes thathave a higher difference than otherattributes, but these seem to encompass agroup of about 6 or 7 attributes, andthese groups of attributes are the samefor all three brands. So, while associationwith a brand attribute is positive, inabsolute quantitative terms there appearto be minimal differences betweenbrands. So, while it is possible to identifyattributes which have a strongerrelationship with loyalty, it is difficult toisolate a specific attribute for each brandthat would be dramatically better thanany other attribute for that brand, orwould not also be suitable for otherbrands in the market.

This provides some limited empiricalsupport for Hypothesis 1. Some attributesdo seem a tiny bit more associated withloyalty than others.

Of course this does not preclude thepossibility that the key drivers areattributes other than those included in



Table 1: Difference in loyalty based on brand to attribute association

Brand 1 Brand 2 Brand 3 Row Mean

Economical 1.1*** 0.8*** 1.1** 1.0Good service 0.8*** 0.7** 1.4*** 1.0Cheap 0.8*** 0.8** 1.0** 0.9Listens 1.0*** 0.7** 0.9*** 0.8Ahead 0.7*** 0.7** 1.0** 0.8Easy to work with 1.0*** 0.8*** 0.3 0.7Trustable 0.8*** 0.6* 0.8** 0.7Thinks ahead 0.8*** 0.4* 0.7** 0.6Smart 0.8*** 0.4* 0.6* 0.6Knowledgeable 0.8*** 0.7** 0.4 0.6Solves problems 0.6*** 0.6** 0.5 0.6Helps 0.5*** 0.7** 0.5 0.6Works together 0.8*** 0.4 0.5 0.5Cares 0.8*** 0.4* 0.2 0.5Responds 0.8*** 0.4* 0.3 0.5Important 0.4*** 0.5** 0.3 0.4Worthwhile 0.7*** 0.1 0.2 0.3Insights 0.6*** 0.4 0.0 0.3

Column mean 0.8 0.6 0.6 0.6

***�p<0.001, **�p<0.01, *�p<0.05

be minimal (they are all very close to themean score of 1.5). No brand seems tohave an ideal position for its customerbase. Of course, this is a subjectiveassessment, but it would be a tall order toargue that any difference is meaningful.From a marketing managementperspective, based on these results, there islittle justification for any brand forchoosing one position over another.

Results for hypothesis 3: Brandsalience/share of mind

To test Hypothesis 3, a calculation wasmade of the number of times thebrand was mentioned across all of theattributes presented to the respondent.This is referred to as the brandsalience. This ranged from zero times(the brand not mentioned for any ofthe attributes) to mentioning the brandup to 19 times (the brand mentionedfor all attributes). The mean loyalty ateach level of brand salience wascalculated and revealed a positiverelationship between the number ofattributes the brand was associated withand loyalty. Brand loyalty was measuredon a 0–10 scale, which is the range(shown on the y axis). This providessupport for Hypothesis 3. Figure 1shows this relationship.

Further examining Figure 1, it isevident that this relationship is alsorelatively consistent between brands,

— combination 1 (value, service,expertise and listening)

— combination 2 (proactive, pricecompetitiveness, partnership and easeof working together)

— combination 3 (responsive,understanding, solutions and value).

As previously mentioned, loyalty ismeasured on a 0–10 scale. For eachbrand and position the loyalty ofcustomers who perceived the brand tobe associated with all of the attributes inthe position was compared with theloyalty of customers who did notperceive the brand to be associated withany of the attributes in the position. Forexample, if those who perceived thebrand in that position had a loyalty of8.4 and those who did not perceive thebrand to be in that position had a loyaltyof 6.4, the difference would be 2.0 scalepoints. It is this difference in loyalty thatis shown in Table 2.

Examining the results across all brands,the difference in loyalty ranges from 0.7to 2.1; however, for most brands onmost positions the difference is about 1.5(15 per cent difference in loyaltybetween those who did associate thebrand for all the attributes in the clusterand those who did not).

While it is possible to identify aposition with the greatest difference inloyalty for each brand, the differencebetween that position and others looks to


Romaniuk and Sharp

Table 2: Multiattribute positions and loyalty

Brand 1 Brand 2 Brand 3

Combination 2 1.9*** 1.2*** 0.7Combination 1 1.8*** 1.7*** 2.1***Combination 3 1.7*** 1.4*** 1.5***Service 1.7*** 1.2*** 1.7***Relationship 1.6*** 1.2*** 1.0#Information/Solution 1.5*** 1.0** 1.1**Expertise 1.4*** 1.6*** 1.0#Pricing 1.3*** 1.1*** 1.4***

#p<0.10, **p<0.01, ***p<0.001

This consistent linear relationship alsoexplains the lack of variation foundbetween attributes and positions intesting the single and multiple attributepositioning theories.

