Unit economics and subscription businesses
Whether a business makes money on each order or customer: contribution per order, customer acquisition cost (CAC), lifetime value (LTV), churn, and cohorts.
Key takeaways
- Before asking how to grow, check that each order or customer makes money; then grow the units that do.
- Build the economics of one unit first (order or customer), find the lines that make it negative, and size the levers before discussing growth.
- Unit-economics questions are common in technology, consumer, and investor-focused cases, including private-equity and venture-capital style cases.
- The strong answer checks that each unit makes money before scaling. The weak one scales losses.
- Why a generic structure scores lower: Pointing to fast revenue growth without checking whether each unit makes money approves more spend that loses more money.
What this case type is and when it shows up
Unit-economics cases ask whether a fast-growing business makes money on each unit: each order for a delivery app, each customer for a subscription service. They are common for technology, e-commerce, and consumer subscription businesses, and in private-equity and venture-capital questions. Growth hides losses: if each order loses money, growing faster loses money faster.
Key idea
Before asking how to grow, check that each order or customer makes money; then grow the units that do.
The underlying theory, in plain language
Contribution per order is what one order adds after all costs that vary with orders: product margin minus delivery, packaging, payment fees, discounts, and a share of local running costs such as a small warehouse (a dark store, in quick commerce). Costs that are fixed per store fall per order as orders grow, so volume matters.
For subscriptions, customer acquisition cost (CAC) is marketing and sales spend divided by new customers won. Churn is the share of customers who leave each month. Average customer lifetime is roughly 1 divided by monthly churn, so lifetime value (LTV) is monthly contribution per customer divided by monthly churn (a simple, undiscounted version; discounting lowers it).
Two tests are common. LTV/CAC: many investors look for about 3 or more, meaning a customer is worth three times what it costs to win. CAC payback: months of contribution needed to earn back CAC; shorter is safer.
A cohort is a group of customers who joined in the same month. Tracking how many remain month by month shows churn more reliably than company-wide averages, because new customers can hide old ones leaving.
What the prompts sound like, from simple to hard
- Simple: does a Dubai food-delivery order make money.
- Medium: can a quick-commerce grocery app in India reach positive contribution per order.
- Hard: should a European software company raise marketing spend, given its churn and CAC.
Build the structure from the goal
Three moves that give you the structure
- 1Start from the decision. Does each order or customer make money, and if not, what would make it, before the client spends more to grow?
- 2Write the maths of the goal. Per order: contribution = order value x margin, minus delivery, packaging, payment, and discounts, minus local fixed costs / orders. Per customer: lifetime value = monthly contribution x months a customer stays (1 / monthly churn), compared with the cost to acquire the customer (CAC).
- 3Let the business pick the branches. The model tells you which unit to use. Delivery and marketplace businesses live on contribution per order, so order size and orders per rider or store matter. Subscription businesses live on churn and CAC. Hardware sold with a subscription needs both.
- Is a customer worth more than it costs to win?
- Contribution per box: price minus food, packing, delivery
- Key: Boxes per customer before they cancel
- Cost to acquire: first-box discounts and advertising
- Levers: smaller first-box discounts, more meals per box
This comes from lifetime value = contribution x lifetime, compared with CAC. Meal kits lose many customers when the discount ends, so lifetime leads.
- Contribution per ride
- Fare x commission kept
- Key: Driver incentives per ride
- Rider discounts per ride
- Rides per driver hour as the city fills
Same maths, per order. In a new city, incentives to drivers and riders swallow the commission, so they lead.
Pointing to fast revenue growth without checking whether each unit makes money approves more spend that loses more money. Interviewers score the unit first.
Crack any case in five movesFinding and narrowing the real problem
Key idea
Build the economics of one unit first (order or customer), find the lines that make it negative, and size the levers before discussing growth.
- Unit economics
- Key: Per order
- Order value x product margin
- Minus delivery, packaging, payment, discounts
- Minus local fixed costs / orders
- Per customer
- Monthly contribution
- Churn and lifetime
- LTV versus CAC
- Levers
- Bigger orders
- More orders per store
- Lower churn
- Cheaper acquisition
Orders for transaction businesses, customers for subscriptions.
Names you may hear, kept as questions that fall out of the goal maths
- Contribution waterfall: Walking from order value down, which line turns each order negative? Where it stops helping: Shared costs need a fair allocation rule.
- LTV/CAC and CAC payback: Is a customer worth more than it costs to win, and how many months until the cost is earned back? Where it stops helping: LTV rests on a churn estimate; young companies have little history.
- Cohort view: How does each monthly group of new customers behave over time: do they stay longer or leave sooner than earlier groups? Where it stops helping: Needs enough months of data.
