Identification of Hot Leads to focus more on
them and thus enhancing the conversion ratio
for X Education
X Education - Lead Scoring
Case Study
•Group Members:
•Lohit Labhala
•Harshal Lokhande
● X Education , An education
company named sells online
courses to industry
professionals
● Many interested
professionals land on their
Background website
● The company markets its
courses on several websites
X Education Company like Google. Once these
people land on the website,
they might browse the
courses or fill up a form for
the course or watch some
videos
● When these people fill up a
form providing their email
address or phone number,
they are classified to be a
lead
● Once these leads are
acquired, employees from
Background the sales team start making
calls, writing emails, etc.
X Education Company Through this process, some
of the leads get converted
while most do not
● The typical lead conversion
rate at X education is around
30%
● X Education gets a lot of
leads but its lead conversion
rate is very poor
● To make this process more
efficient, the company
Problem wishes to identify the most
potential leads, also known
Statement as ‘Hot Leads’
● If they successfully identify
this set of leads, the lead
X Education Company’s Problem conversion rate should go up
as the sales team will now
be focusing more on
communicating with the
potential leads rather than
making calls to everyone
● We will help them to select
the most promising leads,
i.e. the leads that are most
likely to convert into paying
customers.
Problem ● We are required to build a
model wherein we need to
Statement assign a lead score to each
of the leads such that the
X Education Company’s Problem customers with higher lead
score have a higher
conversion chance
● The CEO, in particular, has
given a ballpark of the target
lead conversion rate to be
80%.
Lead – Conversion Process
Visit to X
Conversion
Lead Generation: Visitors either
Lead to
process
Education provide Email id
1. Ads on Tele calling and
website by these & Contact Details ~30% leads get
websites like Emailing activity
potential converted
Google Or to all the leads
customers
2. Referrals (professionals) View videos etc
Proposed Solution:
A model to filter leads
so that leads to
conversion ratio is
80%+
Proposed Solution
Selection of Hot Communicating Conversion of Hot
Leads with Hot Leads Leads
Leads Clustering Focus Communication Increase conversion
We cluster the leads into Since we would have a Since we focussed on hot
certain categories based on smaller set of leads to have leads, which were more
their tendency or communication with, we probable to convert, we
probability to convert, might make more impact would have a better
thus, getting a smaller with effective conversion rate, and hence
section of hot leads to communication. we can achieve the 80%
focus more on. target.
For our Problem Solution, the
crucial part is to accurately
identify hot leads.
The more accurate we obtain the
Solution hot lead, the more chance we get
of higher conversion ratio.
Selection of Hot Leads
Since we have a target of 80%
conversion rate, we would want
to obtain a high accuracy in
obtaining hot leads.
Implementation
Loading & Observing the Univariate, Bivariate, and Performing pre-requisites
past data provided by Heatmap for numerical for RFE and Logistic
the Company and categorical columns Regression
Data Performing Data Model
Data Gathering
Cleaning EDA Preparation Building
Duplicate removal, null value Outlier Treatment,
treatment, unnecessary column Feature-Standardization
elimination, etc.
Selection of top 25 Reduction of columns Verifying our Final Model
features using RFE and Model re-building Accuracy etc. with model
built with PCA
Feature Model Model Verifying
Final Model
Selection Building Improvement with PCA
Model building using RFE for Final Model Analysis and
selected columns performance on Test Data
Plots (Visualization)
•EDA plots depicting variation in numerical columns for
those who Converted and those who didn't.
•EDA plots depicting variation in Lead Source Trend
•EDA plots depicting variation in categorical column
(Occupation, Specilisation) for those who Converted
and those who didn't.
•EDA plots depicting variation in
Numerical column for those who
Converted and those who didn't.
EDA plots depicting correlation (Heat Map) of all
selected columns (numerical columns and dummy
columns).
•Linear Regression Final Model Parameters
•Area under ROC = 0.80
•Final cut-off = 0.35
Inference / Conclusion
Overall accuracy on Test set:
0.756
Model Analysis Sensitivity of our logistic
regression model: 0.571
Performance of our Final Model
Specificity of our logistic
regression model: 0.869
Inferences from Top 3 variables in model, that
contribute towards lead
Model conversion are:
● Total Time Spent on Website
Business Insights Derived from our ● Lead Source with element
Model Google
● TotalVisits
Top 3 variables in my model, that
Inferences from should be focused are:
Model • Lead Source with Olark Chat
• Lead Source with organic
Business Insights Derived from our Search
Model • Non-Maharashtra Cities
Conclusion 1 (LR Model)
Our Logistic Regression Model is decent and accurate enough, when compared to the
model derived using PCA, with 75.6 % Accuracy on Test Set, 57.1 % Sensitivity and 86.9
% Specificity.
We can vary these parameters by varying the cut-off value and thus predict Hot leads
based on scenarios like availability of extra resources and vice-versa.
Conclusion 2 (Recommendation)
X Education Company needs to focus on following key aspects to improve the overall
conversion rate:
● Increase user engagement on their website since this helps in higher conversion
● Increase on sending SMS notifications since this helps in higher conversion
● Get Total visits increased by advertising etc. since this helps in higher conversion
● Improve the Olark Chat service since this is affecting the conversion negatively