Startups & Business › Business Models
Two-Sided Marketplace
A platform that matches buyers and sellers and earns from the transactions between them.
Also known as: two-sided marketplace, marketplace model, platform business
A two-sided marketplace matches two groups — buyers and sellers, riders and drivers, clients and freelancers — and earns a cut (take rate) of transactions between them. Done right it scales beautifully: each side attracts the other, and the platform taxes the flow without holding inventory or employing providers.
buyers ⇄ marketplace (trust, discovery, payments, rules) ⇄ sellers
earns: take rate × volume — so both volume AND rate matter
The model demands liquidity above all: enough density of both sides in one place and time that matches happen fast. Ten thousand scattered users transact less than one thousand concentrated ones — which is why marketplaces launch city by city, category by category, never “globally”.
The classic mistakes:
- Launching both sides everywhere. Thin spread means no matches, which means no retention on either side. Constrain geography or category until transactions hum, then expand.
- Subsidising both sides forever. Paid demand and paid supply can fake liquidity through a fundraise or two, then collapse. Subsidies must buy a path to organic liquidity, with a dated plan to taper.
- Ignoring the chicken-and-egg order. One side must come first — usually supply (sellers tolerate low buyer volume better than buyers tolerate empty shelves). Sequence deliberately.
- Take rate fantasy. Assuming 20% on transactions where 5% is the norm (or vice versa, leaving money). Rate follows value added and competitive alternatives — research both before modelling.
- Disintermediation blindness. If parties can transact off-platform after meeting, they will. Earn the cut continuously (trust, payments, guarantees, discovery) or leakage eats the model.
Founder fit check: marketplaces are operations-heavy, slow to liquidate, and viciously competitive. They reward patience and local depth — see chicken-and-egg for the sequencing playbook.