Ravindra BagaleCourses & study guides मराठी Track your progress

How to Understand a Dataset Before Analysis: Blinkit Quick Commerce Example (Business Questions)

By Last updated:

You now know enough functions to answer a lot of questions. But knowing formulas is not what makes someone a data analyst. The real skill comes before the first formula: understanding what the data is, what every column means, and which business questions it can answer. In this chapter we take a small orders file from two quick-commerce apps, Blinkit and Amazon Now, in six Maharashtra cities, and go through it the way an analyst does on day one of a project.

What you'll learn in this chapter

  • E-commerce vs quick commerce
  • What one row of the data means
  • Every column, grouped by what it tells you
  • Cost_Price vs Selling_Price, revenue and profit
  • Status and delivery columns, and the rules they need
  • A data dictionary
  • Business questions and the decisions they lead to
  • First answers with COUNTIF, COUNTIFS and SUMIF

Practice file: quick-commerce-orders-practice.xlsx has the Orders sheet (24 orders with Revenue, Total_Cost and Profit formulas), the Questions sheet (first answers) and the Data dictionary sheet.

Made-up data: every order, product, price and number in this chapter is made up for practice. Product names are generic (for example "Milk 500 ml (local brand)"), not real brands.

Classroom line

Applying formulas doesn't make you a data analyst. The one who understands the data is the data analyst.

1. E-commerce vs quick commerce

E-commerce Quick commerce
Delivery time one day or more about 10 to 15 minutes
Ships from large warehouses, far away small "dark stores" inside the city
Typical order anything, often one big item daily needs: milk, bread, eggs, snacks
What decides success price and range speed, stock in the nearby store, distance

That last row already tells you what to look at in the data: delivery time, distance and what each store needs to keep in stock.

E-commerce vs quick commerce E-commerce Ships from: large warehouse, far away Delivery: one day or more Order: anything, often one big item Success: price and range Quick commerce (made-up data in this chapter) Ships from: small dark store in the city Delivery: about 10 to 15 minutes Order: milk, bread, eggs, snacks Success: speed, stock nearby, distance The last line tells you what to study: delivery time, distance and stock per store.

Figure 1. E-commerce ships from far-away warehouses in a day or more; quick commerce delivers daily needs from a nearby dark store in minutes.

2. First look: what is one row?

Open the Orders sheet. Before reading any number, answer three questions:

  1. What is one row? One order: one customer, one product, one delivery.
  2. How many rows and columns? 24 orders, 14 data columns plus 3 calculated ones.
  3. What period and places? 1 to 8 September 2026; Pune, Solapur, Nashik, Sambhaji Nagar, Kolhapur and Nagpur; two apps.

The first rows:

Order_ID City Product_Name Quantity Cost_Price Selling_Price Status Distance_Km Delivery_Time_mins Delivery_Status
BLK-2001 Pune Milk 500 ml (local brand) 4 26 30 Delivered 1.8 11 On Time
BLK-2002 Pune Curd 400 g 2 57 71 Delivered 2.4 13 On Time
AMN-2003 Nashik Atta 5 kg 1 205 240 Delivered 3.9 19 Late
BLK-2004 Solapur Bread 400 g 2 35 45 Delivered 1.2 9 On Time
BLK-2006 Nagpur Cold drink 750 ml 6 32 40 Cancelled 4.6 (blank) (blank)

3. The columns, grouped

Reading 17 column names one by one is tiring. Group them by the question they answer:

  1. Which order? Order_ID (BLK- for Blinkit, AMN- for Amazon Now), Order_Date, Platform.
  2. Where? City, Area.
  3. What? Product_Name, Quantity.
  4. Money: Cost_Price, Selling_Price, and the calculated Revenue, Total_Cost, Profit.
  5. How paid? Payment_Mode: COD (cash on delivery) or Online.
  6. What happened? Status: Delivered, Cancelled or Returned.
  7. How fast? Distance_Km, Delivery_Time_mins, Delivery_Status (On Time or Late).
17 columns, 7 questions 1. Which order? Order_ID, Order_Date, Platform 2. Where? City, Area 3. What? Product_Name, Quantity 4. Money Cost_Price, Selling_Price, Revenue, Total_Cost, Profit 5. How paid? Payment_Mode 6. What happened? Status 7. How fast? Distance_Km, Delivery_Time_mins, Delivery_Status Green-group columns Revenue, Total_Cost and Profit are calculated with formulas.

Figure 2. Seventeen columns become seven easy groups: which order, where, what, money, payment, what happened, and how fast.

