Unit economics is the profit or loss on one unit of what a business sells, after the costs that unit directly causes, and before the overheads that exist whether or not it sells anything. If that number is negative, growth makes the hole bigger. If it is positive, the business can eventually cover its fixed costs and the money it spends to acquire customers. Investors ask about it because it separates a company that is early from one that is broken. This guide is informational, not investment advice.
What a “unit” is, and how to pick yours
The unit is the thing whose revenue and costs you can measure on their own: one order for a delivery platform or an online store, one subscriber for a SaaS company, one ride for a mobility app, one seat for a coaching class. Pick the one that repeats most often and that your costs attach to most cleanly.
Many businesses need two layers. A food-delivery company measures the order (does each delivery earn more than it costs?) and the customer (does a customer order often enough, for long enough, to repay what it cost to acquire them?). Order-level numbers tell you whether the product works; customer-level numbers tell you whether the marketing works.
Contribution margin, step by step
Managerial accounting defines contribution margin as “the amount by which a product’s selling price exceeds its total variable cost per unit”, with variable costs being “those costs that vary per unit of production” and fixed costs the ones that “remain constant within a relevant range”, according to OpenStax’s Principles of Accounting. Divide the per-unit contribution by the price and you get the contribution margin ratio.
For a real order the list of variable costs is longer than most founders expect:
- Price the customer pays, excluding GST, which you collect for the government and never keep.
- Discounts and coupons you fund. Count them as a cost, not as marketing that lives somewhere else.
- Cost of goods, or, for a marketplace, the payout to the restaurant, seller or driver.
- Delivery and packaging, including the part of delivery you subsidise when the customer pays less than the courier charges.
- Payment fees: card MDR, gateway charges, cash-on-delivery handling. Our guide to MDR explains who pays what on each payment type.
- Returns, refunds and customer support, averaged across all orders, including the ones that go wrong.
- Other per-order costs: SMS and OTPs, insurance, onboarding of a delivery partner, fraud losses.
What is left is contribution per unit. Marketing, salaries, rent and software stay out; they are what the contribution has to pay for.
Gross margin, contribution margin and EBITDA are three different lines
Gross margin is price minus cost of goods only. Contribution margin also removes the other variable costs above. EBITDA (earnings before interest, tax, depreciation and amortisation) then removes fixed operating costs such as salaries, rent and marketing. Each line points to a different problem: healthy gross margin with negative contribution means delivery, discounts or returns are eating the product margin; healthy contribution with negative EBITDA means the fixed base is too big for the volume. (A separate Pointales guide to EBITDA is planned.)
CAC, LTV, payback and the 3:1 rule
Customer acquisition cost (CAC) is sales and marketing spend in a period divided by the new customers won in that period. Andreessen Horowitz’s 16 Startup Metrics says it should be “the full cost of acquiring users, stated on a per user basis”, and distinguishes blended CAC (all customers, including organic) from paid CAC (only customers from paid channels). The authors consider paid CAC “more important than blended CAC in evaluating the viability of a business”.
Lifetime value (LTV) is, in the same guide, “the present value of the future net profit from the customer over the duration of the relationship”. The common mistake is to compute it on revenue. The useful version multiplies contribution per customer per month by the expected lifetime, and lifetime is the inverse of churn: at 3% monthly churn, “the Customer Lifetime will be 1/0.03 which is 33 months”, as David Skok’s SaaS metrics definitions put it. Retention is the same idea from the other side: the share of a cohort still active after a given number of months.
Payback period is how many months of contribution it takes to recover CAC. Skok’s SaaS Metrics 2.0 offers two guidelines: the best SaaS businesses have “a LTV to CAC ratio that is higher than 3”, and “profitability is anemic if the time to recover CAC extends beyond 12 months”, with the best recovering it in five to seven months. He adds that “these are only guidelines”. The limits matter: LTV is a forecast built on churn you have only partly observed, 3:1 was written for subscription software, and a ratio can look fine while payback is so slow that the company runs out of cash reaching it. Payback protects you from that, because it is measured in cash and months.
