Olist E-commerce Dashboard

Which part of the operation costs the most?

An e-commerce business can grow its sales and still lose margin to late deliveries, unhappy customers and the wrong product mix. I wanted to find out where that money goes, using real transaction data instead of guesses.

Start from the questions, then build the views.

• How do revenue and order volume evolve month by month?
• Which categories carry the revenue?
• Where are deliveries slowest, and does that show up in review scores?
• How do customers pay?

Clean, join, aggregate, visualise.

The Olist dataset is split across several tables (orders, items, payments, reviews, customers, products). I cleaned and joined them in Python, computed the indicators, and built an interactive dashboard on top that can be filtered by date range and by state.

The stack.

Analysis

Python, Pandas, SQL

Dashboard

Streamlit, with a React front end

Data

Brazilian E-Commerce Public Dataset by Olist

A working dashboard.

The dashboard brings together the headline figures (revenue, orders, unique customers, average review score), monthly sales, revenue by Brazilian state, the best-selling categories, order statuses and delivery times.

Public data. The figures in the screenshot come from the public Olist dataset, not from a client. I’ll add my conclusions to this page once I’ve checked them.

The dashboard.