In the last week, four different people have asked me questions about how to make a two-sided marketplace work. They’ve included a travel site, a jewelry startup, a photography marketplace, a task marketplace. All of them have the same challenges.

So I figured I’d write down some of the things that have come up.

You’re not a unique and special snowflake

Any two-sided marketplace faces three classic problems:

1. Creating demand
2. Creating supply (inventory)
3. Maximizing the number of matches between a buyer and seller (based on price, filters, and so on.)

You may think you’re different, but you’re not. These are always the three challenges. The first two need to be resolved for you to launch; the latter, once you’re seeing attempted transactions. Later, you need to deal with two other problems: sustainability, and avoiding fee circumvention.

Saying “we will spend money on SEO to create demand or supply” sucks. This is not a sustainable business model, because you’re not creating growth from actions your users take. Later, once you’re making money, you can spend some of that (less than a third) on pouring that money into acquiring new customers.

But that’s not what makes marketplaces successful. Do you buy things from Etsy, Amazon, eBay, AirBnB, or others by clicking on paid results in search engines? No. You go to their sites, or install their apps.

Fake it ’till you make it

In most cases, marketplaces are a chicken-and-egg challenge, and the company has to find a hack to grow one of the two sides (buyer or seller) artificially.

• One rental-by-owner company, for example, crawled Craigslist and scraped anyone listing a house for temporary rental, and mailed them—a complete violation of Craigslist’s terms, but that didn’t matter much at the time. Artificial supply, which helped them compete in organic rankings.
• When Uber launched in Seattle, they paid town car drivers to idle. This generated supply; once the money started rolling in, they switched those drivers to a commission model. They artificially created supply, improving the way the service worked from the outset.
• Going after a niche where you have an unfair advantage, then expanding, is a good way to build demand. Amazon sold books; then other products. Once its customers were used to buying from the company, it introduced second-hand markets into its search results, bringing the demand it already had elsewhere.
• Fril, a Japanese used-clothing startup, capitalized on the fact that Japanese women sell their entire wardrobe twice: when they enter college, and when they leave. By making an app that required users to authenticate via Facebook—men simply cannot use the marketplace!—they targeted a narrow market using the tools and tone it needed. And because each buyer was also a seller, they grew both sides of of the market simultaneously.

Most of pretty much every marketplaces that you see succeeding started like this. When Ben and I wrote Lean Analytics we discussed stopping and writing a book on the exploits and skeletons in the closets of the most successful startups. If we were talking about your two-sided marketplace, all I would want to know is “what is your secret trick for creating supply or demand artificially?” Nothing else matters until you figure that out.

By default, it’s better to create demand than to create supply.

Imagine two startups launching cabbage marketplaces. One has a thousand cabbage farmers; one has a thousand cooks eager for cabbage. Which you you rather be? The one with the cooks—because you can always go get cabbage.

Proving you have supply is only interesting if that supply is somehow hard for the buyers to collect themselves (like comprehensive, trusted apartment listings; or something in a foreign language or currency; or something rare like in-demand tickets.)

Your role as matchmaker

As the person running the marketplace, you have one key insight: how to make buyers and sellers come together. You know which products sell for what prices; what search terms are driving traffic; and so on. If you don’t share that knowledge about demand with your suppliers, you’re failing.

• Give hints about what to offer: Etsy noticed that there were a lot of inbound searches for owls. So they told their designers this, which resulted in more owl-related art for buyers to choose from.
• Help sellers get their price right: If you know what price different pieces of art are selling for, you can build a predictive model of what the “best” pricing is. If you then share this with artists as a guideline for what they should charge, you’ll make transactions more likely.

You may not realize it, but in a two-sided market, you’re the analyst for all your suppliers. Think of your search results as a perpetual survey of your buyers—and think of searches with no results as potential need your suppliers can fill.

Once you make it

Once they get some traction, they survive either because:

• They achieve critical mass in a market with network effects. They are the default place to go—this is why Craigslist, AirBnB, eBay, Etsy, and other places with most of the inventory thrive.
• They provide a tool set ecosystem. The product or service has certain functions that buyer and seller need (escrow, analytics, a CRM, etc.) that the marketplace can provide with an economy of scale that beats everyone doing it themselves—Uber has ratings and a payment system, and a tool that shows the driver where to idle.

This is why most successful marketplaces wind up becoming technology plays. Not just because tech scales better than humans, but because if you don’t gild your offering with valuable tools and services that help to lubricate transactions, you’re not erecting barriers to entry. Value you create makes it harder for someone to steal your supply or demand.

Making sure you get paid

For larger-ticket items (such as jewelry sales, or wedding portfolios, or custom artwork) marketplaces face an additional challenge of circumvention. Buyers and sellers transact outside the system, robbing the marketplace of its revenue cut.

This can be countered in several ways. For example:

• By charging for a listing (as a C2C marketplace like Craigslist does);
• By taking a tiny cut from a huge market;
• By including features like payment (why eBay needed Paypal, and probably why Facebook wants to offer a payment tool.)

Marketplaces that resort to trickery (for example, by constraining the ability of the buyer and seller to see one another until the deal is done) often wind up undermining their user experience so badly they fail because of the obstacles they’ve put in the way of the transaction.

Marketplaces are not for the meek

Two-sided marketplaces look deceptively easy to start. You don’t need to make something; you don’t need to sell something. You just get to be matchmaker, and charge a tax.

Unfortunately, because they look easy to start, everyone’s doing them. Every accelerator I talk with has a housecleaning marketplace, a dog-walking marketplace, a local food marketplace, or one of a dozen seemingly easy two-sided markets in their cohort. And nearly every one of them is struggling with growth, because markets don’t sell themselves.

Herbert Simon called this over forty years ago: we live in an attention economy, and those who can capture the attention will win. Unless you have a hack to artificially kickstart supply or, preferably, demand, you shouldn’t be building a marketplace in the first place.