Using Growth Patterns to Segment Tourist Markets
Demand forecasting is critical in seasonal industries such as tourism, practitioners need to know what is driving demand and when to expect growth. The most common methods for predicting demand, at least in the tourism sector, focus on predicting demand growth levels. This can be very useful in identifying high growth markets, such as from emerging east Asian economies, and has shaped much of Australia’s tourism policy and strategy.
However, research is increasingly highlighting the importance of understanding the pattern of growth, not just the level of growth. Often the factors that contribute to a destination’s initial attractiveness to tourists (uniqueness, on-trend, exclusivity) are inherently unsustainable, resulting in a commonly experienced S-shaped exponential growth pattern, with demand tapering off as the market matures. These aggregate S-curves are critical to identifying the life cycle of a tourism destination. However, by focusing on the aggregate data across all tourists this approach hides much of the complexity of tourism behaviour.
A destination is usually visited by different people doing different things at different times, and so by splitting the market into smaller, more discreate markets, growth patterns can be used as a segmentation strategy to identify the life cycle of audiences rather than just destinations. This approach reflects the dynamic and complex needs of tourists who are increasingly dissatisfied with one-size-fits-all experiences. However, while a segmentation approach to predicting demand patterns can be critical in the development of diversification strategies and the timing of product development and promotion, the complexity of mathematics required by many existing methods has left this approach to demand forecasting underused by industry.
The paper “Segmenting Tourism Markets Based on Demand Growth Patterns” introduces a new approach to using growth patterns that is practitioner-friendly and uses algorithms available in commonly used statistical software such as SAS, SPSS, R and Stata. It then walks the reader through the application of this approach to identify five key segments of Australian’s inbound tourism markets.
Methodology
In general, demand growth profiles are characterised by level and pattern. In tourist markets, level refers to the number of arrivals while pattern would refer to changes over time in that level. For example, both the level of demand from both China and New Zealand are very high (large number of arrivals), but while New Zealand has consistently shown a high level of demand, the level of demand from China has increased exponentially since 2009. Therefore, we would say that China and New Zealand have high levels of demand, but New Zealand shows a mature pattern of growth while China’s demand is still developing.
The typical approaches to segmentation using a demand profile are Profile Analysis (identifying similarities in demand pattern), Cluster Analysis (identifying groups of individuals or countries based on similar average demand characteristics) and Factor Analysis (identifying factors that explain the variance shared between individuals or countries, such as level of income or shared languages).
This paper, however, proposes the use of Multidimensional Scaling, a class of statistical analysis that looks at similarities and differences in pairs of data points (such as specific tourist markets) across a range of defined variables. This approach focuses on similarities and differences in the pattern, not the level of change, and is capable of identifying both individual level differences and broader context or demographic factors that are impacting on demand patterns. Importantly to practitioners, this approach results in the visual mapping of these similarities and differences in multidimensional space onto a 2-dimensional graph, greatly increasing how approachable the output of the analysis is to decision-makers.
Growth Patterns in Australia’s Inbound Markets
To demonstrate how Multidimensional Scaling can be used to segment tourism markets, this paper applies this approach to Australia’s annual inbound short-term visitor arrivals from 1991-2016 using publicly available Australian Bureau of Statistics (ABS) data.
Across this paper, the process and interpretation of data is detailed step by step, from multidimensional scaling, assessing goodness of fit, the regression of growth patterns, and the clustering of inbound markets. The resulting five factor solution highlights different inbound markets in different stages of development.
The first segment, “development” is characterised by an increasing linear straight-line pattern of growth and reflects markets with modest but stable increases in demand such as New Zealand, United States and Germany. The second segment, “involvement and exploration” demonstrates exponentially increasing tourism demand from rapidly expanding markets such as Mainland China, India and Malaysia. The third segment, “consolidation and stagnation” shows a leveling off of demand, from mature markets no longer in growth such as United Kingdom and South Korea. The fourth segment, “rejuvenation” shows an inverse s-curve in demand, with markets such as Singapore, Hong Kong and Taiwan rising again after a period of stagnation. The fifth segment, “decline” reflects markets in retreat, such as Japan, that are experiencing rapid collapse in demand.
Importantly this approach can also show associations between various factors captured in ABS data and the observed patterns in demand. This paper also demonstrates how to produce scatter plot diagrams to visually confirm clusters of these variables that best capture the drivers of demand. In particular, countries with lower growth rates of income and immigration are more likely to show linear straight line demand curves. However, high rates of income and immigration lead to exponential increases in demand.
Conclusion
While the segments revealed by this paper largely reflect an existing understanding of the lifecycles of tourist markets, the demonstration of how to this new approach to data driven segmentation strategies has much broader applications. Depending on the data available, this approach can be applied to individuals as well as countries, and incorporate psychographic variables in addition to demographic ones available in ABS data.
This type of analysis can be achieved through general-purposes statistical packages already in use by industry on simply structured and publicly accessible data, with results that are straightforward to interpret with visual illustration.
It is this second point that is perhaps the most important contribution of this paper to industry, as nearly 70% of marketing managers have concerns about interpreting data-driven segmentation analysis. Often the complexity of the mathematics behind these models leaves managers feeling it is a “black box”, with decision making using these models feeling like a question of faith. This approach produces graphs that can be interpreted visually, closing the theory-practice divide and helping to facilitate the transformation of research findings into managerial decisions. Data driven decision making is vital to successful outcomes across retail industries and making that data approachable by management is key to bringing the experience and intuition of managers back into decision making.
Researcher
More information
The research article is also available on eprints.