DISCUSSIONAn overall assessment of the analysisappears to fail to support the view thatthere are particular brand perceptions, orcombinations of perceptions/positions,that are more (or even less) associatedwith loyalty. Instead a consistent linearassociation is shown between the numberof image attributes that a respondentassociates the brand with and theirloyalty to that brand. And this is enoughto explain why clusters of attributes didbetter at distinguishing between loyalcustomers and those vulnerable to defect— simply because to mention all the

particularly when sample size is takeninto account (the brands with the mostvariability are the smaller sample sizebrands).

To quantify the relationship betweenthe two variables, regression at theaggregate level was undertaken, ie thedata points in Figure 1 (see Table 3). Arelatively consistent relationship betweenperceptions and loyalty across brands wasevident. The baseline loyalty isapproximately six out of ten. Eachadditional time a brand is mentioned as aperceptual response, the loyalty increasesby about 0.1 points out of ten. It is alsopossible to see a slight brand size effect,with the incremental increase in loyaltyfor each additional brand mentionincreasing as brand size increases. Giventhe large difference in the number ofusers for each brand, the difference in Bvalues is, however, relatively minor.



Figure 1 Brand salience and brand loyalty

Table 3: Regression results

Adj. r-squared b-value Constant

Brand 1Brand 2Brand 3

0.820.570.45

0.140.110.09

6.25.96.2

0

2

4

6

8

10

0 5 10 15 20

Brand salience

Bra

nd lo

yalty

Brand 1 Brand 2 Brand 3

that further testing to see if the basis ofthis relationship is salience, attitude orsome combination of the two would bea useful endeavour to improveunderstanding of how consumerperceptions influence future behaviour.

It is also worth noting that theseresults complement that of past research,which found (despite aggregate stability)a systematic instability in the associationof specific attributes with specific brandsat individual level.35 Systematic instabilityof brand mention at individual level callsinto question the ability of a specificbrand/attribute link to be crucial in thefuture purchase decision.

IMPLICATIONSIt is suggested that marketers should belooking to maintain and increase thesalience of their brands in the minds ofcustomers; that is, to develop andreinforce the breadth of the networkabout the brand in consumer memory.This takes the focus away fromspecifically what attributes customersassociate with brands and towards howmany attributes customers associate withbrands. This provides greater scope forvariation of the messages sent out tocustomers, which should provide moreopportunities to create entertaining anduseful advertising.

Techniques such as perceptual mappingmay be used as a stimulus for new ideasabout creative advertising content.Indeed the results here highlight abenefit of perceptual mapping and imageanalysis, that is, identification of attributesthat the brand is not currently associatedwith. If the objective is to ensure thatcustomers link the brand with as many(non-negative) attributes as possible, thenit is necessary to identify those attributeswhere improvement is possible.

The results of this research also suggesta new way to measure the performance

attributes in a cluster meant a respondentwas mentioning the brand more.

The total number of times a brand ismentioned by respondents in an imagesurvey is a relatively ignored method ofmeasuring brand performance thatappears to warrant inclusion in any toolto assess brand image.

These findings support firmsundertaking market research into brandimage/perceptions because it is importantfor a brand to be well known by itscustomer base. These findings suggest,however, that it is less importantspecifically how the brand is perceived atany one point in time (so long as it isvaguely positive). That is, brand managersshould not be concerned if customers thinkof the firm as offering competitive prices,good service or experts in the field, just aslong as they know something about thefirm — and the more the better.

This finding does fit in with what isunderstood about how consumers holdinformation about brands. That is, in theform of an associative network33 wherethe brand name is linked to relevantattributes. This provides the mechanismfor retrieval for brands in choicesituations. Consumers use these attributesas retrieval cues for accessing choiceinformation. Therefore, it seems thegreater the share of mind (regardless ofthe attributes actually linked to the brandname) the more loyal the customer.

It could be argued that themultiattribute measure used here issimply a crude measure of attitude in theform of a sum of the beliefs a customerholds about the brand.34 Therefore, it isthe strength of the attitude to the brandthat is related to future behaviour. Thisis a possibility that needs furtherempirical testing to distinguish clearlybetween a ‘salience/retrieval’ and an‘attitudinal’ perspective for interpretingthe empirical results presented here. Theresults of this paper do suggest, however,


Romaniuk and Sharp

communications at obtaining ‘cutthrough’.

Strategic brand management

There are, the authors suggest, twodistinct objectives for short-term andlong-term brand building. In the shortterm, managers need to identify a specificattribute or position for communicationto the market. The similarity in resultsacross multiple attributes suggests that thespecific attribute for a short-term focuscan be chosen based on which messageprovides the best creative execution. Thekey aim is to develop likeable adverts, sothat cut through in the marketplace canbe achieved. An important note is thateven if a manager was to take theattributes with the strongest relationshipwith loyalty, these are likely to be thesame attributes that competitors wouldconsider to be attractive for positioning aswell. This makes it particularly importantto ensure that any marketingcommunications are prominently anddistinctly branded. This makes it easier forcustomers to know who is advertising.

