In my last post, I said that network effects are apparent when the users of a given network care about how many other users are in the same network. Put more formally, this means that willingness to pay for access to a network is a function of how many other users are on the network. If thirty years ago someone had offered to sell you the very first fax machine in the world, you-absent the ability to foresee the future-would probably have been willing to spend next to nothing for it, because owning the only fax machine in the world would be pointless. Fifteen years ago, however, once fax machines had proliferated widely, you would probably be willing to spend substantially more, since the large number of other machine owners would make it quite a useful piece of technology.

Understanding that network effects, value, and price are closely related suggests an important strategy for avoiding the penguin problem: subsidize early adopters… In other words, reward the first to sign up for a given network with discounts to encourage them to join, then raise prices over time as the participation of those early users raises the value of membership in the network to everyone else.

In fact, the existence of network effects suggests that platform providers should consider ‘under-pricing’ (relative to consumers’ willingness to pay) until the market has become penetrated. The laws of supply and demand tell us that whenever you raise prices, the number of users willing to pay those prices shrinks. But since the number of users in a network affects the willingness to pay of potential users, it behoves platform sponsors to keep prices as low as possible until they’ve become firmly established in the marketplace. Then they can raise prices to capture the value that they’ve created.

A look at the history of pricing for Microsoft Windows reveals a classic example of this strategy. Today Windows is several times more expensive than it was ten years ago. Why? Because the dominance of Windows in the desktop operating system market means that most users (especially business users) value very highly membership in the Windows ‘network’, because it allows them to use all of the marketplace-standard applications. As such, Microsoft is able to command a high price without worrying about discouraging new users to adopt Windows.

There are two big implications of all this for operators of mobile money services. First, penetration pricing may be the only way to grow a network to the point where network effects will make network membership sufficiently valuable to justify raising prices to profitable levels. This in turn means that the payback period will probably be long for investments in mobile money. Trying to recoup such investments more quickly, by raising transaction fees too soon, is a trap that can stifle growth and destroy the long-run profitability. But of course, this strategy is risky: deferring the potential payoff to the future means running the risk that that payoff will never materialize.

This is a good moment to make a related point about pricing. We’ve seen a few mobile money deployments that try to charge sign-up and subscription fees to users, rather than charging on a per-transaction basis. Sign up fees are attractive because they help to pay for costs incurred in customer acquisition; subscription-based models are appealing because they afford the operator a regular, predictable revenue stream. But both throw up barriers to adoption when users aren’t able to confidently predict the value they’ll get from the service, which is usually the case with new technologies like mobile money. Moreover, the existence of network effects makes it even more important that operators make it as easy (and therefore inexpensive) as possible for users to sign up, since that will make the service more attractive to others. This strongly suggests that per-transaction pricing is a better approach when it comes to mobile money.