21 July 2009
The lucky ones fail early
One of the core concepts behind the statement “everything digital and online will eventually become free” that is being popularised by various hacks and journalists is Marginal Cost, or, how much it costs to manufacture/produce an additional unit of whatever product you are pimping online.
The pitch is that, according to some economic theories, prices will inevitably gravitate towards the marginal cost of production. Add to that the popular myth that the marginal cost of production of anything digital is effectively zero and you have a neat, seemingly compelling argument that everything online will eventually1 become free so you should just get with the program and stop arguing already.
But, aside from that the theory is just a theory—only applicable in specific circumstances—and that, as I’ve previously argued, we’re already accustomed to having a variety of non-free, virtual goods in our marketplaces, and that this argument ignores externalities, most capital expenses and operating expenses, marginal cost online isn’t zero.
It’s ‘effectively’ zero, which is a completely different thing. There are bandwidth costs and storage costs, which, although amount to little more than pennies per user, are decidedly non-zero.
What makes this non-zero marginal cost even more important is the viral and psychologically compelling nature of ‘free’. A free online product can, at least for a while, gain a lot of users even though the product represents no value or utility to the user and has no prospects whatsoever of ever attaining any semblance of a sustainable business.
Consider the following, simplistic and made up example:
A small group of entrepreneurs create an online social product that takes off through word of mouth. Let’s assume that they have no customer acquisition costs—unrealistic as in the real world customer acquisition costs for a ‘freemium’2 web product can easily go over a dollar per eventual paying customer.
It quickly becomes clear that the product is taking off and that they’ll reach a million users within a year or two. They then work to find investors, spend that money to build up an infrastructure that will scale to that many users and optimise the hell out of it.
The usage load on the infrastructure always exceeds their expectations, at least to begin with, as in social networks people spend a disproportionate amount of time setting up and using their accounts at first as they connect with their friends and try things out.
At a (hypothetical) marginal cost of 10 cents per user per year, the entrepreneurs plan to just scale like mad, bind their users with so-called network effects and then figure out a freemium business model with a million-plus users in the bag.
Then, as the plucky entrepreneurs get over the million user hump, the number of active users falls off a cliff. It seems that in most social network most of the users at least try the product out to begin with (it only costs them time as it’s free) but regular and active use is only compelling to a small proportion of the user-base.
The entrepreneurs end up with 100 000 active users in an infrastructure designed for a million, an effective marginal cost of a dollar per user, a rigid system that can’t adapt to new opportunities because it’s been optimised to near-death and still no semblance of a business model.
As sad as the situation of these fictional founders may seem, they are the lucky ones. Imagine having to scale to a 100 million users to learn the same lesson.
-
One of the problems with ‘eventually’ is that in any economic model of this sort most of the actual profit is in the inefficiencies before this ‘eventually’. That is, most of the money in the industry will be made before the forces of competition manage to drive the prices down to marginal cost. That’s assuming everything goes according to the theoretical model, which, as with all best-laid plans, rarely happens. ↩
-
Freemium is just a buzzword for a mishmash of older economic concepts, the most important one being ‘versioning’. Most of the users get the product for free, some pay for additional features, some pay much more for really exclusive features and so on. ↩
