In principle, we can describe the nature of spatial competition in a market either in demand terms or in supply terms. Market segmentation repre- sents the demand perspective on structure, whilst competitive positioning represents the supply perspective.
Market segmentation takes as its starting point assumptions about the differing requirements that individual customers have with respect to bun- dles of benefits in particular use situations. Most obviously in this context it is an ‘ideal’ approach in that it is effectively assumed that each customer can/does specify their own ideal benefit bundle and their purchase choice in the relevant use situ- ation is based on proximity to this ideal point. In consumer psychology this is equivalent to an assumption that individuals have strong and stable preferences.
The competitive positioning approach uses consumer judgements, normally on an aggregate basis, on the similarities and differences between
specific competitive offerings. In principle this provides an analytical output roughly equivalent to the spatial distribution in the Hotelling model. Such an analysis can also be used to provide an estimate of the dimensionality of the discriminant space, but in many situations for ease of presenta- tion the results are presented in a constrained 2D format. Equally benefit segmentation studies can be used along with techniques such as factor analysis to try and arrive at an estimate of the dimensionality of the demand side.
We can be reasonably certain that the atti- tude space for customers in any particular market is generally, say, N 3: factor analytical studies might suggest at least four or five in general and that of competitive offerings is of at least a similar order. Indeed, in the later case if we considered the RBV of the firm very seriously we might go for a dimensionality as high as the number of competitors.23
Of more interest from a strategy point-of- view is how we represent what happens in terms of actual purchase behaviour in a competitive mar- ket through time. Although there is relatively little high-quality empirical and indeed theoretical work in this area, so far there are intriguing results to suggest that the dimensionality of the market space for this purpose can be much reduced, although we may still then have problems with some second order effects in terms of market evo- lution. There have been a number of attempts to apply segmentation analysis to behavioural data with much less information as to attitudes or intention. In one of the more detailed of such stud- ies, Chintagunta (1994) suggested that the dimen- sionality of the revealed competitive space was two dimensional but even this might be really an over-estimate.24 In his own interpretation of the
results he focuses on the degree to which the data analysis reveals interesting differences in terms of brand position revealed by individual purchase patterns through time.
In terms of second order anomalies, we can also consider some of the issues raised by so-called ‘compromise effect’ in choice situations where the choice between two alternatives depends on other, less attractive, alternatives. In an intriguing paper, Wernerfeld (1995b) argues that this effect can be systematically explained by the notion that consumers draw inferences about their own personal valuations from the portfolio of offer- ings. However, it may be that a compromise effect could also be seen as the result of mapping an
N 1 attribute and preference space on to an N 1 set of purchase decisions.25
A simple model of spatial competition might therefore be one in which a considerable amount of competition can be seen as along a single dimension, in circumstances in which multiple offerings are possible, and where there is no reason to believe a priori that individual offerings will be grouped either by common brand or specification, with a fixed entry cost for each item and a distribu- tion of demand which is multi-modal. To this extent it may actually be true that the very simpli- fications that many criticize in the Porter ‘ three generic strategies’ approach may be reasonably appropriate in building a first order model of com- petitive market evolution (see Campbell-Hunt, 2000). In the short-run, following the notion of ‘clout’ and ‘vulnerability’ (Cooper and Nakanishi, 1988), we might also expect changes in position in this competitive dimension could be a function of a whole range of what might often be seen as tac- tical as well as strategic marketing actions.
Cooper extended his own approach to understanding market structures by marrying two different data types – switching probabilities and attribute ratings (Cooper and Inoue, 1996). Despite the fact that the models developed appear to perform well against the appropriate statistical test, there remain basic issues which link to the issue of the time dynamic evolution of the market or demand space. When the model is applied to the well-established dataset on car purchase switching behaviour (Harshman et al., 1982), it is clear that it provides an interesting and informa- tive analysis of the ways in which various cus- tomer ‘segments’ have evolved over time both in terms of their size and attribute preferences. However, given the nature of the data and the form of analysis the dynamic process whereby customer desires change in response to both new
The basics of marketing strategy 65
23Since the RBV approach focuses particularly on the
nature of idiosyncratic and difficult to imitate resources in each individual firm. For a thorough presentation and analysis of the RBV approach see Barney and Clark (2007).
24In fact, on closer inspection, it is clear that we can achieve
a high level of discrimination with the one-dimensional map where there are two distinct groupings, and one intermediate brand and one ‘outlier’ brand. It is significant that these groupings are not either brand or pack-sized based but a mix- ture. In fact, the only result in moving from the one-dimen- sional to the two-dimensional analysis, is that one brand has become less discriminated (see Wensley, 1997). Hence it would appear that we can rather surprisingly reduce the effective competitive space to a single dimension with the possibility of only some second order anomalies.
25The classical Victorian monograph ‘Flatland’ (Abbott,
1992) provided an early illustration of many perceptual prob- lems of moving between spaces of different dimensions.
competitive offerings and other endogenous and exogenous factors can only be seen it terms of changes in attributes and specific switching deci- sions. We must now consider, however, particu- larly in the context of understanding the time- based nature of market strategies, how we might incorporate in more detail a longer-term time dimension with a stronger customer focus.