A word about averages

• Tweet
• Pin It

In service-based games, averages are misleading.

Humans tend to understand averages through the lens of physical properties. If I tell you that the height of an average male human is 5′9″ and ask you to guess the height of the next male human to walk through the door, 5′9″ would not only be a pretty good guess, you are unlikely to be more than 6 inches wrong in either direction.

Wealth, on the other hand, is not constrained by physical properties. The median wealth of a US household is $68,828. The richest man in the world, Jeff Bezos, had a net worth of $91 billion in 2017. Jeff Bezos is more than a million times richer than the median household.

In practical terms, this means that when you read an “average” figure about service games, you need to think hard about what it tells you. In the world of physical products, if a game generated retail revenue of $40 million and sold a million units, you could surmise not only that the average sale price was $40, but that most sales took place at around that value.

If a service game says it has a million players and $40 million in revenue, you have no idea whether that means that most players spend about $40, or one crazy rich tycoon is spending $40 million on the game and everyone else is playing for free.

For most games, the distribution is likely to look like this (read more about this graph in “How much is your game worth?” )

Most people spend nothing, then those who do follow a distribution that is closer to a power law than a Normal distribution. A useful test is to query your data for the mean spend, the median spend, and the modal spend. The closer they are together, the more your game generates revenue according to a Normal distribution. The further they are apart, the more your game follows a Power Law distribution.

This is an extract from Nicholas’s new book, The Pyramid of Game Design – get your copy here!