4. The money columns

Two prices, and it matters which is which:

  1. Cost_Price: what the company pays its supplier for one unit.
  2. Selling_Price: what the customer pays for one unit.

From them, three calculated columns (formulas on the Orders sheet):

Column Formula Curd, BLK-2002 (2 units, cost 57, selling 71)
Revenue =G5*I5 (Quantity × Selling_Price) 2 × 71 = 142
Total_Cost =G5*H5 (Quantity × Cost_Price) 2 × 57 = 114
Profit =O5-P5 (Revenue − Total_Cost) 142 − 114 = 28
Per-unit prices × Quantity Quantity 2 per unit Cost_Price 57 | Selling_Price 71 Revenue =G5*I5 2 × 71 = 142 Total_Cost =G5*H5 2 × 57 = 114 Profit =O5-P5 142 − 114 = 28 Order BLK-2002, Curd 400 g. Ask of every money column: per unit or whole order? With tax? Paid by whom?

Figure 3. Cost_Price and Selling_Price are per unit. Multiply by Quantity for Total_Cost and Revenue; the difference is the profit.

Ravindra Bagale's Tip

Whenever a dataset has a column called "price", "amount" or "cost", ask: per unit or for the whole order? With tax or without? Paid by whom? Write the answers in your data dictionary. Half of all wrong reports come from mixing up two money columns that look alike.

5. Status and delivery: rules before formulas

These columns need a decision before you calculate anything:

  1. Cancelled orders were never delivered. Their Delivery_Time_mins and Delivery_Status are blank, and that is correct, not missing data. Their Revenue column still shows a number (Quantity × Selling_Price) because the formula doesn't know the order was cancelled.
  2. Returned orders were delivered and then sent back. The company earned nothing and paid a rider twice.
  3. So decide the rule: count revenue and profit only for Delivered orders. On this data that gives revenue 2,655 and profit 447, with =SUMIF(Status column,"Delivered",Profit column).
  4. On Time means 15 minutes or less, the target in this data. Late means more than 15.

Write these rules down. If two people analyse the same file with different rules, they get different answers, and both look right.

6. The data dictionary

A data dictionary is a small table with one line per column: name, meaning, type and an example. It is the first thing to make, and the first thing to show your manager.

Column Meaning Type Example
Order_ID Unique ID of one order; BLK- = Blinkit, AMN- = Amazon Now Text BLK-2001
Quantity Units in the order Number 2
Cost_Price What the company pays for one unit Number (Rs) 57
Selling_Price What the customer pays for one unit Number (Rs) 71
Status Delivered, Cancelled or Returned Text Delivered
Delivery_Time_mins Minutes from order to delivery; blank if cancelled Number 13
Delivery_Status On Time (15 minutes or less) or Late; blank if cancelled Text On Time

The full dictionary for all 17 columns is on the Data dictionary sheet of the practice file.

Classroom line

Being able to explain what the data is, is a skill in itself.

7. Business questions

Data is useful only when it answers a question that leads to a decision. For each question, note which columns answer it and what the business would do with the answer.

Business question Columns Decision it supports
Which products sell most? Product_Name, Quantity what to stock, where to change the price
How many orders are on time vs late? Delivery_Status is the delivery promise kept?
Are late orders the far ones? Distance_Km, Delivery_Status open a store nearer, or stop taking very far orders
Which products come back? Status, Product_Name check quality, reduce stock, cut rider cost
Does each city prefer different products? City, Product_Name stock each dark store for its own city
How do customers pay? Payment_Mode cash handling for riders
Where do we make or lose money? Revenue, Total_Cost, Profit, Status prices, discounts, which products to push
Question → columns → decision Business question Columns Decision Which products sell most? Product_Name, Quantity stock and price On time vs late? Delivery_Status is the promise kept? Are late orders the far ones? Distance_Km, Delivery_Status new store or distance limit Which products come back? Status, Product_Name quality, stock City-wise preference? City, Product_Name stock each dark store Where is the profit? Revenue, Total_Cost, Profit, Status prices, discounts

Figure 4. A business question points to the columns that answer it, and the answer points to a decision: stock, price, store location or delivery rules.

