Olist E-Commerce Analytics
- Late deliveries cut review scores from 4.03 to 2.27 — the single largest lever on customer satisfaction in the data.
- The top 20% of sellers generated 82.54% of GMV — a concentration risk nobody had flagged.
The Problem
Olist's marketplace data existed, but nobody had connected delivery performance to customer satisfaction, or checked how concentrated revenue was across sellers. Both were assumed risks. Neither had a number attached — which meant neither could be prioritized against anything else.
The Method
Joined 9 raw datasets into one analytical base covering 96,478 orders across 27 states. Segmented orders by delivery timing, seller, and region to isolate which variable actually moved customer sentiment and revenue concentration.
Recommendation & Outcome
Recommended cutting the late-delivery rate from 6.8% to under 3%, and flagged seller concentration as a standing risk to monitor rather than a one-time finding — with the top fifth of sellers carrying over 80% of GMV, losing even one materially exposes revenue. This was delivered as a self-directed analysis; the 3% target and the concentration flag are the concrete asks a real ops team would act on next.
Why It Mattered
This turned two assumed problems into two prioritized, numbered ones. Leadership no longer has to guess whether delivery speed is worth fixing — the review-score gap makes the business case on its own.