Leni Théry Customer Relationship Management
The Gabriel Hansen Foundation
Individual Assignment 1 : Customer Valuation and Prediction
Question 1
The signs show whether there is a positive or negative effect on each parameter. A positive
coefficient indicates that when the value of the independent variable increases, the dependent
variable also increases.
For the question 2 & 3, we take the results from the out-of-sample prediction on the spreadsheet
“Results (Q1)” and for the question 4 & 5, the out-of-sample prediction on the spreadsheet “Results
(Q4)”
Question 2
In case the Gabriel Hansen Foundation does not change its direct marketing strategy at all, we can
calculate the average of the donations predicted by the observations. Then we have a marginal cost
of a solicitation for the March campaign at 1.10 € and a fixed cost of the campaign of 7500 €.
The profitability calculation is the following :
Average Donation 5.985 €
Observation group 2 293
Donors/Observation group 200
Total Revenue 2 744 782.87 €
Marginal Cost 1.10 € 504 460.00 €
Gross Margin 2 240 322.87 €
Marketing Cost 7 500.00 €
Net Revenue 2 232 822.87 €
Question 3
In case the Gabriel Hansen Foundation solicits all its active donors, but only those expected to be
profitable (donation above the marginal cost : 1.10€). So in our calculation, we only take into account
predicted donation greater or equal to 1.10€.
To calculate the profitability, we have the following :
Average Donation 12.071 €
Observation group 1 111
Donors/Observation group 200
Total Revenue 2 682 218.17 €
Marginal Cost 1.10 € 244 420.00 €
Gross Margin 2 437 798.17 €
Marketing Cost 7 500.00 €
Net Revenue 2 430 298.17 €
2 430 298.17 € – 2232 822.87 €
=8.84 %
2 232822.87 €
We can conclude here that it is more profitable if we solicit group of observation who donate more
than 1.10€. Indeed, there is an increase of 197 475.30 € (8.84% relative) on the net revenue.
1
Leni Théry Customer Relationship Management
Question 4
We now take into account the cold donors in the calculation. The profitability of cold donors
solicitation is the following :
Average Donation 0.072 €
Observation group 1 023
Donors/Observation group 200
Total Revenue 14 773.82 €
Marginal Cost 1.10 € 225 060.00 €
Gross Margin (Cold Donors) – 210 286.18 €
Gross Margin (Active Donors)(Q3) 2 437 798.17 €
Marketing Cost 7 500.00 €
Net Revenue 2 220 011.99 €
We can see that soliciting all cold donors is not profitable. Soliciting all of them would just generate a
210 286.18 € loss and decrease the net revenue to 2 220 011.99 € (2 430 298.17 € without cold
donors). It is better for Gabriel Hansen Foundation to establish an adequate strategy for example
selecting only cold donors who are profitable.
Question 5
Now, we are going to see the effect of calling only donors who are expected to be profitable
(donation above marginal cost : 1.10 €). We can see that only 1 cold donors respects that criteria.
We can make the following calculation :
Donation 65.369 €
Donors 200
Total Revenue 13 073.79 €
Marginal Cost 1.10 € 220.00 €
Gross Margin (Cold Donors) 12 853.79 €
Gross Margin (Active Donors)(Q3) 2 437 798.17 €
Marketing Cost 7 500.00 €
Net Revenue 2 443 151.96 €
12 853.79 €
=0.53 %
2 4 43 151 . 96 €
We can see that it is still profitable if we solicit that only cold donor. The net revenue increase is 12
853.79 €, i.e. a relative increase of 0.53%. However, the increase is not notable given that it is “risky”
to solicit cold-donors (we have to take into account the probability that this profitable observation
group donate). Indeed, there is only one group of observation over 1023 who respects the criteria.
2
Leni Théry Customer Relationship Management
Question 6
In order to improve scoring model, it is necessary to implement the IPO process (Input, Process,
Output). That is, the RFM, which leads to segmentation, giving segments of consumers. Then the
prediction which would classify and finally the recommendation in order to use adequate marketing
actions for each segment.
Data Mining Using RFM
Analysis, Birant D,
Chapter 6 in Knowledge-
Oriented Applications in
Data Mining (2011)
Concerning segmentation, demographic data (age, gender, city, education, preferences, etc.) would
be required. Then we can classify them according to the RFM criteria. To go further, and define the
criteria that are more relevant than others, we can do a WRFM. Every variable of the model have an
assigned weight which can help for clustering analysis and decision-making. Thus, we can make a real
prediction on the behavior of each segment and know their intentions to make a donation. In
extension, we can add recommendations concerning these segments. For example, if a certain
segment gives a certain amount, then the marketing campaign should be focused on an appropriate
action.
In addition, we can add the Markov-Chain Model, in order to know the evolution of consumer
behavior, indeed we can know the probability that such a consumer group will change group and
become for example a good and valuable donor.
Finally, it would be interesting to establish a profit function where we can maximize it. We can use
logistic regression for this. We must first calculate the probability that an observation group will
make a donation. Then we have to define a profit function and solve it.
Question 7
The case of Gabriel Hansen Foundation is special because it is based on donations and not a purchase
of services or products. Marketing campaigns in this case should focus more on its message in order
to influence consumers to make a donation, similar to blood or organ donation.
It is therefore important to know how to differentiate the segment groups and to have a complete
understanding of the consumer and how they react to a marketing action. For example, we may have
a loyal consumer segment and making donations regularly (R↑F↑M↑). On the other hand, we have
a donor who gives a lot of money but at an irregular frequency (R↑F↓M↑), often correlated with
an event which pushes the consumer to make a donation. We have to take into account these
different behaviors and understand which are the most profitable in the long term.
The marketing message should therefore be positioned on an incentive to donate regularly than only
once. This requires playing with psychology to ensure that consumer behavior is shaped. Only people
convinced of the message and the usefulness will make a donation.
However, by having previous data, it is possible to refine the actions and understand the types of
consumers interested in the donation, but also their behavior when faced with a marketing action.
Thus, the manager will be able to establish adequate strategies in order to switch an irregular donor
to a regular one and get a high RFM score.