Methods for solving this type
- Build contribution per order or per customer
- Find the lines that make it negative
- Size levers: order value, density (orders per store), churn, CAC
- Check LTV/CAC and payback
- Grow only the units that make money
The math patterns it relies on
- Contribution = order value x margin minus variable costs minus fixed cost per order
- Fixed cost per order = daily fixed cost / orders per day
- LTV = monthly contribution / monthly churn
- CAC = acquisition spend / new customers
- CAC payback = CAC / monthly contribution
Worked cases
Worked case
Quick commerce in India: does an order make money?
The prompt
A quick-commerce app in India delivers groceries in about 10 minutes from small local warehouses (dark stores). The exhibit shows the economics of an average order. Is it profitable, and what would it take?
Interviewer-led: the interviewer shows the order waterfall and asks the questions in order.
Clarifying questions, with the interviewer's answers
- Should contribution include the dark store's running costs?Answer: Yes, include all costs that vary with orders plus the store's own running costs; exclude head-office costs.
- How many orders does a typical store handle, and what does it cost to run?Answer: About 2,000 orders a day; about INR 60,000 a day to run.
- How does the app earn money on an order?Answer: It runs a marketplace, so it earns a take rate (commissions, fees, and ad income from sellers) of about 18 percent of order value; the average order is about INR 500.
A hypothesis to say out loud: Quick-commerce stores have high fixed costs, so my hypothesis is that orders lose money at today's volume and order size, and that more orders per store plus bigger baskets can fix it.
The structure
- Contribution per order, then the leversThis comes from contribution per order = take-rate income minus rider, packaging, payment, and discount costs, minus dark-store cost / orders.
- Take-rate income per order
- Variable costs: rider, packaging and payment, discounts
- Key: Dark-store cost per order (falls with volume)
- Levers: order value and orders per store
The exhibit
Waterfall chart: Economics of an average order today (INR, illustrative). Values in INR per order. Take-rate income, total: 90; Rider cost, change: -45; Dark-store cost, change: -30; Packaging and payment, change: -10; Discounts, change: -15; Contribution, total: -10.
Working it through
1. Take-rate income
INR 500 order value at an 18 percent take rate.
Take-rate income per order (INR):500 × 0.18 = 902. Dark-store cost per order
INR 60,000 a day spread over 2,000 orders.
Store cost per order (INR):60,000 ÷ 2,000 = 303. Contribution per order today
Take-rate income minus rider (INR 45), store (INR 30), packaging and payment (INR 10), and discounts (INR 15).
Contribution per order (INR):500 × 0.18 - 45 - 60,000 ÷ 2,000 - 10 - 15 = -104. Break-even orders per store
Before store costs, each order contributes INR 20. The store needs this many orders a day to cover INR 60,000.
Break-even orders a day:60,000 ÷ (500 × 0.18 - 45 - 10 - 15) = 3,0005. Store cost per order at 3,000 orders
More orders spread the same store cost.
Store cost per order (INR):60,000 ÷ 3,000 = 206. Both levers together
At 3,000 orders a day and an average order of INR 600.
Contribution per order (INR):600 × 0.18 - 45 - 60,000 ÷ 3,000 - 10 - 15 = 18
What the exhibit shows
Each order loses about INR 10. The rider and the dark store take most of the margin, and the store cost falls as orders per store rise.
The recommendation
Each order loses about INR 10 today, so growth should focus on density and basket size, not on opening more stores. First, store costs are fixed, so at 3,000 orders a day the store breaks even at today's basket. Second, raising the average order to INR 600, through minimum order values and adding higher-value items, lifts contribution further, to about INR 18 per order at 3,000 orders. Third, discounts of INR 15 per order are larger than today's loss, so cutting them for loyal customers is the quickest lever. Pause new store openings until existing stores pass 3,000 orders a day.
Risks: Cutting discounts may reduce orders; Higher minimum orders may push small-basket customers to rivals; Rider costs may rise with wages.
Next steps: Track contribution per order by store weekly; Test a minimum order value in two cities; Reduce discounts for customers who order at least weekly.
A strong candidate
Built contribution line by line, saw that store cost is fixed, found break-even orders per store, and sized both levers.
A weak candidate
Recommended more marketing to grow orders across more stores, adding stores that each lose money.
Worked case
A European software company: spend more on marketing?
The prompt
A German software company sells a subscription to small businesses across Europe. The board wants to double marketing spend to grow faster. Should it?
Candidate-led: you ask for the data and drive; the interviewer answers what you ask.
Clarifying questions, with the interviewer's answers
- What does a customer pay, and what is the gross margin?Answer: EUR 100 a month, with an 80 percent gross margin after hosting and support.
