Академический Документы
Профессиональный Документы
Культура Документы
Jenni Romaniuk
is 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 also
involved in measurement and strategic issues in brand research.
Byron Sharp
is Director of the Marketing Science Centre and Worldwide Director of the Centres R&D Initiative for Marketing. Participating
companies, around the world, include Procter & Gamble, Coca-Cola, General Motors, Commonwealth Bank of Australia,
Unilever, AC Nielsen and many others.
218 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218229 Henry Stewart Publications 0967-3237 (2003)
Measuring brand perceptions: Testing quantity and quality
these scores either indicate current or memory and thus become part of that
future market performance or that they brands image.9
actually cause current/future
performance. (The practice also assumes
that these perceptual scores can be BRAND IMAGE AND CUSTOMER
influenced by the managers in question.) LOYALTY
This is because of the belief that the The following section presents three
perceptions individual customers hold theories regarding how brand perceptions
somehow affect their buying behaviour. can influence buyer behaviour. These
Generally, it is held that the more theories have been developed drawing
positively the brand is perceived by the from relevant marketing, psychology and
marketplace, the more customers will memory literature. From these are
buy. developed three hypotheses, which are
This paper investigates the relationship subsequently tested. The ability to
between brand image and customer develop and test (potentially) competing
loyalty and broadly asks the question: do hypotheses is a crucial aspect of this
customers who hold different perceptions research. The use of competing
exhibit different levels of loyalty? The hypotheses reduces the risk of
concern here is with one type of loyalty, confirmation bias.10,11
differentiation loyalty, which is the The first two hypotheses concern the
sensitivity of a customer to competing quality of the perceptions held of the
brands. This is operationalised as the brand, while the third hypothesis focuses
customers self-reported likelihood of on the quantity of perceptions devoted
defecting to a competing brand.3 The to the brand.
investigation reported here concerns a
subscription market where sole brand
loyalty is the norm (eg home loans, Single attribute positioning
electricity supply) and so where buying It has been proposed that not all
another brand is an act of defection perceptions are as important as others,
rather than being a part of a process of that there are certain perceptions that can
cycling through a repertoire of acceptable operate as triggers for purchase. That is,
brands (such as occurs in grocery if a customer holds a specific
goods).4 belief/perception about the brand, they
Brand image: brand perceptions can will buy it. Therefore, from a marketing
come from a variety of sources, perspective, there is benefit in getting
including consumer experiences, more people to link the brand with that
marketing communications and/or word attribute. For example, if a customer
of mouth.5 They can consist of thinks Sprint (brand) offers good value
descriptive information (eg comes in a for money (perception), then largely
blue carton), benefits (eg will not raise irrespective of what else the customer
cholesterol), evaluations of specific thinks about Sprint, the customer is
aspects of the brand (eg provides likely to purchase from that brand.12,13
excellent service) and/or This brand belief can be unique to the
purchase/consumption situations (eg a brand (ie I perceive Sprint is good value
treat for the kids).68 Essentially, any for money, but I do not perceive this
information that is encountered with the about other brands) or relative to other
brand name can, if sufficiently processed, brands (ie I believe that Sprint is better
become linked to the brand name in value for money than other brands).
Henry Stewart Publications 0967-3237 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 219
Romaniuk and Sharp
220 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218229 Henry Stewart Publications 0967-3237 (2003)
Measuring brand perceptions: Testing quantity and quality
face value to be desirable positions were links to attributes that could become
tested. The hypothesis is as follows: brand retrieval cues (ie the brands share
of consumer mind) increases the
H2: There will be specific clusters of probability of retrieval (ie greater brand
attributes that will be related to higher salience). In contrast to the prior two
loyalty to the brand. theories, this places less emphasis on the
nature/quality of the specific
attributes/beliefs and more emphasis on
Brand salience/share of mind the customers propensity to mention the
The third theory draws on a relatively brand, of the quantity of information in
small stream of research that has memory. The attributes provided in any
examined the role of the number of research are not meant to be
attributes associated with the brand and all-inclusive, but rather sufficient to
brand choice. This theory draws on the provide an indication of the brands
Associative Network Theories of ability to be retrieved by the consumer.
