Scribd
Upload
360 views13 pages

Marketing Insights from Discriminant Analysis

The document discusses how WSES, a software services company, analyzed over 4 years of sales data using discriminant analysis to develop a model that predicts deal wins and losses. The analysis found that relative strength in the customer's segment, as well as the customer's profit and profit percentage, were the most important factors for differentiating between won and lost deals. Based on this analysis, recommendations were made to Jack, the CEO of WSES, to reduce marketing budgets for certain products.

Uploaded by

Rini Rafi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
360 views13 pages

Marketing Insights from Discriminant Analysis

The document discusses how WSES, a software services company, analyzed over 4 years of sales data using discriminant analysis to develop a model that predicts deal wins and losses. The analysis found that relative strength in the customer's segment, as well as the customer's profit and profit percentage, were the most important factors for differentiating between won and lost deals. Based on this analysis, recommendations were made to Jack, the CEO of WSES, to reduce marketing budgets for certain products.

Uploaded by

Rini Rafi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 13

Marketing Head’s Condurum

Group 7 :
Madan
Senthil
Rini
Syed
WSES is a software services company

 Characters :
 Jack - the CEO
 Ben - Vice President-Marketing
 Mark – Young Data analysist holding phd in statistics.
Products and Services at WSES inc

 Market leaders in there segment


 GTMsys
 Finsys
 Procsys
 Lifesys

 Learnsys
 Contactsys
 Logisys
Being market leader what worries Jack

 Competition
 Which brings margins down
 Which requires reduction in production
cost.
Where ?

 6% of revenue which is $500mn goes into marketing.


 Which includes
 Travel costs, boarding, lodging costs, time spent etc.. By sales team
 Along with that customer wants to travel development centers before buying
the deal. Which add to the cost
 And the time spent by technical experts, support staff etc..
How?

 WSES has more than 4 years of data. But how to interpret the data.
 Here is where Mark comes into picture.
 Mark with the help of Ben developed a model to run the analysis to
understand the data.
Disriminant analysis

 Mark used discriminant analysis to analyze the data.


 Why not regression.
 Discriminant analysis is technique for analyzing data when the dependent
variable ( products like procysys, lifesys etc.., and probability to clinche deal
or lose it) is categorical and independent variable (revenue split for sales, on
a point scale to identify products sold etc..) are metric.
 Example : dependent variable ( PC brand, A, B and C) independent variable
( ratings of attributes of PCs on 7 point likert scale)

 Whereas for regression analysis requires both dependent and independent


variables as metric.
Disriminant analysis – Are variable values different
across groups?

Relative strength significantly differentiate


between the two groups

We can Observe, that the Relative strength in the segment, Profit


of the customer and profit % are clearly different between the
two groups.
Disriminant analysis – Are assumptions about
data satisfied?

Assumption of equal
The correlation numbers indicate no multicollinearity
variance-covariance is
not met
H0 = Equal Var -CoVar
H1 = Not Equal Var -CoVar
Disriminant analysis – How good is the model?

Wilks'λ is significant at 0.05 levels,


The square of Canonical indicating that the null hypothesis
correlation (0.791)2 = 0.625 (63%) which is that the variables ( like
is the variance in the dependent segment etc) have no impact on win
variable explained by this or loss is rejected.
discriminant
So, discriminant function has real
discriminating powers.
Disriminant analysis – How good is the model?

The model shows 97.3% correct


classification.

Considering we would have only 50%


correct without the model , we see
a 47% improvement in the classification
with Discriminant analysis.
Disriminant analysis – Interpretation

All measures seems to be indicating


Profit to the customer , Profit and Relative Strength in the segment in
that order contribute to the discrimination function
Ben’s suggestions to Jack

 After discussion with Mark Ben gave following suggestions to Jack


 For Procsys product pipeline reduce budget by 50%
 Expected value of sales to be 90% of sales opportunities.

You might also like