Analytics & SQL · Live demo

Customer Retention & Revenue Analysis

Explore what happened after customers placed their first observed order.

identified purchasers: 4,297; identified purchasers
repeat within 90 days among eligible customers: 47.4%; repeat within 90 days among eligible customers
net transaction value from the top 10%: 59.6%; net transaction value from the top 10%

Real historical transactions from Chen (2015), UCI Online Retail, CC BY 4.0. The fixed 2010–2011 snapshot is not automatically updated. First observed purchase is not necessarily acquisition; anonymous transactions are excluded and refunds have limitations. Findings are descriptive, not causal.

Overview

A reproducible analysis and interactive explorer turn historical UCI Online Retail transactions into cohort activity, repeat-purchase and monthly views for sales and finance.

Inside the interactive explorer

Retention explorer overview and key figures — 01Overview of the historical snapshot and its key figures.

Cohort selector and monthly purchase activity — 02Choose a first-observed cohort and follow its monthly activity.

Monthly activity chart with a metric selector — 03Compare monthly active customers, orders or gross purchase value.

90-day repeat rate and observed order distribution — 04Compare 90-day repeat with the distribution of observed orders.

The Challenge

Incomplete customer histories, unequal follow-up periods, anonymous transactions and refunds can make retention and revenue measures misleading.

The Solution

Python builds customer and order metrics, SQL independently checks key aggregates, and a static interactive demo lets visitors explore cohorts, monthly activity and order frequency with definitions and caveats.

Explore the analysis

01
Choose a cohort
Select the month of first observed purchase and follow purchasing activity in subsequent complete months. Unavailable months are not zero.
02
Compare monthly activity
Switch between active purchasers, qualifying orders and gross purchase value for complete calendar months.
03
Understand repeat purchasing
Review 90-day repeat among customers with enough follow-up and the distribution of observed order counts.
04
Read the business brief
Inspect revenue concentration, definitions, refund treatment and questions for further investigation in the linked report.

Results & Impact

Of 3,324 eligible customers, 1,577 (47.4%) placed a second order within 90 days. The analysis covers a UK retailer from December 2010 through November 2011; observed patterns do not establish causes of churn.

What the analysis offers

Comparable follow-up

The 90-day measure excludes customers without a full 90 days of observation.

Reproducible

Python produces aggregates and a report; SQL checks key metrics independently.

Careful interpretation

Censoring, anonymous transactions and refunds are disclosed; patterns alone do not explain churn.