Skip to main content
Zakaria Maachou
Selected projects

Independent portfolio case study

02

Case study / CRM & Retention Analytics

Customer Segmentation RFM

Business question

Which customers should CRM prioritize?

Verified signalVIP customers represent 27.9% of customers and generate 75.4% of revenue.
Inspect the evidence
Customers
5,000
Orders
45,356
Total revenue
€4,522,014

Dataset disclosureSynthetic but business-realistic e-commerce order data generated with Python. The dataset contains 5,000 simulated customers and 45,356 orders and does not represent a real company.

01

Executive decision brief

The decision, before the documentation.

A concise chain from the commercial question to the recommended action.

  1. Question

    Which customers should CRM prioritize?

  2. Observed signal

    VIP customers represent 27.9% of customers and generate 75.4% of revenue.

    VIP share
    27.9% customers · 75.4% revenue
    Lost share
    23.62% customers · 2.95% revenue
  3. Interpretation

    Revenue is highly concentrated: VIP customers (27.9%) drive 75.4% of revenue, while Lost customers (23.62%) contribute only 2.95% — clear CRM prioritization signals.

  4. Recommended decision

    VIP retention: loyalty programs, exclusive offers, proactive account management

02

Primary evidence

Inspect the analytical exhibit.

The dashboard is presented as reviewable evidence, with the full analytical context preserved.

Exhibit 01

RFM Segmentation / primary analytical output

Customer Segmentation RFM dashboard showing rfm segmentation dashboard — segment distribution & revenue
Figure 01

RFM segmentation dashboard — segment distribution & revenue

Synthetic but business-realistic e-commerce order data generated with Python. The dataset contains 5,000 simulated customers and 45,356 orders and does not represent a real company.
03

Analysis and diagnosis

From signal to commercial meaning.

This independent case study develops a data-driven customer segmentation approach based on recency, frequency and monetary value to prioritize retention and win-back actions.

Customer concentration

Customer share versus revenue share
  1. 01

    VIP

    Retention priority

    27.9%
    75.4%
    27.9% customers · 75.4% revenue
  2. 02

    Lost

    Lower broad-campaign priority

    23.62%
    2.95%
    23.62% customers · 2.95% revenue

Revenue is highly concentrated: VIP customers (27.9%) drive 75.4% of revenue, while Lost customers (23.62%) contribute only 2.95% — clear CRM prioritization signals.

Finding / interpretation

Revenue is highly concentrated: VIP customers (27.9%) drive 75.4% of revenue, while Lost customers (23.62%) contribute only 2.95% — clear CRM prioritization signals.

Decision layer

Recommended business action

Recommendations follow the evidence in this independent case study; no tested uplift is implied.
  1. 01

    VIP retention: loyalty programs, exclusive offers, proactive account management

  2. 02

    At-risk win-back: targeted email campaigns before churn to Lost segment

  3. 03

    Upsell / cross-sell on Loyal segment to move toward VIP status

  4. 04

    Deprioritize broad campaigns on Lost segment — focus budget on recoverable At-risk

04

Method and quality

How the conclusion was built.

The technical record stays inspectable without displacing the business question.

Methodology

  1. 01

    Computed RFM scores on 5,000 customers and 45,356 orders

  2. 02

    Segmented customers into VIP, Loyal, At-risk and Lost clusters

  3. 03

    Quantified revenue concentration per segment

  4. 04

    Built CRM prioritization rules based on segment economics

  5. 05

    Delivered actionable recommendations per segment

Quality controls

  1. 01

    Python and pandas customer-level aggregation

  2. 02

    Documented RFM scoring and segmentation rules

  3. 03

    Generated KPI outputs and visual analysis

  4. 04

    Reproducible GitHub repository

Tools in service of the question

  • Python
  • pandas
  • CRM analytics
05

Transparency record

What this work does—and does not—claim.

Dataset origin, ownership and material limitations remain part of the main narrative.

01Project type
Independent portfolio case study
02Ownership
Individual end-to-end project
03Dataset origin and boundary
Synthetic but business-realistic e-commerce order data generated with Python. The dataset contains 5,000 simulated customers and 45,356 orders and does not represent a real company.
06

Evidence handoff

Inspect the work.

Open the underlying repository, methodology and analytical artifacts.

  1. 01
    Synthetic dataset generator

    Seeded Python generator defining customer behavior profiles, order frequency, recency and revenue distributions.

    (opens in a new tab)
  2. 02
    RFM scoring pipeline

    Customer-level aggregation, quintile scoring, mutually exclusive segment assignment and KPI export logic.

    (opens in a new tab)
  3. 03
    Committed run metrics

    Machine-readable output containing customer count, order count, revenue and exact segment-level results.

    (opens in a new tab)
  4. 04
    Segment summary

    Published segment-level customer counts, revenue shares and average revenue per customer.

    (opens in a new tab)
  5. 05
    CRM playbook

    Business actions, campaign ideas and monitoring KPIs for VIP, Loyal, At-risk and Lost customers.

    (opens in a new tab)
GitHub repositoryAll projects
Next case / Growth Profitability AnalyticsProfit Leak Analysis

Where is margin being destroyed?