Understanding customer behaviour and improving customer
experience is critical in competitive
market place. Today’s customer expects personalized experiences with retailers
and brands. To meet such expectations and to improve the business growth, companies
seek various customer analytics techniques and apply insights from customers
across the globe. The dataset we have chosen for analysis consists of data of a
company called dunnhumby. Dunhumby enables Retailers and brands to tap into the
most advanced customer analytics technique to provide the best services to meet
their customer’s expectations. As part of this project, we intend to apply
various data analytics techniques which can help retailers to decide on the
amount to be spent in acquiring customers.
There are 7 tables provided for the analysis. These tables needs
to be consolidated and the data needs to be cleansed, transformed, and turned
into meaningful information before Analysis.Our Analysis is based on the Acquistion
stage of Customer Lifecycle which provides useful insights to the company to
decide upon the amount to be spent on acquiring the customers.Below is the list
of the tables.
This table consists of demographic information for households with
the details such as household_key,Age_Desc,Marital_status_code,Income_desc,homeowner_desc,hh_comp_desc,household_desc,hh_comp_desc,housegold_size_desc,kid_Category_desc.Household_key
field is unique, and can be used to join the tables as required.
This table consists of
all products purchased by households with the details such as
uniquely identifies each household.Basket_ID uniquely identifies the purchase
table lists the campaigns received by each household in the study. Each
household received a different set of campaigns.
4)Campaign_desc : This
table gives the length of time for which a campaign runs. So, any coupons
received as part of a campaign are valid within the dates contained in this
5)Coupon : This table
lists all the coupons sent to customers as part of a campaign, as well as the
products for which each coupon is redeemable.
table identifies the coupons that each household redeemed.
7)Product dataset: This
table contains information on each product sold such as type of product,
national or private label and a brand identifier
Below techniques will be
implemented to solve our business case of deciding on the customer acquisition
1)Market Basket Analysis:
On the basis of the
basket of the products purchased,what recommendations can be made by the
company to improve the sales?
What offers can be given
by the company on the basis of the products purchased ?
Assess LTV for each
Assess Retention Rate.
Identify the best
marketing methods according to each cluster.
Assess which marketing
segment has the highest Return of Investment on marketing campaigns.
factors (e.g. household size, presence of children, income) appear to affect
Does direct marketing
improves customer acquisition?
How many customers are
spending more/less with time?Which segment of customers are growing faster?
2)Next Best Offer :
What is the next best
offer the company can provide to its customers?
Some of the challenges we may face as part of the analysis
process are as follows:
Sampling the data for
train,test and validation datasets.
Evaluating accuracy of
the prediction model.
Assessing overall return
of investment on the cost associated with acquiring the customers.