Brazilian E-Commerce: 3% Repeat, 8% Late

BRAZILIAN E-COMMERCE ANALYTICS

Business question

Where should a marketplace focus first to improve repeat purchase, delivery experience, and profitable growth?

Evidence base

99,441 customers and orders; 112,650 order items; 103,886 payments; 99,224 reviews; and 3,095 sellers. I built a PostgreSQL schema, nine SQL analysis modules, a Python ingestion workflow, and a five-page Power BI dashboard.

What the analysis showed

• Only about 3% of customers placed multiple orders, making retention the clearest growth gap.

• Roughly 8% of orders arrived late, and delayed orders were associated with lower review scores.

• Freight-to-revenue ratios varied materially by category, so revenue rankings alone could hide margin pressure.

• More than 3,000 sellers reduced broad concentration risk, while a smaller high-performing group still drove a disproportionate share of revenue.

Business decision

Prioritize lifecycle campaigns for one-time buyers, investigate the operational causes of late delivery, and use category-and-seller scorecards that pair revenue with freight burden and satisfaction.

Uncertainty

This is observational marketplace data, so the dashboard identifies where to act—not causal impact. Delivery delays and review scores may share confounders; proposed interventions should be validated with controlled experiments.

The completed repository includes the SQL pipeline, Power BI file, dashboard screenshots, and full technical report.