- ARPDAUPosted 5 years ago
- What’s an impressive conversion rate? And other stats updatesPosted 5 years ago
- Your quick guide to metricsPosted 5 years ago
Experiment and Learn: Free-to-play design rule 14
- Experiment often
- Make sure the experiments are survivable
- Learn from the failures (and the successes)
For a creature in the primordial goo, the lessons are particularly harsh. Individual organisms, prompted by random genetic mutation, experiment with different survival tactics. Some of these experiments fail, and the individual protozoa dies, together with its particular genetic makeup. The species as a whole thrives as it learns how to adapt through a Darwinian process of trial and error.
That is not much consolation to the expired protozoa. Similarly, you will need to make sure that you are part of one of the successful experiments, not an evolutionary dead-end, as you transition your business or skills to free-to-play.
The good news is that you already know how to experiment. Nearly every game released has some experiments contained within. Experiments in gameplay, in technology, in story-telling, in distribution method; in something.
The games industry is good at experimenting. It is not so good at making sure that when an experiment fails, the company thrives. RealTime Worlds bet $100 million on the belief that it knew exactly what gamers wanted from a new MMO. It was wrong. Until we stopped counting, the GAMESbrief Job Loss Tracker followed all of the companies that collapsed or downsized as retail game sales declined and the publishers and developers who made them failed to adapt. Who knows whether the current titans of gaming like Sony and Nintendo will be the titans of tomorrow?
I work with clients who argue that they can’t make a game for less than $30 million. My response: you had better learn.
The market is changing incredibly rapidly. We now work in an industry beset by creative uncertainty (as it always was), technology uncertainty (as it always was) and business model uncertainty (this is new). Against that background, processes that helped companies achieve high Metacritic scores (which probably don’t matter any more) and reduce technological risk may actively hinder the experimentation necessary to carve out a new position in the new market.
The key to experimenting often and surviving the failures is to reduce the scope of the project that you are building. You are trying to discover whether there is a market for your product, not to drop it into a market with a limited number of products released every year through a limited number of physical distribution channels.
If I had $20 million, I would rather run 20 $1 million experiments, not one $20 million dollar experiment. If you work in free-to-play, I urge you to do the same.
Learn and keep learning.
Humans are amazingly good at seeing patterns. It helps us to make sense of the world. It also makes us imagine connections, correlation and causation where none exists.The secret of a good experiment is to know what you are testing for. Write down what you are doing, and how you will measure the results. Run the test and revisit the numbers. Explain to someone else the results of the test so that they can help you see if you are just seeing the results you want to see (confirmation bias). Even better, try to run a test to disprove your pet theory. Proving a theory wrong is much easier than proving it to be right.
Having learned something, make sure you use it. Share the information throughout your organisation. Keep a list of the tests you have run and the results. It will save you time later.
And then… Start again. This process never stops.