Tmall Order Analysis

E-commerce transaction analysis on 28,010 orders from China's largest B2C platform

Dataset: Feb 2020 Source: Tmall (天猫) Stack: MySQL + Tableau
Total orders
28,010
GMV
¥1.90M
Completed
18,955
67.67% conv.
ARPU
¥100.38
Conv. rate
67.7%
01 — Conversion funnel

Order funnel analysis

Tracking user progression from order placement through payment, transaction completion, to full-price completion. Each stage reveals where potential revenue is lost.

Placed
28,010
100.00%
Paid
24,087
85.99%
Completed
18,955
67.67%
Full price
16,650
59.44%
Key finding: The largest drop-off (18.32 pp) occurs between payment and completion, indicating a substantial refund rate. 14.01% of orders that were placed were never paid, while 21.36% of paid orders resulted in partial or full refunds.
02 — Daily performance

Daily KPI trends

Core metrics tracked by payment date, covering the observation window of February 1 through February 10, 2020.

Date Orders Completed GMV (¥) ARPU (¥) Conv. % Refund %
Feb 01163877,03180.8253.3789.66
Feb 021991078,50879.5153.7791.59
Feb 0323913611,31783.2156.9078.68
Feb 0441725421,92686.3260.9166.54
Feb 0532819115,49581.1358.2372.25
Feb 06128726,02583.6856.2581.94
Feb 07157967,18574.8461.1564.58
Feb 08213838.0050.00100.00
Feb 0935226622,12383.1775.5732.33
Feb 1024161,01063.1366.6756.25
Key finding: Feb 04 and Feb 09 emerge as peak days with GMV exceeding ¥21,000. The anomaly on Feb 08 (only 2 orders) suggests a data collection gap or system event. Feb 09 shows the highest conversion rate (75.57%) paired with the lowest refund rate (32.33%), representing the healthiest transaction day in the dataset.
03 — Temporal patterns

Hourly and weekly distribution

Understanding when customers transact helps optimize marketing spend, inventory management, and customer service staffing.

Hourly transaction volume

Completed orders by hour (0:00 to 9:00, partial data)

0:00
¥67.7K
1:00
¥36.4K
2:00
¥23.4K
3:00
¥13.8K
4:00
¥9.4K
5:00
¥7.3K
6:00
¥15.7K
7:00
¥34.4K
8:00
¥55.8K
9:00
¥94.6K

Weekly transaction volume

Revenue distribution across weekdays

Sun
¥184K
Mon
¥180K
Tue
¥319K
Wed
¥280K
Thu
¥290K
Fri
¥353K
Sat
¥297K
Key finding: The midnight hour (0:00) shows unexpectedly high activity with 683 completed orders, likely driven by post-midnight browsing behavior. Revenue peaks sharply at 9:00 AM with ¥94.6K. On a weekly basis, Friday generates the highest revenue (¥352.9K), while Sunday and Monday are the slowest days, suggesting that weekday marketing investments may yield better returns.
04 — Geographic distribution

Revenue by province (Top 10)

Transaction concentration across Chinese provinces, ranked by total completed revenue.

01Shanghai (上海)
¥264,067
2,470 orders · 73.67% conv.
02Beijing (北京)
¥166,470
1,489 orders · 72.49% conv.
03Jiangsu (江苏)
¥159,377
1,459 orders · 68.63% conv.
04Guangdong (广东)
¥147,845
1,585 orders · 64.35% conv.
05Zhejiang (浙江)
¥141,686
1,438 orders · 69.77% conv.
06Sichuan (四川)
¥127,668
1,380 orders · 68.35% conv.
07Shandong (山东)
¥103,930
1,145 orders · 63.47% conv.
08Tianjin (天津)
¥90,001
838 orders · 72.68% conv.
09Liaoning (辽宁)
¥74,698
812 orders · 68.41% conv.
10Chongqing (重庆)
¥71,525
691 orders · 66.70% conv.
Key finding: Shanghai dominates with ¥264K in revenue and the highest conversion rate (73.67%) among major provinces. Tier-1 cities (Shanghai, Beijing) demonstrate both higher volume and higher conversion rates compared to other regions. Notably, Guangdong, despite ranking 4th in order volume, has the lowest conversion rate (64.35%) in the top 10, suggesting potential regional differences in purchase intent or product-market fit. The average ticket size is highest in Beijing (¥111.80) and Tianjin (¥107.40).
05 — Methodology

Technical approach

End-to-end data pipeline from raw CSV ingestion to interactive dashboard.

Data pipeline

Three-stage processing workflow

1. Ingestion: CSV (UTF-8 with Chinese characters) imported into MySQL 8.0 via a custom Python script to handle encoding issues.
2. Transformation: SQL-based cleaning including province name standardization, derived date fields, order status classification, and refund flags.
3. Visualization: Tableau Desktop connected to MySQL, with both live queries and custom SQL data sources powering the dashboard.

Metric definitions

Consistent calculation methodology

GMV: Sum of buyer's actual payment where amount > 0
Conversion rate: Completed orders / Total orders placed
ARPU: GMV / Number of completed orders
Refund rate: Orders with refund / Completed orders