Marketing Data Analysis
Rian BEISE-ZEE
From Marketing Metrics, 2nd Ed. by Paul W. Farris, et. al. (ISBN: 0321750403) Copyright © 2012 Pearson Education, Inc. All rights reserved. Perceptual Mapping Positioning
… is establishing a specific customer perception about an existing product or brand.
Positioning
• Happens in the mind of the customers. • We can position the same product differently. • Defined relative to competitors, therefore can keep a product/service “different” from competitors products in the mind of the customer even if it is physically the same Positioning Principles Same Product - Different Perception Mouthwash Perceptual Mapping (or: Multidimensional Scaling)
• We can think of a variety of product attributes that possibly differentiate a product; however, consumers are unlikely to keep many attributes in mind. • We can position a brand, but we don’t know if consumers are really “buying it”. 1. Perceptual mapping is a tool to measure what dimensions consumers use to differentiate products. 2. For marketers two or three dimensions allow a graphic representation, which makes strategy building a lot easier (but it does not mean that the world is really that simple). Perceptual Map
The outcome of a multidimensional scale is called ‘perceptual map’, a visual representation which helps to derive strategic conclusions about:
• What main dimensions are used by consumers to compare Brands? (but not its importance!) • Who are our competitors and how similar are they to our Brand? • What product clusters do consumers perceive? • Are there holes in the market that are not yet covered? Example: Detergent
Methods
• There are several methods – Attribute-based: we start with many attributes and try to reduce them to 2 or 3 main dimensions (by means of factor analysis or discriminant analysis) – Comparison-based: We start with comparisons between two brands and end up with interpreting a distance field Attribute based MDS
Rate each beer brand‘s association Not descriptive Totally at all descriptive 1 2 3 4 5 Heavy Action Good Value Budget Relaxation Popular with women Sporty Popular with men Premium Special Occasion Sex Appeal From Moore and Pessemier (1993). Comparison-based MDS
Beer Chang vs. Singha Beer
Totally Extremely similar dissimilar
------1 2 3 4 5 6 7 Expl: Similarity ratings of 4 Objects (for example products or companies)
A B C D
A
B 3.2
C 1.7 3.9
D 5.1 3.3 4.7 A Two Dimensional Solution of Empirical Distances Dimensionality
• How many dimensions a perceptual map has depends on the complexity of the similarity data. Often, two dimensions are sufficient. More than three are difficult to interpret. But one dimension is also quite possible. • Two dimensions are often selected because it looks better. • The correct method to decide on the number of dimensions is the stress value and the scree plot. Labeling the axes (= interpreting the dimensions)
• MDS only delivers the map but no clue about how to interpret the dimensions • Based on individual judgment of the researcher, e.g. by – known objective attributes – Based on additional questions on the Brands’ attributes • Based on objective vector algorithms with additional attribute information Interpreting the Dimensions: Soft Drinks Example Perceptual Map for Beer in Thailand Interpreting the Dimensions Problems with MDS
• Different methods lead to different maps (use several and analyze the differences) • Different consumer groups have different perceptual maps (e.g. users/non-users) but MDS assumes homogeneity • Different users consider different attributes and attach different importance to attributes • Less than 4 objects per dimensions lead to misleading solutions but a large number of comparisons often overexert respondents. Correspondence Analysis
• Another method to derive a perceptual map based on the relationship between objects (brands, products) and nominal attributes (gender, age group, lifestyle) • A perceptual map can be derived from any cross-tabulation (frequency count) Expl. Cross-tabulation
Product Sales A B C Total Young adults 20 20 20 60 (>35)
Middle age (36-55) 40 10 40 90
Mature (55+) 20 10 40 70
Total 80 40 100 220 Process of Correspondence Analysis
• Expected cell count under independence = (column total/total * Row total/total)*total • Chi-quare value for each cell = actual cell count – expected cell count expected cell count • Chi-sqare is used as a measure of association (similarity) between row and column : – The greater chi-square the higher the similarity – The more negative the higher the dissimilarity
Derived Perceptual Map The Ford Ka Which of these cars is more similar to the Ford Ka? VW Polo VW Polo Ad VW Polo Ad Renault Twingo Ads Toyota Rav4 Fiat 500 (Cinquecento) Opel Tigra One Dimension Solution of all Respondents
Object Points Common Space
FIAT500 MICRA POLOP106CORSAFIESTA TWINGO RAV4 KA TIGRA
-1.0 -.5 0.0 .5 1.0 1.5
Dimension 1 Two dimension solution of all respondents
Object Points Common Space
.6 KA FIAT500 .4 TWINGO MICRA .2 CORSA
-.0 TIGRA FIESTA -.2 P106 RAV4 -.4 POLO
Dimension 2 -.6 -1.0 -.5 0.0 .5 1.0 1.5
Dimension 1 Solution for Respondents with Ka Preference
Object Points Common Space
.6 POLO RAV4 .4 FIESTA .2 P106 TIGRA
0.0 CORSA MICRA -.2 KA FIAT500 TWINGO
Dimension 2 Dimension -.4 -1.0 -.5 0.0 .5 1.0 1.5
Dimension 1 Solution for Respondents with low Ka-liking
Object Points Common Space
.8 KA .6
FIAT500 .4 MICRA
.2 FIESTA -.0 TWINGO CORSA TIGRA RAV4 -.2 P106 -.4 POLO
Dimension 2 -.6 -1.0 -.5 0.0 .5 1.0 1.5
Dimension 1