The Challenge
A mid-market fashion boutique was making inventory and pricing decisions based on intuition. Managers had no visibility into which brands were driving returns, which product categories were underperforming, or how much revenue was being lost to markdowns. Without a structured view of their 2,176-record transaction dataset, dead inventory accumulated and pricing decisions were reactive rather than strategic.
The Solution
I built an ETL pipeline (Python + pandas) that cleaned the raw CSV, engineered 8 derived KPIs (revenue_lost, price_band, rating_band, is_dead_inventory, etc.), and exported to Parquet for fast loading. On top of the clean data I delivered a 6-page Streamlit dashboard — Executive Summary, Category Deep Dive, Brand Performance, Markdown & Pricing, Returns Analysis, and Inventory Status — each page answering a specific decision managers actually face. A reusable Plotly/Streamlit design system ensures visual consistency across all chart types.
Results
118
Dead inventory SKUs identified (>30% markdown + low rating)
$5,762
Capital tied up in dead stock — surfaced for clearance
$25,460
Revenue lost to markdowns quantified across catalogue
<2s
Dashboard load time on cleaned Parquet file
Tech Stack
“We finally have a single place to see what's moving, what's dying, and where our markdowns are eating into margin. The dead inventory view alone paid for the project in the first week.”
Store Manager
Luxe & Thread Boutique