Marketing Analytics MBusA MBS 2023 Practice Questions Solution ©Nico Neumann

Practice questions class 3 - Solution

Question A

a) This outcome depends on the students’ choice. We need to trade off the external validity
(how realistic our scenario is) and internal validity (the design we can choose).
b) We choose library(DoE.base) in R to find possible designs. It’s possible that we must revisit
part a) if no orthogonal array exists for our originally planned number of attributes and
attribute levels.

on oint study design trade o s

tep
umber of
a ributes
teps
and levels
xperimental
design
sample si e

xternal Internal
validity validity

Question B

We need to use the R code and Excel calculations from the lecture and just change the input.

a) Relative
Attribute Importance
Drink 40.9%
Candy 4.8%
Spread 11.4%
Salty snack 2.0%
Price 40.9%

b) WTP $ per util (based on average): $0.67

c) Value/ Marginal Willingness-to-pay)

Value of change
Change (MWTP)
Toblerone vs. MMs $0.23
Chips vs. pretzels $0.10
Peanut butter vs Nutella $0.56
Honey vs Nutella $0.44
Peanut butter vs Honey $0.11
Apple juice vs. fanta $2.00
Apple juice vs. coke $0.33
Coke vs fanta $1.67
 Marketing Analytics MBusA MBS 2023 Practice Questions Solution ©Nico Neumann

Question C- a) and b)

Product Logit rule Share-based

1 29.0% 32.5%
2 24.1% 30.9%
3 46.9% 36.6%

Question D

We need to apply matrix multiplication (beer utils times beermarket matrices) to obtain the total
utility for each product and individual. Then can apply the max utility rule and calculate the
market shares:

Heineken: 13.3%
Becks: 9.7%
Guinness: 14.8%
Corona: 33.2%
Amstel: 9.7%
Sapporo: 19.3%

Question E

a) We use Kmeans and the WSS/ elbow method (manual plot or library cluster function) for
various number of clusters (e.g. k=10). Command from R package cluster:
fviz_nbclust(DATAFRAMENAME, kmeans, method = "wss", k.max = 10)

WSS = Total within-cluster sum of square, which we minimise. Manual plot on left side/
plot from library cluster on right hand side (parts of y axis are removed in that graph,
which is why the slopes appear different). The change in the slope is unclear – one could
argue this occurs at 2, 3, or 4 clusters (where the strongest bend/ the knee is). We
proceed with 3 clusters for the subsequent analyses (this is just for illustrative purposes).

Note: For Kmeans cluster analysis, the provided values must have the same units and
ideally similar variances. We therefore often standardise the input matrix first. However,
no need to do this for our utils data.
 Marketing Analytics MBusA MBS 2023 Practice Questions Solution ©Nico Neumann

b) To find out which segment (cluster) to choose, we should estimate for which
segment we obtain the highest market share if we develop a product for the
respective segment. The steps are the following:
➢ Examine means of each segment to examine which attributes are valued the
most.
➢ Develop product for each segment based on most appreciated attribute levels
(those with the highest average per segment).
➢ Simulate market share with the hypothetical product (tip: add the new
product in the same column format as the existing products, see table below).
c) Repeat analysis of b) but use market averages for utilities (of all people/ sample).

Don’t forget to state assumption Only market share matters (as profit/ costs are
linear and equal).

Attribute Heineke Becks Guiness Corona Amstel Sapporo OurBeerC1 OurBeerC2 OurBeerC3
n
European 1 1 1 0 1 0 0 1 1
Mexican 0 0 0 1 0 0 1 0 0
Japanese 0 0 0 0 0 1 0 0 0
Price6.19 0 0 1 0 1 0 0 1 1
Price5.49 1 1 0 1 0 0 0 0 0
Price4.79 0 0 0 0 0 1 1 0 0
RichFullBodied 1 1 1 0 0 0 0 0 1
Regular 0 0 0 0 1 0 0 0 0
Crisp 0 0 0 1 0 1 1 1 0
Strong 0 1 1 0 0 0 0 0 0
Mild 1 0 0 1 0 1 1 1 1
VeryMild 0 0 0 0 1 0 0 0 0
CaloriesFull 0 0 1 0 0 0 0 1 0
CaloriesRegular 1 1 0 1 0 1 0 0 0
CaloriesLow 0 0 0 0 1 0 1 0 1
Six12OzLarge 0 1 0 1 0 0 0 1 1
Six12OzSmall 1 0 1 0 1 0 0 0 0
Four16Oz 0 0 0 0 0 1 1 0 0
GreenBottle 1 1 0 0 0 0 0 0 0
BrownBottle 0 0 1 0 1 1 0 1 0
WhiteBottle 0 0 0 1 0 0 1 0 1

Market share simulation – clusters versus mass market:

Market shares (%)

Cluster41 Cluster43 Cluster15 Mass market
Cluster size: 41 43.8 15.1 100
Product share:
Heineken 12.5 7.9 7.5 6.1
Becks 8.2 7 3.6 3.7
Guiness 14 5.8 6.9 8
Corona 23.4 21.6 20.8 17.5
Amstel 7 5.8 4.2 4.6
Sapporo 14.6 10.9 12.3 12.6
Our new beer 20.4 41 44.6 47.5

When we must choose a single product only, we achieve the greatest marketing share by
producing one product for the mass market (given the conditions of equal profit).