Case Study

CASE STUDY
Presented by Nishant Bharti

ABOUT CASE
STUDY
A retailer of eyewear products (contact lens &
eye-glasses) has been running different
campaigns for its customers over time and have
results from that given in the campaign
performance report. They have run these
campaigns on existing customers using customer
level campaign response probability scores. The
CRM Director, looking at the report wants to
make some changes on it.
. . 2
T TEST FOR ANALYZING THE IMPACT
ON CUSTOMER RESPONSE
1) According to the case study and campaigning report first we have to know
whether the campaign was successful or not which we can clearly notice that
the average of targeted group response is more than the control group
response but in order to prove it statistically I have done T test which will show
the benefit of campaign on targeted group.
2) In the T test we assume null hypothesis (H0) is that there is no difference

between the means of both the group , whereas Alternate hypothesis (Ha) is
that there is a significant difference between the means of both the groups.
No of observation = 1181
3) Degrees of freedom = 1180
95% confidence level 0.05 alpha
4) After taking the necessary assumption and calculation I had calculated the p
value in excel by using this formula T.TEST(Q2:Q1181,P2:P1181,2,1) from
where i get p value 2.12431E-73 less than 0.05 i.e alpha.
5) So, from this T test we can say that since here P value is less than 0.05 so we
reject the null hypothesis that there is no differences between the averages
and accept the alternate hypothesis which state that there is a significant
difference between the means of both the groups.
. . 3
T TEST FOR ANALYZING THE IMPACT
ON CUSTOMER RESPONSE
According to the case study Targeted group are the group on which the
experiment was done and the control group are out of experimentation so after
the T test we find out that campaign was successful as the response in targeted
group are giving better result, performance and average than the control group.
So, yes there is an impact exists and I show this with T test in excel.
. . 4
CORRELATION TO SHOW IMPACT.
INTERPRETATION –
This is the heatmap which I have done in Python programming

language in order to show the correlation between the numerical
variable in the provided dataset.
So here you can see that Response_count and

Response_Sales_AMT are highly correlated as value is 0.99.
So from here also we can show the impacts as if the response

count increases the total sales value also get increased.
. . 5
REGRESSION ANALYSIS TO KNOW
WHICH CUSTOMER IS BEST
In order to answer the second question I have done a regression analysis in R programming which will
provide me with coefficient of each attribute. So more the coefficient more preference is given to that
attribute while making the budget.
According to the analysis it has been noticed that Product 1(prdt-1) has more coefficient, Brand C has
more coefficient followed by brand E, D, B. Milestones 12month has more coefficient followed by
Milestones 15month , Milestones 24month, Milestones 18month , Milestones 21month , Milestones
27month.
In the same way I have shown in the excel that which customer segment I have given more preference
as per their coefficient so with more coefficient customer segments are best customer for campaigning .
In the next slide I am going to show the budget for the next year where I am spending 75% of 2MM
USD to those customer segment, brand, milestone and product type who are having highest coefficient
and the rest 25% with lesser coefficient.
In this way I am going to target the best prospective customer segments.
. . 6
PROSPECTIVE CUSTOMER SEGMENT, BRAND,
MILESTONES AND PRODUCT TYPE WITH COEFFICIENT.
INTERPRETATION -
F. rom this picture I wanted to show the coefficient of each variables and
why I am giving more weightage to this attribute.
.
.
I. NTERPRETATION -
From this customer segment picture I want to show that this are the best
customer segments because they have the highest coefficient. There are more
customer segment but due to page limit I am unable to show that but can see that
from the attached excel sheet.
. . 7
BUDGET ALLOCATION FOR NEXT
YEAR CAMPAIGN
. . 8
BUDGET ALLOCATION FOR NEXT
YEAR CAMPAIGN
INTERPRETATION –
In the above slide I have shown the budget allocation which I have done in excel.
According to this customer segment , brands, milestones and product type having
greater coefficient is given much priority.
So Till 14th preference order I have assumed to spend 75% of 2MM in this customer
segment. The amount is 1.5 M USD.
And From range 15 I will spend the rest amount of 0.5 M USD.
So in this way I am going to send 8 lakh emails to prospective clients with highest
coefficient and make our campaign successful.
. . 9

Case Study

Uploaded by

Copyright:

Available Formats

Case Study

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Case Study

Uploaded by

Copyright:

Available Formats

CASE STUDY

Presented by Nishant Bharti

2) In the T test we assume null hypothesis (H0) is that there is no difference

This is the heatmap which I have done in Python programming

So here you can see that Response_count and

So from here also we can show the impacts as if the response

In this way I am going to target the best prospective customer segments.

You might also like