- How many customers leave each month?Answer: About 2 percent.
- What do we spend to win customers?Answer: About EUR 1.2 million a month on marketing and sales, winning about 400 new customers a month.
A hypothesis to say out loud: Software margins are high, so my hypothesis is that the problem, if any, is churn or acquisition cost, and that raising marketing spend only makes sense if a customer is worth several times what it costs to win.
The structure
- Is a customer worth more than it costs to win?This comes from lifetime value = monthly contribution / monthly churn, compared with the cost to acquire a customer.
- Monthly contribution per customer
- Lifetime value from churn
- Acquisition cost (CAC)
- Key: LTV/CAC and payback
Working it through
1. Monthly contribution
Candidate: "At EUR 100 a month and an 80 percent margin, each customer contributes:"
Contribution (EUR a month):100 × 0.8 = 802. Lifetime value
Candidate: "With 2 percent monthly churn, an average customer stays about 50 months, so LTV is contribution divided by churn."
LTV (EUR):80 ÷ 0.02 = 4,0003. Acquisition cost
EUR 1.2 million a month for 400 new customers.
CAC (EUR):1,200,000 ÷ 400 = 3,0004. LTV/CAC
Candidate: "That is far below the 3 that investors often look for." Interviewer: "The board says revenue is growing fast." Candidate: "It is, but each customer barely earns back its cost."
LTV/CAC:4,000 ÷ 3,000 = 1.335. CAC payback
Months of contribution to earn back CAC: over three years.
CAC payback (months):3,000 ÷ 80 = 37.56. Fix first, then grow
Candidate: "Can churn and CAC be improved?" Interviewer: "Better onboarding could halve churn to 1 percent, and shifting spend to partner channels could cut CAC to EUR 2,500."
LTV/CAC after fixes:(80 ÷ 0.01) ÷ 2,500 = 3.2
The recommendation
Do not double marketing spend yet. First, a customer is worth about EUR 4,000 but costs EUR 3,000 to win, an LTV/CAC of about 1.3, so more spending buys growth that barely pays back. Second, CAC payback is about 37.5 months, which ties up a lot of cash. Third, halving churn to 1 percent and cutting CAC to EUR 2,500 would lift LTV/CAC to about 3.2. Fix onboarding and channels first, measure churn by monthly cohort, and raise spend once new cohorts show the lower churn.
Risks: Churn improvements may take several cohorts to prove; Partner channels may not scale at the lower CAC.
Next steps: Build a cohort table of retention by month of joining; Redesign onboarding for the first 60 days; Test partner channels with 20 percent of the budget.
A strong candidate
Asked for margin, churn, and spend, computed LTV from contribution, compared it with CAC, and showed which fixes would justify more spend.
A weak candidate
Agreed to double spend because revenue growth looked strong, without checking what a customer is worth.
Prompt: "Should we spend more to grow faster?"
Weaker answer
Points to fast revenue growth and approves more spend without checking what each customer is worth.
Stronger answer
Calculates contribution, LTV, CAC, and payback, finds LTV/CAC of about 1.3, and fixes churn and CAC before scaling spend.
Why the stronger answer wins: The strong answer checks that each unit makes money before scaling. The weak one scales losses.
Common mistakes, traps, and curveballs
- Pushing growth while each order loses money
- Leaving discounts or delivery out of contribution
- Computing LTV from revenue instead of contribution
- Using company-wide churn that new customers hide
- Ignoring that fixed costs per order fall with volume
Unit-economics questions are common in technology, consumer, and investor-focused cases, including private-equity and venture-capital style cases. Expect to build contribution per order or LTV/CAC on the page. Formats differ by office and change over time, so check the current process for your target office.
Practice
A Singapore meal-kit subscription earns SGD 30 of contribution a month per customer, and 5 percent of customers leave each month. What is the LTV, in SGD?
A UK app spends GBP 500,000 on marketing in a month and wins 2,500 customers. What is its CAC, in GBP?
Of 1,000 customers who joined a Dubai streaming service in January, 900 remain after one month. What is the first-month churn, in percent?
A delivery app loses money on each order. What happens if it doubles orders with the same economics per order?
LTV is EUR 900 and CAC is EUR 300. What is LTV/CAC, and is it healthy by the common rule of thumb?
Why use contribution, not revenue, to calculate LTV?
Check that each order or customer makes money before chasing growth, and fix the lines that make it negative.
Sources for this lesson (1)
- Recognized public explanations of case-interview concepts and terms
My notes on this lesson
0 of 5,000 characters. Saves automatically.
Try the 6 remaining checks and drills above to complete this lesson (0 of 6 done).
Spotted something wrong or out of date? Report a mistake. We check every report and correct the page.