Building a marketplace business

Every marketplace has two sides. One is the hard side — supply or demand that's structurally scarcer, more expensive to acquire, or less motivated. Solving for the hard side first is the path that works:

• Airbnb — hard side: hosts. They went door-to-door in NYC photographing apartments.
• Uber — hard side: drivers. They paid early drivers fixed weekly minimums to guarantee availability.
• OpenTable — hard side: restaurants. Free hardware + sales team in major cities.
• Etsy — hard side: sellers. Built tools for craft makers before SEO scaled buyers.

The mistake almost every marketplace makes: trying to grow both sides equally. You can't. Hire someone explicitly responsible for the hard side; treat the easy side as a marketing problem you'll figure out after liquidity exists.

How to find your hard side:

• Which side has scarcer alternatives? (Demand for handmade — many sellers but each has limited capacity)
• Which side has higher acquisition cost? (Insurance agents are harder to get than insurance buyers)
• Which side defines the product's quality? (Tutors define StudentRight; both sides exist but tutors are the hard side)

Liquidity = the probability that a transaction happens when someone tries.

For Airbnb: of all guests who search a city on a date, what % book? For Uber: of all riders who open the app, what % get a ride within 5 minutes? For Etsy: of all sellers who list, what % make a sale within a month?

Marketplace metrics that aren't liquidity:

• Number of listings — doesn't matter if buyers can't find what they want
• Number of buyers — doesn't matter if supply is wrong
• GMV — flatters early stages where you're subsidising both sides

Liquidity has a threshold below which the marketplace doesn't work and above which it compounds. For Uber, that threshold was 5-minute pickup in core hours. Below that, riders churned; above that, riders recommended the app.

The hardest part: liquidity is geographic / categorical. Airbnb has high liquidity in Paris and low liquidity in Detroit. Uber has high liquidity at 8am Tuesday and low liquidity at 3am Sunday. You can't average them. Build market-by-market and only declare PMF in the markets where liquidity is above the threshold.

Take rate = the % of GMV you keep. Airbnb is ~15%. Uber is ~25% (effective). OpenTable is ~$1.25/cover. eBay is ~12.5%.

Most marketplaces start with take rates that are too low because founders are scared of the hard side leaving. Take rate goes up as liquidity goes up — once supply needs the marketplace more than the marketplace needs them, you can raise rates.

The cap on take rate is disintermediation — the point at which buyer and seller meet on your marketplace then transact off it. Disintermediation risk is highest when:

• Transactions are high-value but rare (one-off services)
• Buyer and seller need ongoing relationships (housekeepers, tutors)
• The 'product' is information about the counterparty (matchmaking)

Anti-disintermediation tactics:

• Trust and reputation systems that only work on the marketplace
• Payment processing the marketplace owns end-to-end
• Insurance / dispute resolution that doesn't transfer off-platform
• Discovery / matchmaking that's hard to replicate (Airbnb's filters, Cal.com's scheduling logic)

When marketplace is the wrong shape:

• Low frequency transactions where buyer and seller transact once. Hard to build reputation; high CAC per transaction.
• High touch / negotiated deals where each contract is custom. Marketplace dynamics don't apply.
• Niche / supply-side captured markets where one side controls the value chain (selling software to specific industries).