← Back to Home
CUSTOMER INTELLIGENCE • CASE STUDY

Customer Profitability Segmentation via Pareto ABC Analysis

Role: Data Analyst (Personal Analytics Project) Timeline: Sep 2026 Stack: Python 3.x, Pandas, NumPy, Matplotlib, Seaborn, IQR Anomaly Auditing Scope: 51,290 Line-Items (25,728 Orders) / 17,415 Unique Customers

1. Dataset, Scope & Business Scenario

Project Type Personal Project / Simulated Business Case
Dataset Source Simulated E-Commerce CSVs (EcomSales, Customer, Product, Region)
Scale & Granularity 51,290 transaction line-items • 25,728 unique orders • 17,415 customer accounts
Time Range & Currency Jan 2020 – Dec 2023 (4 Years) • Simulated USD ($)

In this simulated year-end appreciation scenario, management faced a fixed cap of 500 physical luxury gift sets and needed to evaluate profit concentration across all 17,415 accounts.

  • Aggregate transaction-level Profit and quantify cumulative profit concentration.
  • Segment customer base into Group A (0–80%), Group B (80–95%), and Group C (>95%).
  • Audit whether the 500-gift quota covers Group A and propose a tiered rollout roadmap.
  • Profile VIP demographic personas (Income, Age, Occupation) and geographic clusters.
Profit Metric Definition (Verified in Code): Rankings use the pre-computed line-item Profit field supplied in EcomSales.csv (separate from gross Sales). Because the simulated schema lacks marketing/CAC, fulfillment shipping, and corporate SG&A expenses, this reflects transaction-level gross dataset profit rather than full bottom-line customer net profit.

2. Key Findings & Pareto Classification

The analysis revealed pronounced profit concentration: 12.94% of customers generate 80% of cumulative dataset profit.

Tier Cumulative Profit Range Customer Count Customer Share (%) Profit Contribution Strategic Designation
Group A ≤ 80% 2,254 12.94% 80.0% Core VIP Accounts
Group B 80% - 95% 3,842 22.06% 15.0% Potential High-Value
Group C > 95% 11,319 65.00% 5.0% Long-tail / Low-Margin

3. Budget Audit & Demographic Persona Profile

  • Gift Budget Audit: Group A has 2,254 VIPs, exceeding the 500-gift limit by 1,754 accounts.
    → Tiered Solution: Deliver physical luxury gifts to Top 1–500 VIPs; award digital VIP packages (vouchers + loyalty points) to ranks 501–2,254.
  • VIP Persona Profile: Average income of $56,681.46 (Median: $60,000) with a mean age of 46.0 years.
  • Occupational Distribution: Led by Professionals (29.64%) and Skilled Manual (24.98%).
  • Top Categories: Body care ($446,768.07) and Make up ($121,098.25).
  • Geographic Hubs: Los Angeles (95 VIPs), San Francisco (74 VIPs), and New York City (72 VIPs).

4. Methodology & Code Refactoring

  • Cross-Platform Portability: Decoupled hardcoded drive paths using os.path.join().
  • Dynamic Timestamp Calculation: Replaced fixed-year logic with pd.Timestamp.now().year.
  • Modular Constants: Parameterized ABC thresholds and budget quotas into maintainable global variables.
  • Data Integrity: Implemented defensive copy operations (.copy()) inside IQR outlier filtering utilities to prevent DataFrame side-effects.
Validation Plan: Historical spend alone does not prove future gift responsiveness; evaluate the tiered physical (Top 1–500) vs. digital VVIP (Ranks 501–2,254) retention pilot against a randomized holdout control group measuring 90-Day VIP Repeat Rate and Incremental Profit vs. Reward Cost.