In the long term, the objective is tobuild up the ‘bank’ of perceptions thatconsumers hold about the brand. Thisgradual accumulation of attributes buildsthe ‘share of mind’ for the brand,making it (a) the one that will bethought of and (b) more difficult forcompetitors to have access to the mindof customers. This suggests that theremight be value in advertising counter tothe brand’s current position, a tactic atodds with ‘positioning theory’recommendations.36 This seems a worthyarea for research.

Limitations and future research

Finally, it is stressed that the authorshave investigated the relationshipbetween image and only one aspect of

of a brand, through measuring the ‘shareof mind’ it holds, or brand salience.

There are several implications formeasurement. The first is that thisfinding calls into question the practice ofundertaking brand image studies thatseek to identify and recommend the bestposition in the marketplace. It is difficultto see on what basis it can legitimatelybe claimed that one position is superiorto another, since the position in itselfdoes not seem to be related to a higherloyalty level. The second aspect is thatthe results also question the practice ofmonitoring a few ‘key’ perceptions, egwith line charts and including these askey performance indicators. Usually, sucha practice is based on the assumption thatsome attributes are particularly importantin influencing buyer behaviour. Thisresearch has found this assumption to beunsupported. The best case that can bemade is for all attributes to be monitoredso that an overall assessment of brandperformance can be made.

The findings should also not beinterpreted as suggesting that the idealposition on a correspondence analysisbased perceptual map is in the centre.While it appears that it is desirable to beassociated with as many image attributesas possible such a result would notnecessarily give a central position. It issuggested that it is important to have aspecific position, which can be any(non-negative) position, so long as it isdistinct. A distinct position does notmean that buyers will be dramaticallymore loyal to the brand. There are,however, other benefits of beingdistinctive that are not covered in thisresearch. For example, it may facilitatethe clear (and easily recognisable)branding of marketing communicationsand give consumers something by whichthey can easily categorise and rememberthe brand. Thus increasing theeffectiveness of marketing



‘Three conceptualisations of loyalty’, Australia NewZealand Marketing Educators Conference 97,Department of Marketing, Monash University,Melbourne.

4 Sharp, B. and Wright, M. (1999) ‘There are twotypes of repeat purchase markets’, 28th EuropeanMarketing Academy Conference, Institute ofMarketing, Humboldt-University, Berlin, Germany.

5 Krishnan, H. S. (1996) ‘Characteristics of memoryassociations: A consumer-based brand equityperspective’, International Journal of Research inMarketing, Vol. 13, pp. 389–405.

6 Joyce, T. (1963) ‘Techniques of brand imagemeasurement’, New Developments in Research, MarketResearch Society, London, pp. 45–63.

7 Barwise, T. P. and Ehrenberg, A. S. C. (1985)‘Consumer beliefs and brand usage’, Journal of theMarket Research Society, Vol. 27, No. 2, pp. 81–93.

8 Aaker, D. A. (1991) ‘Managing brand equity:Capitalizing on the value of a brand name’, TheFree Press, New York.

9 Keller, K. L. (1993) ‘Conceptualizing, measuring,and managing customer-based brand equity’, Journalof Marketing, Vol. 57, January, pp. 1–22.

10 Greenwald, A. G., Pratkanis, A. R., Leippe, M. R.and Baumgardner, M. H. (1986) ‘Under whatconditions does theory obstruct research progress?’,Psychological Review, Vol. 93, No. 2, pp. 216–229.

11 Wright, M. and Kearns, Z. (1998) ‘Progress inmarketing knowledge’, Journal of EmpiricalGeneralisations in Marketing Science, Vol. 3, pp. 1–21.

12 Alpert, M. I. (1971) ‘Identification of determinantattributes: A comparison of methods’, Journal ofMarketing Research, Vol. 8, May, pp. 184–191.

13 Woodside, A. G. and Trappey, R. J. (1992) ‘Findingout why customers shop in your store and buy yourbrand: Automatic cognitive processing models ofprimary choice’, Journal of Advertising Research, Vol.32, pp. 59–78.

14 Rossiter, J. R. and Percy, L. (1987) ‘Advertising andpromotion management’, McGraw-Hill, Singapore.

15 Holden, S. J. S. (1993) ‘Understanding brandawareness: Let me give you a c(l)ue!’, Advances inConsumer Research, Vol. 20, pp. 383–388.

16 Holden, S. J. S. and Lutz, R. J. (1992) ‘Ask notwhat the brand can evoke; Ask what can evoke thebrand?’, Advances in Consumer Research, Vol. 19, pp.101–107.

