Customer-Centric
Segmentation Analysis

Leveraging Data Science to Optimize Marketing ROI

MIT Applied Data Science Program
Capstone Project

August 2024 | www.congs.co.uk

Executive Summary

A multi-channel retailer needed to enhance marketing strategy efficiency by identifying actionable customer segments. The challenge was creating meaningful, distinct, and manageable segments that would enable targeted marketing actions and drive revenue growth.

The Objective

Leverage Data Science Unsupervised Learning and Clustering Algorithms to identify customer segments that are manageable, big enough to warrant resource allocation, and distinct enough to clearly dictate different marketing actions.

Key Findings

High Value

20% of customer base

High income, high spending, highest response rate to campaigns

48% of total revenue

Budget Care

43% of customer base

Low income, low spending, high web engagement

6% of total revenue

Med Spenders

37% of customer base

Moderate income/spending, preference for deals

46% of total revenue

Business Impact

Potential Revenue Boost: 10-15% increase through efficient segmentation
Resource Optimization: Targeted marketing strategies reduce waste and improve ROI
Actionable Insights: Clear customer profiles enable tailored product offerings and promotions
Framework for Growth: Test-and-learn approach allows refinement as customer behaviors evolve

Key Insight

The High Value segment (20% of customers) contributes 48% of revenue - a 2.4x importance ratio. This disproportionate contribution highlights the critical need for targeted retention and growth strategies.

Why Customer Segmentation Matters

Customer segmentation empowers businesses to move beyond one-size-fits-all marketing and create tailored strategies that drive measurable results.

Five Key Benefits

Empowerment

Empowers marketing and sales departments with information about customer needs and motivations

Enhanced Marketing Strategies

Enables tailored product and service offerings to specific customer groups

Increased Satisfaction

Improved customer experience and satisfaction as a result of targeted approaches

Sales Growth

Customer sales maintained or increased over time due to repeated purchasing activity

Loyalty & Retention

Drives customer retention and business growth achievement

The Bottom Line

OPTIMISING ALLOCATED RESOURCES AND INCREASING ROI

The Challenge

Business Requirements

The number of segments should be manageable to avoid complexity
Segments should be big enough to warrant resource allocation
Segments should be distinct and meaningful enough to clearly dictate different marketing actions

Data Science Approach

Use unsupervised learning algorithms (clustering)
Balance statistical rigor with practical business application
Create interpretable results for non-technical stakeholders
Enable dynamic adjustment as customer behaviors evolve

Exploratory Data Analysis

Before clustering, we analyzed the dataset to understand key patterns and relationships that would inform our segmentation approach.

Key Highlights

        Income-Spending Correlation: Higher income is linked to increased spending across categories. High-income customers make more purchases via catalogues and web.
Family Size Impact: Larger households tend to spend less, particularly on products like wines.
Marketing Campaign Effectiveness: Generally low acceptance rates for marketing campaigns with limited correlation to other factors.
Digital Behavior: High web visits do not translate to high purchases - negative correlation between web visits and overall expenses, suggesting opportunity to improve online conversion rates.
Purchase Channels: Strong correlation between total purchases and purchases made through catalogue/store channels.

    

Key Correlations Identified

Relationship	Correlation	Business Implication
Income → Spending	+0.73	High-income customers drive revenue - focus retention efforts here
Income → Catalogue Purchases	+0.71	Catalogue channel effective for high-value customers
Web Visits → Spending	-0.54	High web traffic but low conversion - optimization opportunity
Family Size → Spending	-0.50	Larger families spend less - may need budget-friendly offers
Campaign Acceptance → Other Factors	Weak	Campaigns not well-targeted - segmentation can improve this

Critical Finding

The negative correlation between web visits and spending reveals a significant conversion gap. Customers are engaging online but not purchasing - this represents a major opportunity for the Budget Care segment, which shows high web engagement but low spending.

Customer Cluster Profiling

Using Hierarchical Clustering (Ward Linkage), we identified three distinct customer segments, each with unique characteristics and marketing implications.

HIGH VALUE - 20% of Customer Base

Profile:

High income (avg. $76K)
Small family size
High spending (avg. $1,487)
Highly engaged (585 days tenure)
High campaign response (21%)

Behavior:

Highest average spending: $77 per purchase
Balanced multi-channel: store, catalogue, web
21 purchases per customer
48% of total revenue

Recommended Actions

Enhance omni-channel experience to drive incremental purchases
Target with premium products and services
Implement points-based incentive to reward campaign participation

BUDGET CARE - 43% of Customer Base

Profile:

Low income (avg. $34K)
Large family size
Minimal spending (avg. $84)
Lower engagement (503 days)
Lowest campaign acceptance

Behavior:

Highest web visits (6 times/month)
Low conversion rate
Store and web channels primarily
7 purchases per customer
6% of total revenue

Recommended Actions

Create web-specific, limited-time offers to convert traffic to orders
Develop targeted promotions and discounts
Introduce budget-friendly "recommend a friend" and gamification initiatives

MED SPENDERS - 37% of Customer Base

Profile:

Moderate income (avg. $59K)
Moderate family size
Significant spending (avg. $750)
Moderate engagement (559 days)
Moderately responsive to campaigns

Behavior:

Highest number of deals purchases
Prefer web and store purchases
20 purchases per customer
Oldest customer group (long-term loyalty)
46% of total revenue

Recommended Actions

Develop profile-appropriate campaigns and "long-term customer" benefits
Implement targeted offers to prevent churn
Encourage multi-channel purchasing through enhanced omni-channel experience

Cluster Analysis Highlights

Detailed analysis of spending patterns and channel preferences reveals distinct behaviors that inform targeted marketing strategies.

