Case 030Forecasting and scenariosWarm up

A D2C skin care brand adds 10,000 new customers a month, and each cohort keeps reordering at a declining rate. What is revenue in month 6?

The situation

Ksheera Skin Care sells online, direct to consumers. It acquires 10,000 new customers every month, and each places a first order in the month it is acquired. Of each month's new customers, 40% order again in their second month, 25% in their third, and 20% in every month after that. Each active customer places one order a month, and the average order is worth Rs 700.

The brand launched in month 1. The finance lead wants a revenue forecast built from cohorts, not from a growth rate.

Your task

What is Ksheera's revenue in month 6, and what does the build tell you that a simple growth rate would not?

Quick check

In month 6, how many orders does Ksheera receive?

Worked solution

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

30-second answerThe answer to give first

Month 6 revenue is Rs 1.575 crore, from 22,500 orders at Rs 700. Ten thousand are first orders; the other 12,500 come from older cohorts still reordering: 4,000 from month 5, 2,500 from month 4 and 2,000 from each of the three before. Repeat business is already 56% of orders, which a forecast built from new customers alone would miss entirely.

Step 1Why build revenue from cohorts at all?

Think of a gym. Its January revenue is not just the people who joined in January; it is everyone from every earlier month who is still turning up. A cohortA group of customers acquired in the same period, tracked together over time. build adds up, for each calendar month, every earlier group of customers who are still buying. It forces you to state how many stay, which is the number that decides whether a consumer brand is a business or a leaky bucket.

Step 2How do you lay out the grid?

Put cohorts down the side and calendar months across the top. Each cohort starts on the diagonal with 10,000 orders and then follows the same retention curve: 4,000, then 2,500, then 2,000 a month for as long as the 20% holds. Month 6 revenue is simply the sum of the month 6 column, times the order value. Every cell in that column is a different age of customer, which is why one retention curve generates the whole table.

Six cohorts of 10,000 customers each, reordering at 40%, 25% and then 20% a month, place 10,000, 14,000, 16,500, 18,500, 20,500, 22,500 orders in months 1 to 6, so month 6 revenue is 22,500 orders at Rs 700, or Rs 1.575 crore.

Month	New orders	Repeat orders	Total orders	Revenue, Rs lakh
1	10,000	0	10,000	70.0
2	10,000	4,000	14,000	98.0
3	10,000	6,500	16,500	115.5
4	10,000	8,500	18,500	129.5
5	10,000	10,500	20,500	143.5
6	10,000	12,500	22,500	157.5

New orders stay flat at 10,000 a month while repeat orders build from zero to 12,500, so revenue more than doubles from Rs 70 lakh in month 1 to Rs 157.5 lakh in month 6 with no change in acquisition.

Step 3What does the build tell you that a growth rate would not?

Revenue is growing while acquisition is flat, and all of that growth is repeat orders. A growth-rate forecast would read 125% growth over five months as momentum and extrapolate it. The cohort build shows the growth slowing by construction: each new month adds one more 2,000-order layer, so month 12 would bring about 34,500 orders, not a compounding curve.

It also shows where the forecast is fragile. The 20% that reorders every month forever is the assumption doing the most work, because it is the layer that keeps stacking. Real cohorts usually keep decaying, so a careful analyst would test 20% falling to 15% after month 6 and show how much lower month 12 lands. Say that limit out loud: the cohort method is only as good as the retention curve fed into it.

Where candidates lose it

The most common slip is multiplying 10,000 by six months and the retention rates in some blended way, or counting only new customers and reporting Rs 70 lakh. Draw the grid and add one column.

The second is applying the 40% to everyone ever acquired rather than only to last month's cohort. Each retention rate belongs to one age of customer, which is why the grid has a diagonal structure.

What the interviewer asks next

If acquisition costs Rs 1,200 per customer and gross margin is 60%, how many months does a cohort take to pay back?
How would the month 6 number change if the 20% decayed by a point a month?
What would you ask the brand for to test whether 40% second-month reorder is realistic?

Company names and figures are illustrative.