Markets With Two Sides

The idea in plain words

A newspaper sells to readers and to advertisers. A card network serves cardholders and merchants. A search engine serves people searching and businesses that want to be found. In each case one group is valuable to the platform mainly because the other group is there. Readers attract advertisers; advertisers pay for the readers.

Such businesses face a problem ordinary firms do not. They must get both sides on board, and each side’s willingness to join depends on the other’s. The solution is usually to make one side cheap or free and recover the cost from the other, and the choice of who pays is not a detail of pricing but the core of the strategy.

Where it comes from

Economists had long studied network effects, the way the value of a telephone or a standard rises with the number of users. Michael Katz and Carl Shapiro formalised them in 1985. In the early 2000s, studying card payments, Jean-Charles Rochet and Jean Tirole showed that platforms serving two groups set prices according to how sensitive each side is and how much each values the other, and that it can be efficient to charge one side less than cost. Mark Armstrong added in 2006 that when one side uses only one platform and the other must use all of them, each platform holds a bottleneck over its single-homing users and can charge the other side heavily. Tirole received the Nobel Prize in 2014, partly for this work.

What it does to the mind

The side that does not pay rarely notices that it is the product being sold. People treat free services as gifts and judge them as gifts, overlooking the interests of whoever is paying. The side that pays, meanwhile, often feels it has no choice, because the audience it needs is only reachable through one gate. Both sides adapt their expectations to the platform’s pricing, which makes the arrangement feel natural long after it has stopped being necessary.

Where it shows up

The essay on web search describes users who pay nothing and advertisers who pay for every click. The essay on payments shows card networks charging merchants and rewarding cardholders, and a study estimating how the cost reached people who paid in cash. The essay on app stores shows Armstrong’s bottleneck: developers must be on both phone platforms, users carry only one, and the commission stayed at 30 percent for a decade. The who pays and network reversal patterns collect further cases.

What it asks of someone deciding

Decide deliberately who pays. The paying side’s interests will shape the product over time, so choose the payer whose interests you can live with.

Solve the first side first. Every platform must cross a threshold where each side has enough of the other; find a group that already wants to be there, or bring one side from an existing business.

Watch for the bottleneck. If your users single-home and your suppliers cannot avoid you, your margins may be sustained by position rather than value, which invites regulation.