Memory (ANT).19 ANT suggests that This leads to the final hypothesis:
information in memory consist of
concepts that are linked together in a H3: There will be a positive
network. Retrieval of information relies relationship between the number of
on the stimulus of a concept, which then image attributes the brand is associated
activates connected nodes in a spreading with and loyalty to that brand.
activation phenomenon.20 These links all
have a chance of being retrieved at any
point in time, but only a certain amount METHOD
of information will be.21 In a buying The market: the research was conducted
context, these attributes have the in a subscription market (such as banking
potential to act and interact as retrieval or insurance), where consumers typically
cues or evaluation influencers in choice used only one brand for the service. In
situations.22 Thus the more attributes to this instance the corporate brand and the
which the brand is linked, the greater brand under which customers buy the
the likelihood the brand will be retrieved service are the same. In this context the
and fit the criteria of interest to the use of an alternative provider constitutes
customer, therefore chosen. defection from the original brand, rather
Some empirical support has emerged than cycling through a repertoire.25 As the
for this theory. In a series of experiments, sample was drawn from the total
a positive relationship has been found population, the sample sizes vary across
between the number of attributes brands according to their size in the
associated with the brand, and marketplace. The specific sample sizes for
subsequent choice of that brand.23 Other each of the brands were 4,000 for Brand
researchers who have examined quantity, 1, 900 for Brand 2 and 350 for Brand 3.
rather than quality, of information have These three brands represent over 95 per
also found encouraging results.24 In some cent of the total share of the market. The
instances it has been suggested as an three brands are in direct competition
indicator of the strength of an overall with each other in the marketplace.
attitude. The authors, however, take a The respondents: the participants in
more retrieval/memory theory approach the research were business users of a
and refer to this as a salience/share of service who were screened to check they
mind measure. That is, the increasing had influence over the selection of the
Henry Stewart Publications 0967-3237 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 221
Romaniuk and Sharp
supplier for this service. Respondents than leaving, ie being loyal was
were contacted, recruited and considered the inverse of the probability
interviewed via telephone by professional of defecting from the brand.
market research interviewers. Respondents gave loyalty scores that
The attributes: both the ad agency ranged from nought to ten, ie, from no
and market research department of chance of staying with the brand through
Brand 1 selected the brand image to 100 per cent chance of staying. Thus
attributes. These were the attributes any of the image attributes perceptions
considered to be important in customer had the potential to distinguish between
decision making. Details of the specific highly loyal and very disloyal customers,
attributes are contained in the results however it turned out that no individual
section. They covered a variety of image attribute or even bundle of
different aspects of brands that were attributes has this much discriminatory
perceived to be relevant to buyers in power.
this market. This included facets such The questions regarding the image
as pricing, service, expertise and attributes and loyalty to the brand were
relationship quality. The attributes were separated by approximately five minutes
drawn from the market and effort was of questions on other topics relevant to
made to ensure that attributes relevant the market.
to all three brands (and not just one
brand) were included.
The measures: image attribute Results for hypothesis 1: Single
responses were collected using a free attributes
choice, picking any format where both Each image attribute for each brand was
brands and perceptions are provided to analysed by dividing respondents into
respondents.26 That is, respondents were two groups, (1) those who associated the
presented with an image attribute (eg brand with that attribute and (2) those
tastes good) and asked which, if any, of who did not.
the listed brands they associated with this The difference in loyalty scores for
attribute. This has been found to be a each group by individual attribute for the
valid and reliable method of collecting three brands is shown in the Table 1.
perceptual data;27 it is also commonly ANOVA tests were used to determine if
used in the market research industry. All the differences in loyalty means between
attributes were positive and evaluative in these two groups were statistically
nature.28 significant.
