On two-sided markets – or why free may be the right price in some cases, and in some cases not

So you just coded this really cool Internet application, which is kind of a mixture between, youtube, facebook and flickr, but targeted stamp collectors. Everybody likes it, even your former high-school sweetheart thinks it has enormous potential. Congratulations! Time to uncork the Champagne. Mmmm, but how are you gonna make money? No matter if you plan to get a Boeing 737 or you are just happy to cover the costs of running your service, you need some way of getting some income.

Are you gonna handle it like twitter (quoting their about twitter page):

Twitter has many appealing opportunities for generating revenue but we are holding off on implementation for now because we don’t want to distract ourselves from the more important work at hand which is to create a compelling service and great user experience for millions of people around the world. While our business model is in a research phase, we spend more money than we make.

Or listen to the 37 signals crew, and start charging from day one? And who is right, anyhow? Do their cases even generalize, or does it just work for them? Should I build a user base first? Or is Freemium the right way of doing things? I am glad to announce that the final and definite answer is: it depends!

It really is! Way to little attention is put on that fact, instead too many people argue with absolutisms. (Example: DHH’ startup school video. This is by far one of the most notable talks I’ve seen, but his arguments just don’t hold for every type online business. An explanation why will follow below.)

Before doing anything, you first have to determine what kind of ecosystem your business actually is positioned in. Common sense might tell you this, but common sense is sometimes not so common.

I had my personal AHA moment recently when I browsed through some of my girlfriend’s Business Administration course material. The keyword that put order to the chaotic chitchat about business models on the Web is two-sided markets.
A two-sided market connects two different groups of customers. For instance, credit cards connect card owners (buyers) and store owners, game consoles connect players and game programmers, Google connects searchers and advertisers, or PDF connects readers and creators of documents. In a nutshell, you usually encounter a two-sided market situation, when you provide a platform kind-of-service for two customer groups. (Side note: in this situation turns out perfectly fine to serve two customer groups at once, it’s even essential to understand this. Even if some smartass advisor or investor tells you have to decide who your customer is.)

Now the interesting part of this is, which customer group should be charged how much to use the platform. And even more interesting: it turns out that it is not always the same solution – it depends (again…).
Some of the main factors this depends on is the value proposition for both customer groups and the network effects triggered by the platform.

For example, when Adobe first released PDF, it offered Reader and Writer as a bundle, in essence charging both sides of the market, i.e. document readers and document creators. Adoption did not take off, until Adobe lowered the price of the Reader to 0. The network effects shifted the demand curve at the document creator side so dramatically, that the revenue lost from offering the reader for free was more than compensated by selling significantly more writers at only a slightly higher price. (This is illustrated further in this Wikipedia article and this paper).

So, the learning here is that demand curves for the different customer groups in two sided markets are not independent. You may say it’s obvious: just offer the consumer something for free and things will take off. HA! Time for the next example: PC’s or Apple vs. Windows.

Apple started out the same way as Adobe, charging both software producers and software buyers. (Note, the PC is just the platform for software, software producers and software users are the two customer groups). The high price for developing Software on the Mac resulted in a limited amount of software available on the Mac, which made Windows more attractive, where Microsoft gave away the software development kit for free. So, here the situation is just the opposite, the user pays and the software producer is not charged at all.

Innnnarrrrresting!

Now, it gets even more confusing with example 3, the game consoles market: it may seem it is just the same situation as PC’s, but it turns out that here, the developers and the users are charged, in fact the developers have to pay quite high charges to develop for the game platforms (Playstation, xbox etc.). The reason for this may be quality: the users and platform providers request really high quality games, so not “just anybody” can develop games. (Side note: I think one intermediate way is to offer an SDK for a very low price, to keep away just “the worst” quality producers. It seems to me as if Apple does that with the iPhone SDK (itcosts roughly 100 dollars).)

So, with this we discovered one other factor, which affects pricing and strategy decisions in two sided markets. One further factor to consider is competition: if a competitor offers a very similar platform, pricing decisions are crucial, up to offering service for free initially. The reason is that two sided markets are often winner-takes all markets, where only very few big providers dominate. (Examples: Microsoft/Apple (PCs), Google/Bing (Search engines), Facebook (Social Networks), Playstation/xBox/Wii (Game consoles).)

The goal you want to try with your strategy is to enable the network effects to play freely in order to position your platform as the leader and profit from the maximum possible revenue triggered by the right pricing strategy. All of this is analyzed in more detail in the papers listed at the end of this post.

Now one also has to point out, that not all businesses on the web are platform businesses in two sided markets. For instance, if you offer a project management software (as a service), this is basically just selling a product. (Examples: 37signals’ basecamp, Salesforce, etc.). They only get some taste of a network if an API is offered, since this makes the software a platform on its own.

So, it essential that you find out what kind of context you are in. In Internet and platform-like businesses you will often encounter two-sided market situations. If you are in a two-sided market, pricing strategies are complex and may involve offering a service to end-users for free, indeed. To that end, to enable the network to grow beyond a tipping point and network effects top play, you may need the help of a bit deeper pockets (i.e. VC money) indeed. If you simply offer a service to a single group of customers, pricing is simpler. You may still offer a free trial as a kind of loss leader, but it is a different situation than in a two sided market, where it can make sense to offer a product for free for ever to one group of customers. (This is also why I think a comparison with free muffins in a bakery is simply wrong.)

On a final note, all of this is also related to finding out or determining what kind of company you are, or as Joel Spolsky puts it: Are you Ben and Jerry’s or Amazon? Finding out in what market situation you are in might be a good start to figure this out eventually.

Further reading on two-sided markets:

Wikipedia has a good summary

Harvard Business Review: Strategies for Two-Sided Markets.

Geoffrey Parker and Marshall Van Alstyne (2005). “Two-Sided Network Effects: A Theory of Information Product Design.” Management Science, Vol. 51, No. 1

HARVARD BUSINESS REVIEW ARTICLE: New Rules for Bringing Innovations to Market

Further reading for expanded context

37 signals podcast: Episode #3: Making people pay.

Joel on Software: Strategy Letter I: Ben and Jerry’s vs. Amazon

Joel on Software: Strategy Letter II: Chicken and Egg Problems

Joel on Software: Camels and Rubber Duckies

Chris Anderson: Free, the future of a radical price
Book on amazon.
Article in Wired

Leave a Reply