Unit economics: does each sale make money?

Unit economics means looking at one small piece of the business and asking: after I count every cost tied to this piece, do I keep any money? The piece is called a unit. For a food delivery app the unit is one order. For a software company it is one paying customer. For a lending app it is one loan. For a cab app it is one ride.

The idea matters because a company can show rising sales and still be broken. If a startup sells a ₹500 product and spends ₹600 to make, ship and market it, every extra sale adds ₹100 to the loss. Investors read unit economics to find out whether a business gets healthier or sicker as it grows.

Start with the revenue from one unit, which is the money the company actually keeps from that sale. Then subtract the costs that exist only because the sale happened. These are called variable costs, because they rise and fall with the number of units.

For a typical online order the variable costs include the cost of the goods, packaging, payment gateway fees, delivery, returns and refunds, customer support time, and any discount or coupon given. What is left is the contribution margin: the money that each unit contributes towards the fixed costs of the company, such as salaries, rent and technology. If contribution margin is positive, the unit makes money before fixed costs. If it is negative, no amount of volume will fix it by itself.

• Revenue per unit: what the company keeps, after discounts and taxes like GST are taken out
• Variable costs per unit: goods, delivery, payment fees, returns, support
• Contribution margin per unit: revenue minus variable costs
• Acquisition cost: what it took to win that customer in the first place

Many consumer startups in India have chosen to lose money per order on purpose in the early years. Heavy discounts, free delivery and cash-back offers pull customers in, and the company hopes they will stay and order more later. Investors have funded this approach when they believed that the market was large and that the loss per unit would shrink with scale.

It can shrink for real reasons. Delivery partners handle more orders per hour once a city is dense. Suppliers give better rates at higher volumes. Customers who come back cost nothing to win a second time. But it can also fail to shrink, if customers only buy when there is a discount. The honest test is whether the loss per unit has gone down over time in the company's own numbers.

Unit economics looks different at different moments. On a customer's first order the company has just paid to acquire them, so the numbers are usually poor. By the tenth order the acquisition cost is spread across many sales, and the unit often looks healthy.

This is why a careful reader asks two questions. One: is a single order profitable on its own, before marketing? Two: does a customer, over their whole life with the company, bring in more than it cost to win them? The second question is the subject of the guide on CAC and LTV. Good analysis keeps the two apart, so a profitable order is not hidden by heavy marketing and a cheap customer is not confused with a profitable one.

The shape of the calculation changes with the model, but the logic stays the same. A software-as-a-service company earns a monthly fee and its variable cost per customer is small, mostly cloud hosting and support, so its margin per customer is often high. A marketplace earns only a commission on each deal, called the take rate, so the money per unit is a small slice of the sale value. A lender earns interest but must subtract the cost of funds, collection costs and loans that are never repaid.

When you read a startup story, ask what the unit is, what the company earns on it, and what it pays out on it. Those three answers tell you most of what you need to know about whether the model can work.

The first mistake is leaving out costs. Founders sometimes count the cost of the goods but forget returns, refunds, or payment fees. The second is using the wrong revenue figure: gross sales value instead of what the company keeps. The third is averaging everything. An average can hide a profitable group of customers and a very unprofitable group.

The fourth mistake is treating fixed costs as variable, or the reverse. Rent for a warehouse is fixed in the short term, but extra delivery cost per order is variable. Mixing them up can make a business look better or worse than it is. When numbers look surprisingly good, check which costs were left out, and always check the time period they cover.

Unit economics is a way of reading a business, not a legal test. Different companies define the unit and the costs differently, so compare like with like.

Look for the numbers a company publishes: revenue, cost of goods, delivery and marketing spend, and the number of orders or customers. Divide each by the number of orders to get a rough per-order view. Annual reports and draft prospectuses often list these costs, and a fall in cost per order over several years is a good sign.

You will rarely get a perfect answer from outside, so aim for a direction, not a decimal. Is the loss per order shrinking? Do repeat customers form a growing share? If both answers are yes, the model is probably improving. As this is a reading aid and not advice, check important decisions with a professional.