This tutorial was originally published on the GameAnalytics blog
As game designers, we tend to perceive our activity as a mix of art and science. After all, game design is deeply linked to psychology, as well as design! The design part is pretty straightforward and well-documented. There is plenty of reading on the topic, starting with Jesse Schell’s famous textbook.
It is different with game analytics.
The domain is still young. 10 to 15 years ago, we lacked both the processing power and the connectivity to track player behaviors. But today, it is different! Game analytics offer an exciting array of new possibilities for everyone in game studios! Game analytics can change the face of game development. They are helpful from alpha testing to QA, community management to monetization.
Where we had to guess before, we designers can now make decisions based on real data! This is what we are going to talk about today.
So what are game analytics?
Game analytics are simply the study of our players’ behaviors using statistics. This expression covers all the types of data you may want to track. Most of the time, we tend to associate them with marketing and monetization. However, those statistics are not only for marketing people or producers!
They are a great learning tool, an occasion to get to better know and understand your audience. Game analytics offer us an opportunity to understand players beyond our subjective interpretation.
Analytics boil down to metrics
A metric is a stream of data that is being tracked over time. Metrics can track anything: average session duration, game uninstalls, player demographics…
There are 4 main categories of metrics:
- Customer metrics: They correspond to all the data related to the acquisition and retention of customers. They can also be seen as the marketers’ data. Specific metrics in that category include DAU (Daily Active Users), ARPU (Average Revenue Per User).
- Community metrics: Community metrics focus on the community’s behavior and evolution. They track what happens in your in-game chat for example. All sorts of social interactions fall in that category as well. For instance, both in-game and social network messaging.
- Performance metrics: Performance metrics track your application’s performances and potential bugs or crashes. Be it a response time from your distant server, the game’s loading duration or framerate at runtime. Anything that can help you to improve your back-end systems.
- Gameplay metrics: Gameplay metrics record anything that happens inside the game, between the player and the game. I.e. time spent in a given level, how many times the player died. They empower us to estimate the quality of the user’s gameplay experience.
As game designers, we often like to approach our projects in a mildly artistic way. After all, we are designers. At first, game analytics may seem like a cold tool, a big black box for marketing people to boost the monetization of a game. For one, monetization related analytics reflect the quality of experience offered by the game. E.g. retention is relevant to game designers: they give you a broad overview of how compelling the first experience of your game is.
But game analytics are not only about money. They are about analyzing any type of gameplay related metric!
Despite being an artistic discipline, game design is also a highly technical activity. Game analytics enable us to understand the behavior of players using factual data:
- If you know when and where a player gets stuck, you can easily tweak your game’s difficulty.
- If you know what items people buy and people don’t, you can adapt prices.
- If you know how old your players are, you can better adapt your content to them.
Game Analytics are relevant to everyone
Studios of all sizes are concerned by data tracking. Well, I won’t need to tell you that all AAA studios tracked by beta testers’ and players’ behaviors. They have the largest budgets, thus the highest need for objective data. A team of 300 people cannot only rely on guesses from game designers!
We could all use some basic metrics to understand how well our players react to our game. Good thing is, you don’t need a dedicated server and a backend developer to set up analytics for you. Today, there are plenty of free tools that handle the processing and monitoring for you (e.g. GameAnalytics)!
Game analytics are also relevant from the prototyping or Alpha testing phase. They give us an opportunity to check that our core gameplay works. You can track the player’s movements and know where he gets stuck in a given level. You can also track at what point he or she leaves the game: was it after a death, in the middle of the level? Or was it maybe after reaching the end of a section, just as expected?
Even without tracking the player’s movement in game, you can solve critical issues with basic metrics. E.g. you can estimate the quality of your game’s tutorial, or you game’s appeal during the first gameplay session. On any mobile game, a bad design decision can kill your game’s retention! For example, a lengthy tutorial or a difficult first level! Players will uninstall the application if they fail straight away. Or if they are simply bored.
