Over the last few years there have been tremendous advances in the world of conjoint analysis in terms of software, mathematics, topics considered, and approaches. These advances create exciting new opportunities for conjoint analysis over the internet. This paper seeks to set a context for conjoint analysis by reviewing how we got to where we are now and then to explore the new possibilities that are opening up because of the web. Finally, the paper will try to draw the threads together by raising questions about where these techniques will take us next.
In this paper three topical ways of segmenting retail markets are discussed illustrating the principal advantages and unique insights afforded by each of them individually. By use of a matrix approach it will then be shown how they can be combined into a single easy-to-use tool that retains all the qualities of the measures individually, and adds a substantial depth of perspective. The first approach is the segmentation of markets by Lifestage or age group, and this paper will show how ones lifespan can be divided into 5 year periods. In each of these periods peoples circumstances change, their interests and activities change, their buying behaviour and levels of consumer expenditure change. The second approach is the segmentation of markets by Lifeplane or socio-economic group, with Education being the key. This paper will show how buying behaviour, store choice and consumer expenditure levels are a function of Life Plane. The third approach is the use of Shopper Typologies to segment retail markets and this paper will show how different types of people react to retail and media offerings. In concluding the paper, it will be shown how a matrix approach may be used to combine all three measures.
Factor Analysis (like the similar technique of Principal Components Analysis) is a multivariate statistical method used primarily for data reduction. Itis usually employed to reduce the columns of our data matrix, that is the variables measured for each respondent. The reduction Is done by grouping together those variables which are intercorrelated as measured by the coefficient of correlation.
In our data matrix we have one variable of particular interest and we wish to see how it is affected by movements in a single explanatory variable. For example, the variable we might be interested in is Sales in £mn and we may wish to know how it is affected by advertising expenditure in £000's. The rows of our data matrix may consist of the last seven years of data for these two variables.
Clustering normally operates on the rows of the data matrix. It seeks to find groupings or clusters of respondents who exhibit similar patterns in terms of the variables measured.
This paper focuses on customer relationships in retail banking. Understanding customer relationships and especially the profitability of them requires the development of new concepts and tools. The paper discusses some concepts and tools available to support profitability analysis on a customer base level and on a relationship level. A tool in this context is the relationship configuration matrix, which enables us to investigate distinct relationships and segments of the customer base and develop products and pricing mechanisms in order to fully utilize the business potential of the particular customer or segment. The tools are illustrated by a case. Swedbank, the largest bank in Scandinavia, has carried out a comprehensive customer base analysis in which MIS data about 97.000 customers and market research data about 3226 customers was used. The aim was to calculate customer relationship profitability and develop strategies for ensuring the loyalty of profitable customers and enhancing profitability of the large number of unprofitable customers. Profitability was found to be a function of customers' interaction behavior and thus a pricing strategy was launched. The successful strategy was the first comprehensive and congruent pricing strategy in the Swedish market and has created a fair amount of debate.
In satisfaction surveys amongst initial buyers of small electrical appliances, it is essential, beyond the easy 'disaster check', to obtain: a comparison of each new product with the accumulated database of similar results from previous launches, an analysis of the sources of satisfaction or disappointment, linking the overall attitude towards the new product with the respective levels of satisfaction induced by the various features of the appliance. Problems with traditional overall satisfaction scales appear when these indicators lack sensitivity and show an unwanted concentration of answers on one point of the scale. What is needed is a tool showing real differences across products and giving small, manageable groups of buyers, which can be ranked in order of their level of satisfaction. This paper describes how new satisfaction questions and a matrix, both designed to produce a more sensitive classification, from most to least satisfied buyers, helped to solve this problem with a better and more valid measurement of satisfaction. Part 1 describes the business background of the satisfaction surveys within a manufacturer of small appliances, with the marketing objectives behind these surveys. Part 2 shows the new approach used to bring a better answer to these needs, with two new, longer scales and their combination to form a 77-cell satisfaction matrix, out of which five categories 'A/B/C/D/E' were drawn. Part 3 illustrates the potential of the new improved method, explaining how it was validated before implementation in all satisfaction surveys. Part 4 is a discussion of the stability of the new ABODE scale across products, comparatively with five other satisfaction scales.
