p_02 · Quick Commerce · Power BI
Blinkit Pulse Analytics
Quick-commerce analytics on 5,000 Blinkit sample orders: sales, the 15-minute delivery promise, customer experience and marketing ROAS, on a star schema with 44 DAX measures in ₹ lakh / crore.
- Power BI
- DAX
- Power Query
- Python
- Claude Code
01 / Problem
Where does revenue come from, how far is delivery from the 15-minute promise, how satisfied are customers, and does marketing pay back?
02 / Data
- Blinkit Sales Dataset (Kaggle, public, synthetic): 5,000 orders, 16 Mar 2023 to 4 Nov 2024.
- 9 raw CSVs prepared in Python into a star schema: Orders, Marketing and Inventory facts; Date, Product and Customer dimensions.
- Personal data dropped, source quirks fixed (actual delivery time recomputed, delay reasons kept only for late orders).
03 / Approach
- Me: semantic model, 44 DAX measures in 8 folders, Indian number formats (₹1.10 Cr / ₹5.48 L), and every headline measure reconciled against an independent pandas calculation.
- Claude Code: wireframe, report layout and build automation, reviewed and published by me.
- 4 pages (Overview, Delivery, Customers, Marketing), one insight per chart, in Blinkit's own palette.
stage_1
9 raw CSVs
Kaggle sample
stage_2
Python prep
pandas · cleaning
stage_3
Star schema
3 facts · 3 dims
stage_4
44 DAX measures
₹ lakh / crore
stage_5
4-page report
Layout by Claude Code
04 / Live dashboard
05 / Business impact
- ₹1.10 Cr GMV from 5,000 orders at ₹2,202 average order value, 27.6% gross margin.
- Only 69.4% of orders arrive on time: average delivery 19.4 min against a 15-min promise.
- ₹1.63 Cr ad spend returned ₹3.22 Cr (ROAS 1.97x); Email is the best channel at 2.05x.
06 / Tools
- Power BI Desktop
- PBIP / TMDL
- DAX
- Power Query (M)
- Python (pandas)
- Claude Code