Why blended numbers mislead
A company-wide average mixes customers acquired in different years, through different channels, at different prices. If old, loyal customers are subsidising expensive new ones, the average looks fine while the business quietly gets worse. The fix is cohort analysis: group customers by the month they joined and track each group’s orders and contribution over time. a16z’s 16 More Startup Metrics lists the two good signs: retention that stabilises after six or twelve months, and newer cohorts performing better than older ones. It also warns against cumulative charts: “Metrics that should never be reported in a cumulative fashion include revenue, new users, and bookings.”
Y Combinator’s Anu Hariharan makes the same point for transactional businesses: “user retention is a cohort metric, it’s not one number”, and “it’s really important to measure your cohort retention on a monthly basis”, per YC’s Startup School recap. The same applies to CAC: compute it per channel, because the average of a cheap referral channel and an expensive paid one describes neither.
A worked example (hypothetical)
Every number below is invented and rounded; it is a template, not a real company. Take an online skincare brand that sells a ₹1,000 bundle (price excluding GST).
| Per order | ₹ |
|---|---|
| Price (excluding GST) | 1,000 |
| Less: average coupon discount | (100) |
| Less: cost of goods | (350) |
| Less: packaging | (30) |
| Less: shipping (net of what the customer pays) | (80) |
| Less: payment gateway fee, about 2% | (20) |
| Less: returns, refunds and support, averaged | (40) |
| Contribution per order | 380 |
Contribution is ₹380 on ₹1,000, a 38% contribution margin ratio, even though gross margin (price minus goods) was 65%. If the brand spends ₹2 lakh a month on ads and wins 500 new customers, paid CAC is ₹400. The first order does not pay for itself: the customer must come back once before the brand is ahead. If the average customer places three orders over their life, LTV on contribution is ₹1,140 and LTV:CAC is 2.9. If fixed costs (two salaries, a warehouse, software) are ₹6 lakh a month, the brand needs ₹6,00,000 ÷ ₹380, about 1,580 orders a month, before it makes a rupee of operating profit. Double the refund line to ₹80 and break-even moves to about 1,765 orders. Small per-unit changes move the whole business.
A real example: food delivery per order
Zomato’s red herring prospectus of 6 July 2021 published a per-order waterfall for its food-delivery business, which remains one of the clearest disclosed in an Indian filing.
| Per order (₹) | FY2020 | FY2021 |
|---|---|---|
| Commission and other charges (restaurant commission, ads, other revenue) | 43.6 | 62.8 |
| Customer delivery charge | 15.3 | 27.0 |
| Less: delivery cost (payout to delivery partners) | (52.0) | (45.7) |
| Less: platform-funded discounts | (21.7) | (8.3) |
| Less: other variable costs (payment gateway, support, refunds, onboarding, insurance, SMS, COD handling) | (15.7) | (15.3) |
| Contribution profit / (loss) | (30.5) | 20.5 |
The prospectus notes that “costs associated with marketing, branding and other fixed operating costs are excluded.” In one year the order went from losing ₹30.5 to earning ₹20.5, mostly through higher commission and delivery fees and lower discounts. Scale had not fixed the economics; pricing had.
Five years on, the company (now Eternal Limited) reports the same idea as a percentage. In its results for April to June 2026, dated 22 July 2026, food delivery generated net order value (NOV, order value after all discounts) of ₹10,769 crore, contribution of ₹1,097 crore (10.2% of NOV) and adjusted EBITDA of ₹606 crore (5.6% of NOV). Its glossary defines contribution by deducting the same list of per-order costs as the 2021 waterfall: last-mile delivery, platform-funded discounts, payment gateway and COD charges, support, refunds and delivery-partner onboarding.
The quick-commerce arm, Blinkit, shows why the three margin lines matter. On NOV of ₹17,132 crore across 331 million orders, gross profit was ₹4,710 crore (27.5% of NOV), contribution ₹907 crore (5.3%) and adjusted EBITDA ₹102 crore (0.6%). On Pointales’ arithmetic that is roughly ₹142 of gross profit per order, ₹27 of contribution after dark-store, delivery, warehousing, wastage, packaging, payment and support costs, and about ₹3 of adjusted EBITDA. The letter also publishes the customer layer: on average 46% of a cohort still orders in the fourth quarter after acquisition, and a cohort’s NOV reaches 279% of its first-quarter spend by its twelfth quarter. Our guides to how quick commerce works and what a dark store is explain where those per-order costs come from.
