Travel Reference
In-Depth Information
Data-driven segmentation step #4: Description of segments
The description stage is identical to that in the commonsense approach: segments are compared
to each other with respect to other relevant personal characteristics, which enables management
to get a full picture of the segments which they then select one or more target segments from. A
common mistake made in tourism segmentation research, is to conduct an analysis of variance
using the segmentation base and then arguing that signifi cant differences between segments in
the segmentation base provide evidence of the fact that the segments are distinctly different. This
approach is acceptable for any variables
except
the segmentation base for the following reason:
any algorithm that is used to group respondents using the segmentation base, does this in a way
which maximizes the differences between segments. A signifi cance test determines whether
relationships observed between variables are random. Clearly, after running an algorithm, the aim
of which is to achieve maximum difference between segments, the relationship of variables
between segments is no longer random and cannot be tested. Rather, it is the default expectation
that resulting segments differ signifi cantly in the segmentation base. Tests of differences are,
however, critical for other variables, those not included in the segmentation base, like other travel
behaviours, beliefs or socio-demographics.
Conclusion
Market segmentation is a central part of marketing strategy for businesses, organizations and
tourism destinations. Market segmentation studies are not only popular in academic publications
where they are conducted for the purpose of knowledge development, they are also very popular
and commonly used in the tourism industry with the aim of gaining market intelligence to
ensure future organizational success.
A comparison of current state of the art methods in both the conceptualisation of market
segmentation as well as the methods used to conduct segmentation analysis indicates that there
indeed is a signifi cant gap in both the understanding of what market segmentation can and
cannot achieve, as well as the methods used to actually conduct segmentation studies.
To close this gap it is important for both data analysts and managers involved in segmentation
analyses to understand that, conceptually, market segmentation is an exploratory exercise and
that one single computation is nothing more than one random grouping of many possible
alternatives, and most likely not the best one of them. Once the notion of the exploratory nature
of segmentation is accepted, it is clear that an assessment is needed of the nature of the
segmentation study: is the aim to reveal true segments; identify reproducible segments; or
construct artifi cial segments? This can be achieved by repeating computations a number of times
with different numbers of clusters to determine the level of stability of results across replications.
Low levels of stability puts pressure on decision makers because it is entirely up to them which
of many possible segmentation solutions to choose.
A second key success factor in market segmentation studies goes beyond what is traditionally
understood as segmentation analysis and includes critical steps before and after the actual analysis.
These steps include: a thorough assessment of the constructs which will form the segmenta-
tion base; careful questionnaire design in order to capture the construct under study validly
and avoid data analysis problems further down the marketing decision process; and use of recent
data (collected from a sample that is suitable to the research question at hand, rather than a
convenience sample).
Reports describing segmentation studies should disclose the full details of how the study
was conducted and how all critical conceptual and methodological decisions have been made.
