A marketplace sells liquidity: the confidence that a buyer will find a seller within tolerable time and price. A diner finds a meal, a rider finds a driver, a shipper finds a truck. It holds no inventory. Its revenue is a take rate sliced from a much larger flow of GMV/GOV (gross merchandise/order value). Two consequences dominate the analysis. The first is the accounting. Reported revenue is a policy choice, so you always rebuild the picture from GMV, as you'll see in a moment. The second is the cost structure. The marginal match costs little once both sides are present, but getting both sides there is brutally expensive (discounts, delivery subsidies, driver incentives). So young marketplaces bleed through contribution, the very line that makes mature ones beautiful.
What is the difference between GMV and revenue?
GMV vs net revenue, the trap you must never fall into. Ind AS 115 / IFRS 15 / ASC 606 are aligned here, so you can learn the economics once and relax. An intermediary that is the principal reports revenue gross, meaning the full ticket, with goods cost in COGS. Being the principal means it controls the good or service before transfer. An intermediary that is an agent reports net, just its fee. The control indicators are primary responsibility for fulfilment, inventory risk, and discretion over pricing. Zomato reports commissions and fees, and the value of your biryani never touches its revenue line. A supermarket reports the full basket. Now run the same warehouse-and-riders quick-commerce operation both ways, with identical economics. It reports ~5× more revenue under a 1P (own-inventory) presentation than under 3P (marketplace). Worked Example 3 computes it. The consequences follow directly. Never compare price-to-sales across models. Never call a marketplace's growth "small" against a retailer's until you've converted both to GMV, or both to gross profit. And when a company flips presentation from 1P to 3P or back, restate history before you read trends. Amazon does both at once.
Advertisement
Marketplace take rate examples: Zomato and Uber
Zomato / Eternal (India). These figures are illustrative, modelled on ~FY25 disclosures. Pull the latest quarterly shareholder letter, because the disclosure quality is unusually good. Food-delivery GOV is ~₹38,000 cr and growing in the high teens. The take rate is ~20 to 21% of GOV, counting commissions + customer delivery/platform fees + ads. Food-delivery contribution margin is ~8% of GOV, and adjusted EBITDA is ~4 to 5% of GOV and climbing. Consolidated revenue is ~₹20,000 cr, and reported PAT has recently turned positive (~₹500+ cr). Meanwhile Blinkit, with GOV of ~₹28,000 cr growing >100%, reinvests everything. What should you take from it? First, the P&L is a stack of cohorts of cities at different maturities. Second, watch the take rate's composition. Raising it through ads and platform fees sticks better than raising commissions on angry restaurants. Third, the model's ceiling is the rake question. Charge too much and you invite disintermediation (direct ordering) or a rival, which is ONDC's explicit purpose. Bill Gurley's "A Rake Too Far" is the canonical essay, listed in the resources.
Uber (US). Illustrative ~2024. Gross bookings of ~$163bn turn into revenue of ~$44bn, a blended take of ~27%. Mobility runs at ~29 to 30% and delivery at ~19%. The segment-level take rates are in the 10-K, but part of the mobility figure is a gross-vs-net accounting artifact in some geographies, so read the revenue note before you trend it. Revenue then becomes operating income of ~$2.8bn, with FCF of ~$7bn. Uber had a decade of losses. Then the two-sided network matured, incentives fell away, and operating leverage arrived. You get the whole marketplace lifecycle in one company.
What should you check in the filings?
What to check in filings: GMV/GOV and its growth, by cohort or city where given. The take rate's level and composition. Contribution margin per order or transaction. Incentives and discounts, which are often netted against revenue, so find the gross number in the notes. Frequency and retention on both sides. And float. Marketplaces collect instantly and pay sellers on a cycle, which produces negative working capital. You'll see it as "payable to merchants".