Swiggy’s letter to shareholders of 30 July 2026 uses the same framework as a management target. Instamart’s contribution margin, measured against gross order value, bottomed at -5.6% in January to March 2025, reached break-even in May 2026 and was -0.2% for the June 2026 quarter. Management says it added ₹28 of contribution per order over five quarters, mainly by lifting adjusted revenue per order from ₹83 to ₹108, and needs roughly ₹30 more per order (a 5–6% margin) to reach adjusted EBITDA break-even: ₹10 from product mix, ₹10 from advertising, ₹5 from store density and automation, ₹5 from operating leverage. Because Swiggy measures against GOV and Eternal against NOV, the two percentages are not directly comparable, which is itself a lesson: always check the denominator.
What investors actually ask
Hariharan’s lecture lists what investors want by business model: for e-commerce, revenue growth, gross margin and paid CAC; for marketplaces, GMV, net revenue (the company’s cut, not the value transacted) and user retention; for SaaS, MRR, gross MRR churn and paid CAC. She notes that e-commerce founders often leave shipping and payment-processing costs out of gross margin. In practice the questions sound like this:
- What is contribution per unit today, and what was it six months ago?
- What is paid CAC by channel, and what happens to it when you double spend?
- Show me monthly cohorts: do they flatten, and are new cohorts better than old ones?
- How many months to pay back CAC, in cash?
- Which per-unit lines improve with scale, and which do not?
These numbers set the terms of the next round. A company whose contribution is turning positive can raise on its own schedule; one whose per-unit losses grow with volume is the kind that ends up negotiating a down round. Our pillar on how startup funding works in India covers what each stage looks for.
Common mistakes
- Ignoring returns and refunds. They are a per-order cost, averaged across every order, not an occasional surprise.
- Treating marketing as fixed. Performance marketing needed to win each order is a variable cost of that order; brand spend is overhead. Mixing them flatters contribution.
- Using revenue LTV. a16z calls LTV on revenue or gross margin a figure that “suggests a higher upper limit on what you can spend on customer acquisition” than is safe.
- Reporting blended CAC only. It hides a paid channel that has stopped working.
- Counting GMV as revenue. For a marketplace, revenue is the take, not the order value. a16z’s warning is blunt: “GMV does not equal revenue!”
- “We’ll fix it at scale.” Scale helps costs that are genuinely fixed and does nothing for costs that grow with orders (delivery, payment fees, discounts). Zomato’s turnaround came from pricing and fewer discounts; Swiggy’s plan names each rupee it expects from each lever. Vaguer versions of the claim deserve scepticism.
- Forgetting GST and the denominator. Compare margins on the same base, net of tax, and say which base you used.
The point: Pick the unit your costs attach to, subtract every cost that unit causes (discounts, goods, delivery, payment fees, refunds) and check the result is positive before you buy growth. Then test the customer layer: paid CAC by channel, contribution-based LTV, cash payback in months and monthly cohorts rather than one blended average. Zomato, Eternal and Swiggy’s own filings show the framework in use: per-order contribution that turned positive through pricing, and targets expressed in rupees per order.
Sources
- Principles of Accounting, Volume 2: Managerial Accounting, section 3.1 (contribution margin), OpenStax, Rice University, 2019
- 16 Startup Metrics, Jeff Jordan, Anu Hariharan, Frank Chen and Preethi Kasireddy, Andreessen Horowitz, 21 August 2015
- 16 More Startup Metrics, Andreessen Horowitz, 2015
- SaaS Metrics 2.0: A Guide to Measuring and Improving what Matters and SaaS Metrics 2.0: Detailed Definitions, David Skok, For Entrepreneurs
- Startup School Week 3 Recap: Anu Hariharan and Adora Cheung, Y Combinator, 18 September 2019
- Zomato Limited, Red Herring Prospectus dated 6 July 2021, page 142 (unit economics for Fiscals 2020 and 2021), hosted by BofA Securities India
- Eternal Limited, Shareholders’ Letter and results for Q1 FY27 (quarter ended 30 June 2026), 22 July 2026 (Annexure A quarterly disclosures and Annexure C glossary)
- Swiggy Limited, Letter to Shareholders for Q1 FY27 (quarter ended 30 June 2026), filed with NSE, 30 July 2026