← Back to Home
CUSTOMER ANALYTICS • CASE STUDY
Customer 360 & RFM Segmentation Analysis
1. Dataset, Scope & Business Scenario
Project Type
Personal Project / Simulated Business Case ("CosmeticWorld")
Dataset Source
Simulated E-Commerce CSVs (
EcomSales, Customer, Product, Region)
Scale & Granularity
51,290 transaction line-items • 25,728 unique orders • 17,415 unique customers
Time Range & Currency
Jan 2020 – Dec 2023 (4 Years) • Simulated USD ($)
In this simulated CosmeticWorld scenario, marketing leadership needed to restructure annual promotional spend for a New Year Rejuvenation Campaign, shifting from broad-spectrum discounting toward targeted lifecycle retention. Note that 51,290 represents order line-items across 25,728 distinct OrderIDs.
- Quantify customer behavior across Recency, Frequency, and Monetary dimensions using quantile scoring (
pd.qcut). - Classify the customer base into actionable operational cohorts (
Champions,Loyal Customers,Cant Lose Them,Hibernating,Lost). - Construct a Customer 360 Profile analyzing demographic, geographic, and basket affinities.
- Suppress low-yield outreach on dormant accounts to prioritize budget allocation toward higher-retention tiers.
Methodological Definitions (Verified in Code):
Snapshot Date:
2024-01-01 (max(OrderDate) + 1 day) •
Recency: Days since latest order to snapshot date •
Frequency: Distinct OrderID count (nunique, preventing line-item inflation) •
Monetary: Gross Sales sum in simulated USD (the dataset has no return/refund or ad-spend/CAC tables) •
Scoring: Quintile pd.qcut(q=5) (with .rank(method="first") tie-breaking on Frequency) mapped via deterministic R/F/M threshold rules into 8 segments.
2. RFM Segment Performance Matrix
The analysis revealed strong revenue concentration: Champions & Loyal Customers account for under 29% of the customer base but generate over 53% of company revenue ($3.45M+).
| RFM Segment | Customer Count | Customer Share (%) | Total Revenue ($) | Revenue Share (%) | Strategic Designation |
|---|---|---|---|---|---|
| Champions | 2,361 | 13.56% | $2,138,211 | 32.8% | Core Brand Advocates |
| Loyal Customers | 2,616 | 15.02% | $1,319,383 | 20.2% | Consistent Multi-Repurchasers |
| Big Spenders | 1,979 | 11.36% | $1,354,794 | 20.8% | High Basket Value / Low Freq |
| Cant Lose Them | 833 | 4.78% | $653,354 | 10.0% | Critical High-Value Churn Risk |
| Hibernating | 4,701 | 26.99% | $715,415 | 11.0% | Inactive / Low Engagement |
| Need Attention | 3,274 | 18.80% | $263,137 | 4.0% | Borderline Engagement |
| New / Promising | 1,171 | 6.72% | $61,198 | 0.9% | Recent First-Time Buyers |
| Lost | 480 | 2.76% | $12,182 | 0.2% | Dormant Accounts |
3. Customer 360 View: Demographic & Basket Affinities
- Core Demographics: Peak gross sales (>60%) are driven by ages 25 to 49. High correlation between degree holders (Bachelor’s/Graduate) averaging $60k–$80k income and top Average Order Value (AOV).
- Geographic Hubs: Asia Pacific and Europe serve as the primary revenue drivers in the simulated dataset, outpacing USCA and LATAM markets.
- Basket Preferences: Champions heavily prioritize premium self-care lines:
Nail Care,Hand Creams, andBody Moisturizers.
4. Proposed Campaign Experiments & Evaluation Plan
- Hypothesis — Weighted Budget Focus (70/30 Pilot): Prioritize ~70% of promotional spend as a controlled pilot toward Champions (loyalty perks) and Cant Lose Them (win-back reactivation vouchers, $653k historical pool), while keeping 30% for nurturing Big Spenders and New / Promising tiers.
- Hypothesis — Dormant Outreach Suppression: Pause higher-cost SMS/direct mail to Lost and Hibernating tiers (5,181 combined accounts) to reduce low-yield promotional outreach spend, testing low-cost email-only reactivation instead.
- Hypothesis — Cross-Selling Bridge: Offer top-affinity subcategories (
Nail Care&Hand Creams) as Gift-With-Purchase incentives for New / Promising buyers to test graduation into Loyal Customers.
Observed Insight vs. Business Hypothesis: The revenue concentration (<29% generating >53% revenue) and demographic profiles (ages 25–49 driving >60% gross sales) are empirically observed from the 2020–2023 dataset. The proposed 70/30 split is an initial testing hypothesis rather than a proven optimum—requiring an A/B holdout test tracking channel CAC, Incremental Gross Margin, Reactivation Rate, and 60-day Repeat Purchase Rate before scaling.