Brand, Identity and Reputation: Exploring, Creating New Realities ...

derived map reflects both positive and negative views of the McDonald‘s brand, with two core associations that are
strongly linked (shown as the thickest, black lines) being ‗unhealthy‘ and ‗fast‘. As it stands, the consensus map tells us
that consumers hold very complex and at times conflicting set of associations that differs markedly from the clear cut
structure as shown in Figure 1. This poses a question; is the brand incoherent or is the consensus map confounding
more coherent ‗groups‘ of consumer views about the brand. We consider this in terms of theory and mapping research
next.
Segmentation Using Brand Concept Mapping
The seminal works of Smith (1956) and Wind (1978) have directed marketers to expect that markets are not
homogeneous, but need to be divided into subgroups which share common features and are different from other
subgroups. This is at odds with John et al.‘s (2006) Brand Concept Mapping (BCM) approach, where individual brand
maps are aggregated to produce a single consensus map, like the one illustrated in Figure 2.
Apart from being debatable in segmentation terms, in mapping terms too, aggregated maps constructed from an entire
population have been shown to provide a poor reflection of some individual maps (Henderson, Iacobucci & Calder,
1998).
To improve the BCM approach we therefore amend John et al.‘s (2006) methodology by developing a post hoc
segmentation to produce internally coherent segments based on differing brand maps from the wider sample. A standard
clustering method (Ward‘s minimum variance method (Ward & Reingen, (1990)) is used to generate BCM segment
maps shown below.
The segmentation process produced seven discrete sub-groups of respondents based on their associations and the way
these linked together. These sub-groupings were then subjected to the aggregation process as recommended by John et
al (2006) to produce the seven maps which are shown in Figure 3 (Appendix 1).
Analysis of Segment Attractiveness
By ‗eyeballing‘ the seven maps it is evident that the segments differ in two broad ways. Firstly, some maps are more
dense than others (i.e. they exhibit more core, secondary and tertiary associations with the McDonald‘s brand).
Secondly, some maps appear to have more positive (less negative) associations than others. What is needed however is
an objective way of integrating strength and favourability to objectively measure segment attractiveness.
In order to measure segment attractiveness, we need to consider both the strength and favourability of each segment
map. Previous research (Anderson, 1983b; Krishnan, 1996) indicates that the strength of a map depends on both the
number of associations and also importantly on the number and strength of links between these associations. For each
segment map, we counted the number of associations present and also the total number of links present – taking account
of single, double and triple weighted links. In order to be able to compare between maps, both measures, i.e. the
number of associations and total number of links was normalized by dividing by the maximum over all seven segment
maps.
As alluded to earlier, favourability was measured for each association by averaging the individual favourability score (-
1, 0 and +1) indicated by respondents when creating their maps. In order to produce an overall favourability measure
for each segment map, we simply averaged the favourability scores for each association in the map, thereby producing a
value in the range -1 (all associations considered unfavourable by all respondents) through to +1 (all associations
regarded positively by respondents).
Normalised Association Score = Number of Associations / Maximum Number of Associations over all maps
Normalised Links Score = Sum of Weight of Links / Maximum Sum of Weight of Links over all maps
The relative attractiveness of each segment map can then be measured by multiplying the aforementioned measures
together. The resulting attractiveness score ranges between -1 and +1. The closer to unity, the more attractive the
segment map, whilst those with scores approaching -1 would represent segments with a large number of highly linked
unfavourable associations
Attractiveness = Normalised Association Score x Average Favourability x Normalised Links Score
TAKE IN TABLE 1 ABOUT HERE
Findings
232