ANALYZING CUSTOMER

PURCHASING BEHAVIOR
AND CAMPAIGN IMPACT
ACROSS MARKETING
CHANNELS

CITRA DIANI PUTRI

Business Intelligence Bootcamp Batch 11
 I am a Business Intelligence enthusiast with a
passion for turning data into actionable insights. I
recently completed my MSc in Computer Science with
Data Analytics from the University of York. I have
over 3 years of experience as a data analyst,
working extensively with tools like Excel and
Tableau. My focus has been on using data to support
decision-making processes and optimize business
outcomes. I am committed to applying my skills to
drive impactful results in business intelligence and
analytics.
TABLE OF CONTENT
Below is an overview of the key sections in this presentation, outlining the analysis and
insights derived from the customer engagement and purchasing behavior data.

Project Overview
Background
Business Problems
Data Understanding
Data Analysis and Insights
Dashboards and Visualization
Recommendation and
Actionable Insights

Page 00
PROJECT OVERVIEW TOOLS USED
The primary objective of this project is to analyze
customer behavior through RFM segmentation,
build predictive models to forecast customer
engagement, evaluate channel performance and
campaign impact, and provide insights into the
effectiveness of marketing strategies. These
insights will enable personalized marketing efforts
and optimize customer retention strategies.

Page 01
EXECUTIVE SUMMARY
High-Value Purchases are Driven by Frequent Catalog Buyers:
Customers who make frequent catalog purchases contribute the most to total monetary value, with
spending exceeding $1,500 for high-frequency buyers.
Web Channel Drives Frequent Purchases:
Among the three channels (web, store, catalog), web customers exhibit the highest purchase
frequency. Frequent web buyers make significantly more purchases.
Monetary Value Rises with Purchase Frequency Across All Channels:
Regardless of the channel, there is a clear positive correlation between the number of purchases
and the total amount spent. This trend is most pronounced for catalog and store channels,
highlighting opportunities for campaigns targeting frequent buyers.
Recent Buyers Tend to Purchase More Often:
Customers with lower recency tend to make more frequent purchases across all channels,
particularly in the web and store channels.

Page 02
PROJECT BACKGROUND
CUSTOMER SEGMENTATION CHANNEL PERFORMANCE AND
(RFM ANALYSIS) CAMPAIGN IMPACT
Businesses need to identify different customer Understanding which marketing channels drive the
segments to optimize marketing efforts and most conversions is essential for optimizing
maximize revenue. RFM analysis provides a marketing spend. Analyzing the performance of
powerful way to categorize customers based on each channel helps businesses allocate resources
their purchasing behavior, helping companies focus efficiently and tailor campaigns to the most
on high-value customers and retain those at risk of responsive customers.
churn.
MARKETING CAMPAIGN EVALUATION
PREDICTIVE MODELING
Not all marketing campaigns are equally effective, and
Predictive modeling allows businesses to forecast
evaluating their success is crucial for refining future
future customer actions using historical data. By
strategies. By analyzing customer responses to
identifying trends and patterns, companies can take
different campaigns, businesses can identify
proactive measures to improve customer
successful tactics and improve their return on
engagement and retention strategies.
investment (ROI). Page 03
BUSINESS PROBLEMS

LACK OF INABILITY UNCLEAR INEFFECTIVE

EFFECTIVE TO PREDICT CHANNEL CAMPAIGN
CUSTOMER CUSTOMER EFFECTIVENESS PERFORMANCE
TARGETING CHURN ASSESSMENT

Page 04
DATA UNDERSTANDING #1
The dataset is designed to provide a detailed understanding of customer demographics, purchasing behavior,
and engagement with marketing campaigns. It enables businesses to analyze customer spending patterns,
evaluate the effectiveness of marketing campaigns, and gain insights into customer loyalty and retention.

Demographics Marketing Campaign

Income: The customer’s yearly income.
Enhancement
Kidhome: Number of children in the customer’s AcceptedCmp1-5: Indicates whether the customer
household. accepted the company’s marketing campaigns 1
Teenhome: Number of teenagers in the customer’s through 5.
household. AcceptedCmpOverall: Total number of campaigns
Age: The customer’s age. accepted by the customer.
Customer_Days: How long the customer has been Response: Whether the customer responded to the
with the company (in days). last campaign.
Marital Status: Indicate the customer’s marital status. Complain: Indicates if the customer has lodged a
Education: Indicate the customer’s education level. complaint.

Page 05
DATA UNDERSTANDING #2
Purchasing Behavior
Recency: Days since the customer’s last purchase.
MntWines, MntFruits, MntMeatProducts, MntFishProducts, Dataset
Source
MntSweetProducts, MntGoldProds: Amount spent on various product
categories.
MntTotal: Total amount spent by the customer across all categories.
MntRegularProds: Amount spent on regular (non-discounted) products.
https://www.kaggle.co
NumDealsPurchases: Number of purchases made using discounts.
m/datasets/jackdaoud/
NumWebPurchases, NumCatalogPurchases, NumStorePurchases:
marketing-data/data
Number of purchases made via web, catalog, and store, respectively.
NumWebVisitsMonth: Number of web visits in the last month.