Spending Analysis

Segment	Volume of Purchases	Value per Purchase	Avg Spending per Customer
High Value	21	$77.44	$1,487
Budget Care	7	$10.31	$84
Med Spenders	20	$36.74	$750

Channel Preferences

Segment	Web Engagement	Store Purchases	Catalogue Purchases	Deals Engagement
High Value	26%	39%	29%	6%
Budget Care	26%	42%	6%	26%
Med Spenders	29%	38%	17%	16%

Key Insights

High Value Segment

Balanced multi-channel behavior with strong catalogue preference (29%). This segment values convenience and variety - omni-channel enhancement can drive incremental purchases.

Budget Care Segment

High web engagement (26%) but lowest catalogue use (6%). Strong deals engagement (26%) shows price sensitivity. Web-specific offers can convert traffic to sales.

Med Spenders Segment

Highest deals engagement (16%) combined with moderate web/store usage. This segment responds to promotions - targeted offers can prevent churn and drive retention.

Strategic Implication

Channel preferences vary significantly by segment. Marketing strategies must align with each segment's preferred channels to maximize effectiveness and ROI.

Business Importance & ROI Analysis

The "Importance Ratio" shows how much revenue each cluster contributes relative to its size - revealing which segments deserve the most strategic attention.

Importance Ratio Analysis

% of Customers

High Value: 20%

Budget Care: 43%

Med Spenders: 37%

% of Revenue

High Value: 48%

Budget Care: 6%

Med Spenders: 46%

Importance Ratio

High Value: 2.40

Budget Care: 0.14

Med Spenders: 1.25

What This Means

The Importance Ratio shows revenue contribution relative to segment size:

High Value (2.40): Contributes disproportionately high revenue - clearly the most valuable segment
Med Spenders (1.25): Contributes slightly more than proportional size - also valuable
Budget Care (0.14): Largest in size but contributes least to revenue - needs conversion optimization

Proposed Actions with ROI Analysis

High Value

435 customers | Enhanced omni-channel

128% ROI

Cost: $40k IT infrastructure

Benefit: $91,350 additional revenue

Budget Care

962 customers | Online discount offers

233% ROI

Cost: $1,443 discounts

Benefit: $4,810 additional revenue

Med Spenders

830 customers | Long-term tenure program

125% ROI

Cost: $8,300 program

Benefit: $18,675 additional revenue

Key Finding

All three segments show positive ROI when targeted with appropriate strategies. The Budget Care segment, despite low current revenue contribution, shows the highest ROI (233%) when targeted with web-specific discount offers - highlighting the conversion opportunity.

Solution Design & Implementation

Why Hierarchical Clustering?

What Solution

Primary: Hierarchical Clustering with 3 clusters
Complementary: GMM, K-means micro-segmenting check
Result: Distinct groups: High Value, Budget Care, Med Spenders

Why We Chose It

Flexibility: Adjustable segmentation using HC dendrogram
Interpretability: Clear, distinct segments for non-technical stakeholders
Actionable Insights: Well-distributed clusters (20%, 43%, 37%)

Which Problems It Solves

Addresses business needs for both high-level strategy and detailed targeting
Balances statistical rigor with practical business application
Allows adaptation as customer behaviors evolve
Supports segmented, focused marketing approaches

How It Impacts

Enables targeted marketing strategies for each segment
Improves resource allocation efficiency
Facilitates tailored product offerings and promotions
Increases customer lifetime value

Key Recommendations

        Start Small: Design high-impact, low-risk strategies for each segment (cross-sell, seasonal promotions, website enhancements)
Test & Learn: Implement test-and-learn approach to refine segmentation over time
Involve IT: Ensure systems and infrastructure can support segmented initiatives
Build Action Plans: Create detailed expectations and KPIs for each segment
Consider Micro-segmentation: Explore potential for sub-segments within the three main clusters for even more targeted marketing

    

How We Work

Our Approach

We combine data science rigor with business practicality. Every analysis answers "So what?" and "What do I do?" - turning complex algorithms into clear, actionable strategies.

What This Client Received

Customer segmentation analysis (3 distinct clusters)
Cluster profiling with actionable recommendations
ROI analysis for targeted marketing strategies
Implementation roadmap with risks and opportunities
Framework for ongoing refinement

Data In, Decisions Out

We help SMEs turn messy data into clear revenue actions.

Olga Sakka

olga.sakka@congs.co.uk | www.congs.co.uk