This paper proposes a new technique for measuring consumer's attitudes which might have fuzzy characteristics. In most of the traditional market segmentation, multidimensional configurations are initially divided into several discrete clusters and then those clusters are predicted using socio-demographic and life-style variables. More specifically, scores obtained by factor analysis or correspondence analysis are usually analyzed by cluster analysis, and then analyzed by multiple discriminant analysis. However, the traditional approach might be inappropriate since the market structure consists of fuzzy sets which was proposed by Zadeh (1965). Therefore, the authors applied fuzzy discriminant analysis (FDA) to the survey data on alcohol drinking attitudes. In FDA, the likelihood of subjects belonging to each cluster is expressed by a membership function, which can describe the relationship between the respondents and the fuzzy groups uncrisply. Secondary, the fuzzy multivariate analysis can be used as a weighting adjuster for parameters to be estimated when the sample distribution is biased from population distribution. The elements of the criterion data matrix are dummy variables in the case of discriminant analysis, weight-back coefficients in the case of weight- back, and membership values in the case of FDA. Although these three analytical methods are mathematically equivalent, FDA is the most generalized model and the other two methods are its special cases. The authors illustrate that discriminant coefficients can be estimated accurately by weight-back, and some research implications of weighting coefficients are also discussed. Rationale and mathematical derivation of membership function and weighting on respondents are described in the appendixes.
This paper divides into two main sections: the first section deals with background details to the Country Image study including the reasons why it was conducted, how it was designed, and the specific relevance of the study to the travel and tourism seminar. The second section deals with our interrogation of the survey data, focussing on the application of multivariate analyses; in particular the "Strategic Improvement Matrix". The Country Image study was carried out in 1990 by the INRA network of research companies across Europe and the United States. INRA is a federal or voluntary network of market research companies covering most of the globe. Each research company in each country is independently owned, allowing for separate national development while also benefitting from the membership of an international organisation. Within the European part of the network there is a central co-ordination office based in Brussels.
The topic of my paper, PC-based research in years to come, invites dreaming: talking about research based on the Personal Computer makes it easy to enter the no man's land between dream and reality, that land where cost per interview declines with the speed of a falling star and where a kiss on the computer earns you the optimal solutions. I will try not to do so. Instead, I will first highlight some actual differences between Europe and the USA. Next, I want to discuss the production process of market research and the product range that market researchers (within organisations or as research institutes) are offering. The production process / product range matrix (PP/PM matrix) offers a good base to explain the function and the opportunities of the computer and especially the PC in market research. The PP/PM matrix also offers a good base to show the main developments, threats and opportunities in the market research industry. Finally, I will combine the various developments and will try to draw a picture of PC-based research in the United States of Europe after 1992.
Quest International is an Anglo-Dutch Company and one of the largest suppliers of fragrances to the worldwide consumer goods industry, having been formed recently through the merger of PPF International and Naarden International. PPF has been working with RBL-Research International UK for a number of years on worldwide market and product mapping studies which illustrate how a standardised but highly diagnostic data collection and analysis procedure can be applied to study international brands in any market of the world. Few businesses are more internationally-based than the supply of fragrances to the world's fmcg companies. Although Quest is one step removed from the ultimate consumer, it is only by understanding the latter's relationship to brands that they keep the business of their own clients. Clearly Quest cannot match each of their client's worldwide research resources and budgets. Instead Quest must use its market research budget with considerable efficiency and skill in order to remain in touch with the worldwide consumer. Brand mapping enables them to do this.
The paper shows how different versions of factor analysis applied to the same data can lead to different results. In contrast, factor analyses of data which differ can lead to the same results. These are two major limitations of the technique. Two alternative analysis procedures are described. Firstly, any clusters in the data can readily be seen from the correlation matrix if the correlations are rounded to one or two digits and the variables are ordered according to the average size of their correlations. Secondly, since correlations only reflect the strength of a relationship and not its nature, a further alternative is to establish the actual relationships. This shows up the same kinds of clustering as factor or correlational analysis do at their best, but also adds real hut simple quantification. It leads to a depth of understanding and relative ease of communication which seem impressive.