Late orders are far more likely to receive poor reviews.
The low-review rate rises from 11.38% to 54.02% when delivery is late. This makes logistics reliability the clearest first lever for customer satisfaction.
Oracle SQL portfolio project
Marketplace ops, modeled in Oracle.
1.5M+ public e-commerce rows loaded into Oracle, validated with PL/SQL, and translated into executive analytics.
Strongest signal
54.02%
Low-review rate for late deliveries, compared with 11.38% for orders that were not late.
Orders
99,441
Loaded to Oracle.
Views
9
BI-ready SQL marts.
What the Oracle views found.
The dashboard uses verified query output from Oracle, not mock front-end data.
Item revenue13.59M
Revenue from order item price values.
Avg review4.09
Overall customer review score.
Late rate8.11%
Marketplace-wide late delivery rate.
Products32,951
Distinct products in the catalog.
Delivery delay vs review score
Worst late-delivery states
Revenue needs quality controls.
High-revenue sellers can still be operationally risky. Seller scorecards should combine revenue, late rate, and average review score.
Repeat customers are scarce but valuable.
Repeat customers represent a smaller group, but their average monetary value is nearly double the one-time customer group.
Credit card dominates payment value.
Credit card payments account for the largest payment value and average 3.51 installments.
Category focus is clear.
Health and beauty, watches and gifts, home goods, sports, and electronics accessories drive top category revenue.
Seller revenue does not equal seller health.
The Oracle seller scorecard flags sellers that generate strong revenue but carry delivery or review risk.
High-revenue seller scorecard
4869...229,472.63 revenue, 11.59% late rateWATCHLIST
7c67...189,417.67 revenue, 3.35 avg reviewHIGH RISK
4a3c...202,999.12 revenue, 3.80 avg reviewWATCHLIST
5324...222,776.05 revenue, 4.08 avg reviewHEALTHY
Oracle layer, not just a chart page.
The project includes load scripts, constraints, indexes, views, PL/SQL, documentation, and verified query output.
Tables9
Raw Olist staging layer.
Views9
Dashboard-ready SQL marts.
Indexes17
Join and filter performance support.
PL/SQL1
Validation package and package body.
PostgreSQL monitoring add-on.
A small support pack was added for roles that ask for PostgreSQL basics, routine database checks, and reporting cadence.
Portable monitoring SQL for database-support interviews.
The add-on covers table size, row counts, freshness, active sessions, blocking sessions, slow-query review, and index usage.
Daily operations.
CSV template for freshness, row count, active connection, blocking-session, and slow-query checks.
Weekly review.
Template for table growth, unused indexes, sequential scans, row-count variance, and reporting SLA review.
Monthly summary.
Capacity, performance, reliability, data freshness, and data-quality summary template for stakeholders.