Brand loyalty was captured using a Examining Table 1, it is apparent that
derivative of the verbal probability scale, the majority of attributes/brand
an 11-point probabilistic measure of relationships coincide with a higher
brand switching. This is derived from the loyalty to the brand, though not much
Juster scale for administration via higher. None of the image attributes
telephone29,30 and has been tested in seem to act as a trigger or major
terms of its reliability and validity and determinant of defection, so that those
has also been shown to have a strong respondents who perceived the brand in
ability to predict future loyalty/defection this way are much less likely to defect.
behaviour.31 To make it a loyalty The mean loyalty difference between
indicator, the probability of switching respondents who did and did not
was subtracted from the maximum of ten associate the brand with the attribute is
giving the probability of staying rather only 0.6, less than a single scale point
222 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218229 Henry Stewart Publications 0967-3237 (2003)
Measuring brand perceptions: Testing quantity and quality
(and 6 per cent difference in defection the study. Given, however, that people
probability). There is more variation knowledgeable about the market and the
between attributes (0.4 to 1.0) than brands chose the attributes it seems
between brands (0.8 to 0.6) but neither unlikely that any other attributes would
varies dramatically from the overall provide dramatically different results.
mean. There are some attributes that
have a higher difference than other
attributes, but these seem to encompass a Results for hypothesis 2: Multiattribute
group of about 6 or 7 attributes, and positioning
these groups of attributes are the same To test this hypothesis a series of multi-
for all three brands. So, while association attribute positions (combinations of up
with a brand attribute is positive, in to four attributes) were examined. If
absolute quantitative terms there appear there is a relationship between having a
to be minimal differences between particular position and loyalty a greater
brands. So, while it is possible to identify positive difference in loyalty would be
attributes which have a stronger expected if a customer perceives a brand
relationship with loyalty, it is difficult to in this position (ie has all of these
isolate a specific attribute for each brand attributes) than if it is not. The following
that would be dramatically better than eight positions were tested. The first five
any other attribute for that brand, or are centred around a single theme, while
would not also be suitable for other the final three were combinations of
brands in the market. different themes:32
This provides some limited empirical
support for Hypothesis 1. Some attributes service based (three attributes)
do seem a tiny bit more associated with expertise based (three attributes)
loyalty than others. information/solution based (three
Of course this does not preclude the attributes)
possibility that the key drivers are relationship based (three attributes)
attributes other than those included in pricing based (two attributes)
Henry Stewart Publications 0967-3237 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 223
Romaniuk and Sharp
combination 1 (value, service, be minimal (they are all very close to the
expertise and listening) mean score of 1.5). No brand seems to
combination 2 (proactive, price have an ideal position for its customer
competitiveness, partnership and ease base. Of course, this is a subjective
of working together) assessment, but it would be a tall order to
combination 3 (responsive, argue that any difference is meaningful.
understanding, solutions and value). From a marketing management
perspective, based on these results, there is
As previously mentioned, loyalty is little justification for any brand for
measured on a 010 scale. For each choosing one position over another.
brand and position the loyalty of
customers who perceived the brand to
be associated with all of the attributes in Results for hypothesis 3: Brand
the position was compared with the salience/share of mind
loyalty of customers who did not To test Hypothesis 3, a calculation was
perceive the brand to be associated with made of the number of times the
any of the attributes in the position. For brand was mentioned across all of the
example, if those who perceived the attributes presented to the respondent.
brand in that position had a loyalty of This is referred to as the brand
8.4 and those who did not perceive the salience. This ranged from zero times
brand to be in that position had a loyalty (the brand not mentioned for any of
of 6.4, the difference would be 2.0 scale the attributes) to mentioning the brand
points. It is this difference in loyalty that up to 19 times (the brand mentioned
is shown in Table 2. for all attributes). The mean loyalty at
Examining the results across all brands, each level of brand salience was
the difference in loyalty ranges from 0.7 calculated and revealed a positive
to 2.1; however, for most brands on relationship between the number of
most positions the difference is about 1.5 attributes the brand was associated with
(15 per cent difference in loyalty and loyalty. Brand loyalty was measured
between those who did associate the on a 010 scale, which is the range
brand for all the attributes in the cluster (shown on the y axis). This provides
and those who did not). support for Hypothesis 3. Figure 1
While it is possible to identify a shows this relationship.
position with the greatest difference in Further examining Figure 1, it is
loyalty for each brand, the difference evident that this relationship is also
between that position and others looks to relatively consistent between brands,
224 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218229 Henry Stewart Publications 0967-3237 (2003)
Measuring brand perceptions: Testing quantity and quality
10
Brand loyalty
6
0
0 5 10 15 20
Brand salience
particularly when sample size is taken This consistent linear relationship also
into account (the brands with the most explains the lack of variation found
variability are the smaller sample size between attributes and positions in
brands). testing the single and multiple attribute
To quantify the relationship between positioning theories.
