The last couple of weeks, I wrote about how Bayes’ Rule is the strongest tool for making good business decisions. In this post, I will address one of the most important decision and how Bayes’ Rule can help, deciding what games or products to green light. In the game space, the green light decision is when a company decides whether or not to fund fully a project and put it into production. Some companies have a highly defined process, while others rely on intuition. The lessons of Moneyball already say who is going to win between those using a process and analytics and those using intuition, so I am going to focus this post on how to apply Bayes’ Theorem so you apply the right data. Although I am focusing the post on green lighting game projects, it can be applied to any new product.
One of the most common green light mistakes I have seen in the game industry is companies deciding on the merits of a game primarily based on how much fun the demo or prototype is. Related to this, they look at how the features of the demo/prototype compare with competitors and if it has enough competitive advantages they move forward. With the latter approach, you may feel you are looking at the opportunity very analytically but you are actually neglecting the most important data points.
Bayes’ Rule shows that often the best information for decision making is most likely the data from all previous game releases. As I wrote about last month, Bayes’ Theorem is a rigorous method for interpreting evidence in the context of previous experience or knowledge. Bayes’ Theorem transforms the probabilities that look useful (but are often not), into probabilities that are useful. It is important to note that it is not a matter of conjecture; by definition a theorem is a mathematical statement has been proven true. Denying Bayes’ Theorem is like denying the theory of relativity.
In the case of green lighting a game, rather than focusing on the game you should focus on previous game launches (both yours and other companies). You may have a fantastic idea for an Android pinball game; it is better than almost any other pinball game, it is a lot of fun to play and the focus testers you bring in love it. Without Bayes’ Rule, you may think you have a winner and commit to the project. Conversely, your team has also created a demo for a racing game. It is good and fun but not as unique as your pinball game. Focus testers who come in like it but do not go crazy. Based on these two descriptions, many would green light the pinball game.
Once you understand Bayes’ Theorem, though, you realize this decision making is flawed as it is not taking into account prior experience. Let’s say (and I am not using real numbers) that 95 percent of pinball games fail to make a profit while 75 percent of racing games are profitable. Even if you have a 90 percent chance that your pinball game is a great pinball game and a 50 percent chance that your racing game is a great racing game, all other things being equal you would be better off green lighting the racing game (I promised some people to avoid the math in this post after the last two posts about Bayes but if you want me to run through the numbers, just send me an email or leave a comment). If you are looking to optimize your profit and grow your company, Bayes’ Theorem just increased your chances of success.
When applying Bayes’ Theorem to green light decisions for games, there are at least four areas where you should look at prior experience to build great decisions:
- Mechanic Probably the most important criteria, the type of game you create, should be based on what game mechanic (e.g., hidden object, racing, casino, match-3) has been successful in the past. Some categories are huge and there are many successful titles. Other genres are huge, yet virtually none of the games make money. You must examine the results of games with a similar mechanic to see the probability of success. What you will find is that some genres are much more profitable and likely to succeed than others.
- Theme. Just as in other entertainment industries, your theme has a high correlation to a game’s success. While it is very difficult to make a successful action movie based on driving around the suburbs, your chances are much better if you set it as a battle against international terrorists. The same goes for games. There are themes, from candy to jewels to mystery, that work well and other themes (like science fiction) that repeatedly fail despite gameplay mechanic. By picking a theme for your game that players have gravitated to in the past, your project has a higher likelihood for success.
- Platform. Deciding what platform to focus your project on is also a great application of Bayes’ Theorem. Your project may look like the best social casino on Windows, but if only 10 percent of social casinos on Windows are successful but 85 percent of social casinos on Android are profitable (again these numbers are made up), then you would have a better chance for success with the Android game.
- Business model. You may have one potential project that looks great and is a paid download. Another project is not as impressive but is a free-to-play. Rather than going with the paid download game, you look at the top-100 grossing chart and notice that 95 of the games are free to play. Thus applying Bayes’ Rule you have a much higher likelihood of being a top-100 grossing game with the less impressive free to play title.
This post is not a call to “fast follow” competitors, as I have written before (The dangers of fast following) that is a strategy most likely to fail. Again, apply Bayes’ Theorem and you will start with the data that most fast follows fail to apply. When building your game, Bayes’ Theorem will help you understand what elements are most likely to set it up for success.
The Business of Social Games and Casino 
One thought on “Bayes’ Theorem Part 3: Making the best green light decisions”