The Marginal Cost of Software

Fellow Enterprise Irregular, Bob Warfield calls me out on my claim that software has no marginal cost in my blog post App Stores Galore.

I write about the economics of Information Products often and have used the term marginal costs several times over the years in connection with digital music, e-books and of course the technology industry. So perhaps it is important for readers to understand what I mean when I say marginal costs are zero. In the process, I’ll also examine Bob’s claims to the contrary about the software on an App Store.

Wikipedia’s definition of marginal costs is

…marginal cost is the change in total cost that arises when the quantity produced changes by one unit.

Effectively, when I say that apps on the Mac App Store have a marginal cost of zero, what I am saying is that the next user of Angry Birds or Evernote or Omnigraffle, incrementally costs these companies nothing.

Compare this with Gap jeans. The next pair of Gap jeans sold will cost Gap a pretty sizable portion (maybe 50%?) of the sale price of the jeans.These costs are the material, manufacturing and shipping costs. And maybe a few others as well.

Bob’s rebuttal rests on three types of costs that he says prove that the marginal costs are not zero.

Engineering costs related to updates

Engineering costs are fixed costs. They have no relationship to the units of software sold. Sure, if you sell more units you have more money to spend on engineering, but that is conflating cause with effect. Think of the Gap jeans analogy. Is the next unit of Angry Birds sold going to need any extra cost in engineering? Nope.

Support

Support is a little trickier. If the app you are buying includes support over email or the phone, yes, there is a marginal cost here. But apps, and indeed, most consumer SaaS applications now direct users to their forums. On the forums, they either find the solution to their problem themselves or they post a question and a super user answers it for them. Often, someone from the company’s support team will have to answer the question. And so you might say that as the number of users goes up, the queries on the forum go up and the size of the support team must also go up. 

But, there is another dynamic at play here. As users grow in number, the knowledge base on the forum also gets richer. The number of super users also increases. Problems with the software that lead to support queries are ironed out. Does all this compensate for the upward pressure on support costs? I don’t know. Probably not. But for these kinds of apps, after a certain user base has been achieved, incremental costs are not material, in my opinion.

Server side costs of processing, storage and bandwidth

Even if the software downloaded is simple desktop software with no SaaS component, there is the matter of the first download and subsequent updates. And presumably, the traffic created by polling to see if an update is required.

I understand that these costs, for an additional user, are not zero. But there is something called materiality. Is this particular cost material compared to the revenue from the next user? I doubt that Angry Bird or Omnigraffle worry about it. Evernote perhaps does, but how many apps do you know that users regularly use to scan and upload documents to? 

Two endnotes

In software, there is always more that can be done. More products, more features, better performance, better customer service. The only thing holding it back is the money available to do all this. As a company grows in customers and revenue, it expands its engineering and customer service teams. Because it can. But this is not because of marginal costs. The causation is the other way.

Marginal costs is a construct used in Economics. In real life, you can never really parse out marginal costs from fixed costs with any certitude. Nor should you waste your time doing it. Managerial decision making, most of the time, doesn’t need numbers to be precise. So when I say, marginal costs are zero, but they turn out to be 2% of the unit price, it is not going to result in a different decision on whether one should be on the App Store or not.