Context

I joined BestMarket (fictional) for the “RetailInsight360” project, led by Mélanie (lead data analyst) and Olivier (customer service). The objective was to clearly reformulate business needs, load and enrich the customer feedback database, and write and execute over 20 SQL queries to measure satisfaction and support operational decisions.

Datasets

• Initial database customer_data_feedback.sql imported into SQLite Studio/DBeaver: retour_client table (3,000 rows), produit table (145 entries), ref_magasin table (84 stores)
• Data dictionary (Excel) describing each column (name, type, constraint, description)
• ref_magasin.csv file to complete store location metadata

Workflow

• Requirements gathering: validated the context, business stakes (NPS, returns by category), and KPIs to produce
• Schema loading and update: imported the SQL script, created the ref_magasin table from CSV, updated the relational schema and data dictionary
• Advanced SQL queries: used SELECT, JOIN, aggregations, CTEs, window/date functions, aliases, filtering and ranking to answer 21 business questions (feedback by channel, overall and channel-based NPS, top 5 stores, quarterly evolution, etc.)
• Data integrity checks: uniqueness tests on PK/FK, value range control, mismatch detection (rating vs. recommendation), added SQL constraints to prevent anomalies
• Presentation & documentation: created a 20-slide PDF deck including context, backup/storage best practices, methodology, screenshots of schema and updated dictionary, query snippets and results, and integrity control outputs

Insights

• 639 “delivery” feedbacks and top 3 feedback sources: email (1,032), social media (998), phone (970)
• Average satisfaction by product category: Electronics: 8.16, Leisure: 8.09, Food: 8.04, Home: 7.85
• Top 5 stores by rating: Paris 14th, Saint-Pierre-du-Perray, Paris 19th, Paris 11th, Coulommiers
• Global NPS: 31 pts (Promoters > 8, Detractors ≤ 6), NPS by channel: Phone (34), Email (30), Social media (30)
• Best day in-store: Saturday
• Month with most feedbacks: October
• Notable trends: +4.24% increase in “Leisure” feedbacks from Q1 to Q2 2021
• Additional recommendations: include total sales volume to calculate real return rate ; investigate rating/recommendation mismatches ; expand category dictionary for clarity

Business Impact

My deliverables allowed BestMarket to:

• Prioritize store-level and category-level actions based on actual customer satisfaction
• Implement channel-based NPS monitoring to improve multichannel experience
• Strengthen data governance by adding constraints and integrity controls
• Provide a reusable framework for future customer feedback analysis

Links

Requirements brief

Presentation

Executive summary

Data integrity checks