Retail Sales BI Capstone

Project Overview

This project analyzes retail sales performance across multiple regions, product categories, and store types.

Using Excel, PostgreSQL, and Tableau, I built a complete BI workflow to uncover trends, identify inefficiencies, and generate actionable insights for business decision‑making.

Part of my IBM: BI Analysis final project.

Dataset: Modified subset of the Kaggle Retail Sales dataset (cleaned to fit Excel/Tableau limits).

Objective: Understand how store location, product hierarchy, and promotions influence sales.

Tools & Techniques

  • Excel: Data cleaning, validation, pivot tables, regression
  • PostgreSQL: Joins, aggregations, data cubes, ROLLUP
  • Tableau: Dashboards, regional analysis, product insights

1. Data Cleaning & Preparation (Excel)

I cleaned and standardized 450k+ rows of sales data:

  • Removed duplicates and whitespace
  • Standardized inconsistent date formats
  • Validated product, store, and city IDs
  • Combined sales, stock, and pricing data for analysis

Exploratory Pivot Tables

I built three pivot tables to understand baseline performance:

  • Sales by Store: Identified high‑ and low‑performing locations
  • Sales by City: Highlighted regional differences
  • Product Performance: Flagged high‑potential and low‑demand products

These tables guided the SQL analysis by revealing where deeper investigation was needed.

2. SQL Querying & Aggregation (PostgreSQL)

I loaded the cleaned dataset into PostgreSQL to perform structured analysis.

Key SQL Workflows

  • Joined sales, product hierarchy, and city tables
  • Aggregated sales by store and month
  • Calculated total revenue and average price
  • Built data cubes using GROUP BY ROLLUP to analyze:
    • city → month → total sales
    • product → store → revenue patterns

Example Insight

ROLLUP revealed seasonal fluctuations in several cities, with noticeable dips in March and spikes in February.

3. Statistical Analysis (Excel)

I ran a simple regression to explore the relationship between:

  • Date
  • Sales volume

This helped identify early seasonal patterns and informed the Tableau dashboard design.

4. Tableau Dashboards

I created interactive dashboards to visualize:

  • Regional sales performance
  • Product category trends
  • Store‑level efficiency
  • Revenue vs. stock levels

These dashboards allow stakeholders to quickly compare cities, drill into product categories, and identify operational bottlenecks.

Key Insights

  • Edinburgh and Helsinki consistently outperformed other cities
  • Several products showed high stock but low sales, indicating inventory inefficiency
  • Product categories like Counter‑depth Refrigerators and Convection Ovens were strong performers
  • Seasonal patterns emerged when aggregating sales by month
  • Promotions had uneven impact, with some promo types showing no measurable lift

What I Would Do Next

If this were a real business environment, next steps would include:

  • Forecasting sales using time‑series models
  • Building automated dashboards
  • Adding customer segmentation
  • Conducting price elasticity analysis