the two variables, regression at the
aggregate level was undertaken, ie the
data points in Figure 1 (see Table 3). A DISCUSSION
relatively consistent relationship between An overall assessment of the analysis
perceptions and loyalty across brands was appears to fail to support the view that
evident. The baseline loyalty is there are particular brand perceptions, or
approximately six out of ten. Each combinations of perceptions/positions,
additional time a brand is mentioned as a that are more (or even less) associated
perceptual response, the loyalty increases with loyalty. Instead a consistent linear
by about 0.1 points out of ten. It is also association is shown between the number
possible to see a slight brand size effect, of image attributes that a respondent
with the incremental increase in loyalty associates the brand with and their
for each additional brand mention loyalty to that brand. And this is enough
increasing as brand size increases. Given to explain why clusters of attributes did
the large difference in the number of better at distinguishing between loyal
users for each brand, the difference in B customers and those vulnerable to defect
values is, however, relatively minor. simply because to mention all the
Henry Stewart Publications 0967-3237 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 225
Romaniuk and Sharp
attributes in a cluster meant a respondent that further testing to see if the basis of
was mentioning the brand more. this relationship is salience, attitude or
The total number of times a brand is some combination of the two would be
mentioned by respondents in an image a useful endeavour to improve
survey is a relatively ignored method of understanding of how consumer
measuring brand performance that perceptions influence future behaviour.
appears to warrant inclusion in any tool It is also worth noting that these
to assess brand image. results complement that of past research,
These findings support firms which found (despite aggregate stability)
undertaking market research into brand a systematic instability in the association
image/perceptions because it is important of specific attributes with specific brands
for a brand to be well known by its at individual level.35 Systematic instability
customer base. These findings suggest, of brand mention at individual level calls
however, that it is less important into question the ability of a specific
specifically how the brand is perceived at brand/attribute link to be crucial in the
any one point in time (so long as it is future purchase decision.
vaguely positive). That is, brand managers
should not be concerned if customers think
of the firm as offering competitive prices, IMPLICATIONS
good service or experts in the field, just as It is suggested that marketers should be
long as they know something about the looking to maintain and increase the
firm and the more the better. salience of their brands in the minds of
This finding does fit in with what is customers; that is, to develop and
understood about how consumers hold reinforce the breadth of the network
information about brands. That is, in the about the brand in consumer memory.
form of an associative network33 where This takes the focus away from
the brand name is linked to relevant specifically what attributes customers
attributes. This provides the mechanism associate with brands and towards how
for retrieval for brands in choice many attributes customers associate with
situations. Consumers use these attributes brands. This provides greater scope for
as retrieval cues for accessing choice variation of the messages sent out to
information. Therefore, it seems the customers, which should provide more
greater the share of mind (regardless of opportunities to create entertaining and
the attributes actually linked to the brand useful advertising.
name) the more loyal the customer. Techniques such as perceptual mapping
It could be argued that the may be used as a stimulus for new ideas
multiattribute measure used here is about creative advertising content.
simply a crude measure of attitude in the Indeed the results here highlight a
form of a sum of the beliefs a customer benefit of perceptual mapping and image
holds about the brand.34 Therefore, it is analysis, that is, identification of attributes
the strength of the attitude to the brand that the brand is not currently associated
that is related to future behaviour. This with. If the objective is to ensure that
is a possibility that needs further customers link the brand with as many
empirical testing to distinguish clearly (non-negative) attributes as possible, then
between a salience/retrieval and an it is necessary to identify those attributes
attitudinal perspective for interpreting where improvement is possible.
the empirical results presented here. The The results of this research also suggest
results of this paper do suggest, however, a new way to measure the performance
226 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218229 Henry Stewart Publications 0967-3237 (2003)
Measuring brand perceptions: Testing quantity and quality
Henry Stewart Publications 0967-3237 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 227
Romaniuk and Sharp
228 Journal of Targeting, Measurement and Analysis for Marketing Vol. 11, 3, 218229 Henry Stewart Publications 0967-3237 (2003)
Measuring brand perceptions: Testing quantity and quality
Henry Stewart Publications 0967-3237 (2003) Vol. 11, 3, 218-229 Journal of Targeting, Measurement and Analysis for Marketing 229