Two-sided Market

A company that buys goods and resells them has one kind of customer. Bring in more buyers and revenue rises; as long as what it costs to acquire them stays under what they pay, the structure holds. One lever, simple to manage.

A business that lives on connecting people doesn't work that way. There are two kinds of customer, and the condition that makes the product useful to one side is a number on the other side. Without a separate name for that structure, teams end up managing both groups as a single user count — and counting that way hides which of the two is short.

Pagedeck made exactly that mistake. When Grace proposed a marketplace for buying and selling document templates, Ravi expected it to spin up quickly given 40,000 existing users. Three months later there were 12 templates and 7 purchases. Those 40,000 were candidate buyers; sellers had to be recruited separately. Creators wouldn't spend time on something that might not sell, and buyers wouldn't visit a store with 12 things in it.

What makes up this kind of market is two groups and the value running between them.

When one side grows, what the other side gets grows with it. Go from 12 templates to 400 and buyers have enough to choose from to justify a visit; as buyers arrive, creators start earning and make more. The property where value rises as more people use something is called a network effect, and here it works across groups rather than inside one. Template creators and template buyers are each other's reason to show up.

Which makes starting hard. No buyers means no creators, and nothing made means no buyers. That state is the cold start problem. Four approaches get used in practice.

1. Pay one side first. Advance twenty creators about $150 per template so supply exists regardless of sales.
2. Narrow the scope. Open with weekly meeting-note formats only rather than every kind of template, and 12 items is suddenly enough to choose from.
3. Fill one side yourself. The company makes the first templates, and outside creators are admitted once transactions are moving.
4. Make one side useful alone. If the document tool delivers value with an empty marketplace, users stay, and those users become the demand side later.

Where the money gets charged is also a decision. You can charge both sides, but most businesses load it onto one. The test is which side is less price-sensitive and which side is harder to find. Where creators are scarce, charge buyers; where buyers are scarce, take a percentage of each transaction from creators.

The awkward part of this structure is that too much of one side is also a problem. With 4,000 templates and buyer numbers flat, a creator earns a tenth of what they earned at 400, and creators leave first. So what gets managed isn't the size of either group but what one side actually receives from the other: monthly sales per creator, the probability that a buyer finds what they searched for. Those two figures are the ones that tell you which side to feed next, and they behave differently from headcount — a marketplace can add creators every week while the number that matters, what a creator earns, falls the whole time.

It appears first when a business structure gets written down. Setting out who you sell what to and where the money comes from, the moment the customer column splits in two, this term becomes necessary. Two columns means every downstream plan comes in pairs: two acquisition paths, two retention plans, two sets of metrics. The product form of this market is called a marketplace.

Investor reviews raise it too. You get asked to show growth curves for both sides separately, and a company where only one side is growing fast has to explain why. A combined curve hides the bottleneck, and the side that looks healthy in the aggregate is usually the one that was easy to acquire rather than the one the business depends on.

It carries into the organization. Many companies split the team that recruits and manages supply from the team that brings in demand, because recruiting creators looks like one-by-one sales while bringing in buyers looks like advertising and search visibility.

It also travels with the asset-light model, where a company earns from connection without owning the assets. No inventory to buy means less capital tied up, and balancing the two sides is what arrives in exchange.

Calling things this that aren't is common. A work tool selling one capability to many companies has one kind of customer and is not a two-sided market. The test is whether the product's value to one group disappears when the other group is absent. Getting this wrong in a strategy document is expensive, because it sets up a plan with two acquisition budgets and two sets of metrics for a business that only ever had one customer.

"This is a two-sided market, so stop counting users as one number. Creators and buyers have to be read separately."

A combined figure hides the bottleneck.

"Let's fill the creator side with subsidies first and attach buyers after that."

Paying one side to get past the cold start.

"We have 4,000 templates and creator churn is rising. We need monthly sales per creator."

Oversupply is collapsing the earnings on one side.

Pagedeck narrowed the scope of its stalled marketplace. Instead of accepting every kind of template it opened weekly meeting-note formats only, and advanced twenty creators about $150 per item. Two months later there were 140 meeting-note formats and 210 purchases a month.

After that, the number Mia managed was not the count. It was what a creator earned per month and the share of buyers who found what they searched for. In months where the first figure dropped below about $90, recruiting new creators paused and demand-side traffic got the attention instead.

The most common reading is that a large user base makes a marketplace turn on its own. Layering a store onto a product 40,000 people already use makes demand look pre-assembled. But existing users are candidates for one group, and the other group has to be built from nothing. Left uncorrected, a team opens the store, waits a few months, and concludes demand was insufficient — when the empty side was supply.

Assuming both sides must grow at the same pace is the second misreading. It comes from hearing that balance matters and reading it as simultaneity. In the early phase you deliberately push the harder-to-find side ahead. Grow both a little at a time and both stay under their thresholds for longer, and the people who arrived first leave during the wait. Balancing is work for after transactions are moving.