17 Grant, M, and Opie, T. (2001) ‘Making more thana difference’, Admap, April, WARC,Henley-on-Thames.

18 Thelen, E. and Woodside, A. G. (1997) ‘Whatevokes the brand or store? Consumer research onaccessibility theory applied to modelling primarychoice’, International Journal of Research in Marketing,Vol. 14, pp. 125–145.

19 Anderson, J. R. and Bower, G. H. (1979) ‘Humanassociative memory’, Lawrence Erlbaum, Hillsdale,NJ.

20 Collins, A. M. and Loftus, E. F. (1975) ‘A spreadingactivation theory of semantic processing’, PsychologicalReview, Vol. 82, No. 6, pp. 407–428.

buyer behaviour (loyalty), albeit anextremely important one. The resultsdo not preclude other links betweenparticular perceptions or combinationsof perceptions and behaviour. Forexample, customers who perceive abrand to be very high quality may bewilling to pay more for this brand, orparticular positions may be beneficialfor acquiring customers (the researchhere deals solely with the retention ofexisting customers). This is animportant area for future research.Additionally it may be that the levelof importance of the attributemoderates the effect of that attributefor a specific consumer. That theeffects across the whole market seemto cancel each other out suggests,however, that importance, if it is afactor, is not a very influential one ataggregate level. It may, however, haveimportant implications at individuallevel and so should be investigated.

Replication and extension across awide range of markets and conditions arerecommended, to test the generalisabilityof these findings — particularly to othertypes of markets, such as fast movingconsumer goods markets. This isparticularly so given the research of Lowand Lamb37 which suggested that therewould be variation in results acrossproduct categories, in line with changesto brand familiarity structures.

Finally, it is suggested that researchinto the effect of advertising on brandsalience would be useful, so thatmarketers can understand to what extentthey can influence memory structures inthe short term.

References1 Ries, A. and Trout, J. (1981) ‘Positioning the battle

for your mind’, McGraw-Hill Inc, New York.2 Aaker, D. A. and Shansby, G. (1982) ‘Positioning

your product’, Business Horizons, Vol. 25, pp. 56–62.3 Sharp, B., Rundle-Thiele, S. and Dawes, J. (1997)


Romaniuk and Sharp

‘Obtaining purchase predictions via telephoneinterviews’, Journal of the Market Research Society, Vol.37, No. 3, pp. 241–250.

31 Danenberg, N. and Sharp, B. (1996) ‘Measuringloyalty in subscription markets using probabilisticestimates of switching behaviour’, Australia NewZealand Marketing Educators Conference,Department of Marketing, University of Auckland,Auckland, NZ.

32 In line with Grant and Opie (2001) op. cit.33 Anderson and Bower (1979) op. cit.34 Fishbein, M. and Ajzen, I. (1975) ‘Belief, attitude,

intention and behaviour: An introduction to theoryand research’, Addison-Wesley Publishing Company,Reading, Massachusetts.

35 Dall’Olmo Riley, F., Ehrenberg, A. S. C.,Castleberry, S. B., Barwise, T. P. and Barnard, N. R.(1997) ‘The variability of attitudinal repeat-rates’,International Journal of Research in Marketing, Vol. 14,No. 5, pp. 437–450.

36 Ries and Trout (1981) op. cit.37 Low, G. S. and Lamb, C. W. (2000) ‘The

measurement & dimensionality of brand associations’,Journal of Brand and Product Management, Vol. 9, No.6, pp. 350–368.

21 Raaijmakers, J. G. W. and Shiffrin, R. M. (1981)‘Search of associative memory’, Psychological Review,Vol. 88, No. 2, pp. 93–134.

22 Holden (1993) op. cit.23 Alba, J. W. and Marmorstein, H. (1987) ‘The effects

of frequency knowledge on consumer decisionmaking’, Journal of Consumer Research, Vol. 14, pp.14–25.

24 Krishnan (1996) op. cit.25 Sharp and Wright (1999) op. cit.26 Holbrook, M. B., Moore, W. L. and Winer, R.

(1982) ‘Constructing joint spaces from pick-anydata: A new tool for consumer analysis’, Journal ofConsumer Research, Vol. 9, June, pp. 99–105.

27 Barnard, N. R. and Ehrenberg, A. S. C. (1990)‘Robust measures of consumer brand beliefs’, Journalof Marketing Research, Vol. 27, November, pp.477–487.

28 In line with Barwise, T. P. and Ehrenberg, A. S. C.(1987) ‘Consumer beliefs and awareness’, Journal of theMarket Research Society, Vol. 29, No. 1, pp. 88–94.

29 Juster, F. T. (1960) ‘Prediction and consumer buyingintentions’, American Economic Review, Vol. 50, pp.604–622.

30 Brennan, M., Esslemont, D. and Hini, D. (1995)



Documents

Measuring brand perceptions: Testing quantity and quality