Page 06
 DATA PROFILING
DEMOGRAPHICS CUSTOMER ENGAGEMENT
Average income: $51,622 Average Campaign
Standard deviation of $20,713 acceptance rates: 7%
Average age: 51
Range Customer: 24 to 80 y.o. Most Recent Campaign
Majority: Married Acceptance: 15%

PURCHASING BEHAVIOR OPERATIONAL METRICS

Average total spending: $562 Average Recency: 49 days
Standard deviation: $575 Average purchases made through
Significant Spending on Wine deals: 2.3
Customers visit the website about
Used around 2.3 deals 5.3 times per month on average

Page 07
UNIVARIATE EDA: DEMOGRAPHICS

The majority of customers have no children

(58%) or no teenagers (52%) at home, and most
are either married (38.7%) or single (21.6%).
Customers’ incomes and ages are normally
distributed, with average incomes around
$51,622 and the most common age range
between 40 and 60 years.

Page 08
UNIVARIATE EDA: PURCHASING BEHAVIOR #1

Page 09
UNIVARIATE EDA: PURCHASING BEHAVIOR #2

The majority of customers spend relatively small amounts across all

product categories, with most purchases concentrated at lower
values, especially for wine and meat products.
Additionally, most customers participate in only a few deals, and
recency is spread relatively evenly, with many customers making
recent purchases.

Page 10
UNIVARIATE EDA: PURCHASING BEHAVIOR #3

Web purchases are more common than catalog

or store purchases, while store purchases tend to
have a slightly more evenly spread distribution.
Web visits are concentrated between 5 to 9 visits
per month, but most customers participate in only
a few deals, with 43.4% making only one deal
purchase..

Page 11
UNIVARIATE EDA: CAMPAIGN ACCEPTANCE

Campaign 1-5 Acceptance: Most customers did

not accept the marketing campaigns, with only a
small portion accepting any of the five
campaigns. This indicates low overall
engagement with the company’s marketing
efforts.
Total Campaigns Accepted: The majority of
customers accepted no campaigns, with only a
small percentage accepting more than one,
showing limited effectiveness across all
campaigns.

Page 12
UNIVARIATE EDA: RESPONSE & COMPLAINTS

Response to Last Campaign: Only a

small fraction of customers responded
to the last campaign, indicating low
engagement with the company's latest
marketing outreach.
Customer Complaints: Very few
customers lodged complaints,
suggesting that while engagement may
be low, customer dissatisfaction does
not appear to be a significant issue.

Page 13
MULTIVARIATE EDA: RFM CORRELATION

Frequency and Monetary (MntTotal) have a

strong positive correlation of 0.81, indicating that
customers who purchase more frequently tend to
spend more overall.
There is a very weak correlation between Recency
and the other variables, suggesting that how
recent a purchase was does not strongly relate to
either how often or how much a customer spends.
This highlights that customers’ recency of purchase
is relatively independent of their spending patterns
and frequency of purchases.

Page 14
MULTIVARIATE EDA: MARITAL STATUS GROUP
ANALYSIS
Frequency: Divorced customers have the highest
purchase frequency (14.28)
Monetary (MntTotal): Married customers spend
the most on average ($672)
Recency: The recency for all marital groups is very
similar, ranging between 48 to 50 days, indicating
that recent purchase behavior is consistent across
marital statuses.

Page 15
MULTIVARIATE EDA: MARITAL STATUS GROUP
ANALYSIS

Frequency: As income increases, so does the frequency of purchases. Very high income
customers make nearly 27.6 purchases.
Monetary (MntTotal): Higher income groups also spend significantly more. High income customers
spend $1,146, while low income customers spend only $55 on average.
Recency: Very high income customers have the most recent purchases, with a recency of 34 days,
while low income customers have a recency of 50 days.

Page 16
MULTIVARIATE EDA: MONETARY VALUE BY
NUMBER OF PURCHASES

Customers who purchase more frequently across all channels (web, catalog, store)
tend to spend significantly more. The relationship between frequency and monetary
value is quite strong, particularly for the catalog and store channels, where spending
increases more steeply with frequent purchases.
Page 17
MULTIVARIATE EDA: PURCHASE FREQUENCY
BY CHANNEL

As the number of purchases increases in each channel, the frequency of purchases

rises sharply, especially for customers making web and store purchases. High-
frequency buyers seem to use the web more than the store or catalog.

Page 18
MULTIVARIATE EDA: RECENCY BY NUMBER OF
PURCHASES

For all channels, customers with lower recency (recent buyers) tend to make more
purchases. However, catalog purchases show some inconsistency in recency, with
some high-purchase customers not buying as recently.

Page 19
THANK
YOU

2024

Citra Diani Putri

Business Intelligence Bootcamp Batch 11