8. First answers with COUNTIF and COUNTIFS

You already know the functions for a first look. On the Questions sheet:

Question Formula Answer
Total orders =COUNTA(Order_ID column) 24
Delivered / Cancelled / Returned =COUNTIF(Status column,"Delivered") … 20 / 2 / 2
On Time / Late =COUNTIF(Delivery_Status column,"Late") … 14 / 8
Late and 3 km or more =COUNTIFS(Delivery_Status column,"Late",Distance column,">=3") 7
On Time and 3 km or more same, with "On Time" 0
COD / Online =COUNTIF(Payment_Mode column,"COD") … 9 / 15
Milk orders: local brand / national brand =COUNTIF(Product_Name column,"Milk 500 ml (local brand)") … 6 / 2
Blinkit / Amazon Now =COUNTIF(Platform column,"Blinkit") … 16 / 8

What the sample already says:

  1. Distance and lateness go together. Seven of the eight late orders were 3 km or farther, and no order from 3 km or more arrived on time. The one other late order was 2.9 km away. A store closer to those areas, or a distance limit, would fix most delays. Late deliveries matter: a customer who waits 22 minutes for milk simply switches to another app next time.
  2. Milk is the top product, and the local brand sells more (6 orders vs 2). By city: Pune 2, Kolhapur 2, Solapur 1, Sambhaji Nagar 1 for the local brand; Nashik and Nagpur ordered only the national brand.
  3. Two returns (bananas and curd): small numbers, but each one costs a rider trip and the product.

With 24 rows these are hints, not conclusions. On a full month of data the same formulas give answers you can act on. The Blinkit Maharashtra final project takes this kind of file all the way to a dashboard.

Late orders are the far ones On Time (14) Late (8) 1 km 2 km 3 km 4 km 5 km 2.6 km max 2.9 Late and 3 km or more: 7. On Time and 3 km or more: 0. =COUNTIFS(Delivery_Status,"Late",Distance_Km,">=3")

Figure 5. Every order on a distance line. On-time orders (green) are all within 2.6 km; late orders (red) start at 2.9 km.

Ravindra Bagale's Tip

Before your first formula on any new dataset, write five lines: what one row is, how many rows and columns, the date range, the rules (which status counts as a sale, what "late" means) and three business questions. Those five lines are also your answer when an interviewer says "explain your project".

9. Explaining it in an interview

"Explain your project" is the most common interview question for analysts. Use the same order as this chapter:

  1. The business: quick commerce, delivery in 10 to 15 minutes from dark stores.
  2. The data: one row = one order; 17 columns in seven groups.
  3. The rules: revenue only for delivered orders; late = more than 15 minutes.
  4. The questions: best-selling products, late vs distance, returns, city preferences, profit.
  5. One finding and one decision: late orders are the far ones, so a nearer store or a distance limit.

Classroom line

Don't talk about it as if it were a burden; speak with energy.

10. The whole chapter on one page

  1. Understand the data before you write a formula.
  2. Decide what one row is, and count rows and columns.
  3. Group the columns: which order, where, what, money, payment, outcome, speed.
  4. Cost_Price and Selling_Price are per unit; Revenue, Total_Cost and Profit are per order.
  5. Write the rules: which status counts, what is late.
  6. Keep a data dictionary.
  7. Every business question → columns → decision.
  8. COUNTIF, COUNTIFS and SUMIF give the first answers.

11. Try at home

Try at home

Task 1

  1. Count the late orders in each city with COUNTIFS. (Pune 2, Solapur 0, Nashik 2, Sambhaji Nagar 2, Kolhapur 0, Nagpur 2.)

Task 2

  1. Add up the profit of delivered orders with SUMIF (447). Then add up the profit column without any rule. Why is the second number bigger?

Task 3

  1. Write three more business questions for this data and, for each, the columns that answer it.

Task 4

  1. Explain the dataset in five sentences, out loud, as if to an interviewer.

Recap

In short

  1. Formulas come after understanding.
  2. One row = one order; know the size and the period.
  3. Per-unit prices × quantity = revenue and cost.
  4. Rules first: delivered only, 15-minute target.
  5. Questions lead to decisions: stock, price, store location, delivery.
  6. All data here is made up for practice.

Quick revision cards

Quick revision

Revise this chapter in two minutes. Save the cards on your phone or download them as PNG files. All names and numbers are made up for practice.

Revision card: Understand an orders dataset before analysis. Example data table with What is one row?: one order; Revenue of order O1: ₹100; Delivered within 15 minutes: 1.
Card 1 of 2: Understand an orders dataset before analysis · Download PNG
Revision card: From question to decision. Covers Which products sell most?; Is the price right?; Which city orders most?; Are orders on time?. With a photo of Ravindra Bagale and a short tip.
Card 2 of 2: From question to decision · Download PNG

More cards for every chapter: Ravi Sir's Cards.