Case 077Earnings, models and KPIsWarm up

Build a three-line revenue model for Drovi Quick Commerce: 8 lakh orders a day, an average order value of Rs 450 and a take rate of 18%. What are annual GMV and net revenue, and which input moves revenue most for a 10% change?

Balyasny Asset ManagementNew York · 2026

The situation

Drovi Quick Commerce delivers groceries in under fifteen minutes from dark stores in twelve cities. It handles 8 lakh orders a day at an average order value of Rs 450. Drovi does not own the stock it sells; its net revenue is its take rate, 18% of gross merchandise value (GMV), made up of seller commissions, delivery fees and advertising income.

The portfolio manager wants a model that fits on three lines, because the analyst covers twenty names and cannot maintain a 40-tab spreadsheet for each.

Your task

What are annual GMV and net revenue, and which input moves revenue most for a 10% change?

Quick check

A 10% rise in which single input raises net revenue the most?

Worked solution

Try it on paper, then open one step at a time.

30-second answerThe answer to give first

GMV is about Rs 13,140 crore a year and net revenue about Rs 2,365 crore. 8 lakh orders a day is 29.2 crore orders a year; at Rs 450 each that is Rs 13,140 crore of GMV, and 18% of it is net revenue. A 10% change in any one input moves revenue by the same 10%, about Rs 237 crore, because the model multiplies. The real question is which input you trust least.

Step 1How do you set up a model that fits on three lines?

Think of a tea stall. Cups sold a day times price per cup is the day's sales; if the stall worked on commission for a supplier, its income would be a cut of that. Drovi's revenue is the same chain: how many orders, how big each basket, and what share of the basket Drovi keeps. Each line is a number you can check and argue about on its own, which is why a covering analyst builds it this way.

Line	Input	Result
Orders a day	8,00,000
Orders a year	x 365 days	29.2 crore
GMV	x Rs 450 average order value	Rs 13,140 crore
Net revenue	x 18% take rate	Rs 2,365.2 crore

8 lakh orders a day for a year at Rs 450 each gives Rs 13,140 crore of GMV, and Drovi keeps 18% of it as net revenue, Rs 2,365.2 crore.

Watch the units, because this is where the arithmetic slips. 8 lakh times Rs 450 is Rs 36 crore of GMV a day, not Rs 360 crore. Multiply by 365 for the year before you apply the take rate, and write the unit beside every number. A take rateThe share of the value of goods sold on a platform that the platform keeps as its own revenue. is a revenue line, not a margin: Drovi still pays for riders and dark stores out of its Rs 2,365 crore.

Drovi's 29.2 crore orders a year at Rs 450 make Rs 13,140 crore of GMV and Rs 2,365 crore of net revenue, and a 10% rise in orders, order value or take rate each adds the same Rs 237 crore.

Step 2Why does every input move revenue by the same amount?

Because the model is a product. When revenue is A times B times C, a 10% change in any one of them is a 10% change in revenue. Move all three by 10% together and revenue rises 33.1%, not 30%, since the gains compound on each other. So the interviewer's question has a trick inside it: on the numbers, no input has more leverage than another.

The relationship

R = O \times 365 \times A \times t \qquad \frac{\Delta R}{R} \approx \frac{\Delta O}{O} + \frac{\Delta A}{A} + \frac{\Delta t}{t}

R	annual net revenue
O	orders a day
A	average order value
t	take rate

What it says in wordsIn a multiplicative model the percentage changes in the inputs add up, roughly, to the percentage change in revenue.

Step 3So which input deserves the argument?

The one with the widest honest range, and the ones that push on each other. Order value is fairly stable, perhaps Rs 420 to 480. Take rate is set by Drovi but capped by what sellers and customers will bear. Orders depend on new dark stores, retention and competitors, and could plausibly sit anywhere from 6 to 10 lakh a day. Orders carry about Rs 1,183 crore of revenue range against Rs 315 crore for order value, so that is where the research time goes.

Across honest ranges, orders move net revenue from Rs 1,774 crore to Rs 2,956 crore, far more than take rate or order value, even though all three inputs have the same elasticity.

Then mention the link between lines. If Drovi lifts its take rate 10% by charging higher delivery fees and loses 4% of orders, revenue rises to about Rs 2,498 crore, 5.6%, not 10%. A three-line model is honest only if you remember the lines are not independent.

Where candidates lose it

The frequent slip is the units: 8 lakh times 450 read as Rs 360 crore a day, or the 365 forgotten, which puts GMV out by a factor of ten or of 365. Say the unit at every step out loud.

The second is naming take rate as the biggest lever because it feels like pricing power. In this model every input has the same lever length; the useful answer is about which number is least certain and which ones move together.

What the interviewer asks next

Contribution per order is Rs 12 after rider and store costs. What does Drovi earn at the contribution level?
Which of the three inputs would you check first against channel data, and how?
How would you extend this to five lines without making it a 40-tab model?

Asked at Balyasny Asset Management, Equity Research, New York, 2026 (Wall Street Oasis): Did initial phone screen with BD, then model test, intelligence test.

Company names and figures are illustrative.