Working with user analytics
When we work with user analytics, we work with hard to predict human behaviors. We cannot simply rely on a few generic set of metrics. DAU, ARPU and other common metrics only give us insights on the overall state of the game. We need each part of our games to be accessible to a wide range of users. As designers, we want to understand our players on an individual level. As well as on a broad level! Thus, we need a lot of specific data to run proper analyses.
Yet, we cannot track everything
We have to pick specific streams of data to analyze! All sorts of metrics are at our disposal, but we can’t track everything. Each stream of data we monitor has a processing cost. Both on the server and the human side. Although we can collect and treat a lot of data today, we can’t get everything. We have to find what metrics are relevant to us, and which ones are not.
What metrics should we track?
The nature and amount of metrics you should track completely changes from project to project. You need to plan your use of game analytics during the game’s pre-production phase. Depending on your game’s genre and scope, you will find more or less relevant metrics. For instance, community and monetization related metrics don’t matter in single player games.
Beyond a few generic metrics, each project has different needs.
Small teams don’t need to track large amounts of data though. They don’t need to monitor all user actions and input like big companies do. A few simple metrics can help you solve critical gameplay issues. Here are some useful general picks to balance your game progression. You want to know:
- When players leave the game
- Their average session duration
- How far players go in the game
- At what time players uninstall your app
The few streams of data exposed above are very easy to track. By combining those streams of data, you can get a sense of what content frustrates your users. Mixing this data with user demographics, you can also refine your analysis based on your audience. E.g. you can filter out 10 years old users uninstalling your hardcore game for grown-ups.
Gameplay related analytics go extremely deep. Even the simplest game uses a large amount of variables. Objects have a position, states, parameters, etc. We cannot track them all individually: analyzing the results would cost a lot of time. Thus, we need clever ways to group sets of data together. Basically, we can group actions that are similar in their form. Let us imagine that we want to know how efficient weapons are in a game. We don’t need to track each weapon’s type, modifications, etc. Instead, we can focus on the damage players inflict to a given monster.
What metrics should we not track?
It is probably easier to filter out the data we shouldn’t bother tracking. A game designer doesn’t need to track every bit of available data to get profound insights on the players’ behaviors. A few carefully picked metrics can tell us a lot. Also, the more streams of data we analyze in a given category, the less benefits we gain per metric.
We can only optimize our model of analysis to a certain point. It is tricky to know when to stop, as some very specific design flaws may only be found via a unique metric.
But we basically do not want to track too many individual variables. Let us take a multiplayer RPG as an example. We want to know what is the overall power level of each character. Instead of tracking every characteristic, we can compound them in a few relevant values. We can simply track their damage per second and survivability. That way, we only have to monitor 2 relevant metrics.
Finding the right metrics to follow and to filter out is a tough balancing act. It is not a science. So do not hesitate to experiment!
For a game designer, all metrics matter
Free to play titles have evolving communities. Some players only come back to the game occasionally. Others rush through the available free content and leave. Games with persistent worlds do not only need to retain players. They also need a constant flow of new players to stay afloat. Because of that, their user base keeps evolving! Their community evolves. As a game designer, you need to find new ideas often to keep refreshing the world.
That is when community metrics come in handy. Most of the time, player will express their desires in the in-game chat’s general channels. Part of your role is to answer some of the players’ desires. So tracking general conversations can be a good way to get new ideas and valuable feedback. This is also a great place to discover bugs, as players will tend to express themselves spontaneously in the chat. It is much faster than contacting the support team. I guess that you can see where I’m going. All types of metrics are valuable to a game designer!
On the advanced side: Updating on the fly
Using game analytics opens other, more advance opportunities to improve your development pipeline. One of these is the ability to dynamically update the game.
It takes a lot of time for an update to be validated on any app store. Whenever your game has a critical gameplay issue, you don’t want to be waiting for days for the issue to be solved.
With solid metrics, you can monitor the way players are reacting to your latest update. If you see that players are struggling against a given boss, you want to change its health maybe. And you want to change it now. If you see that players are not buying some objects in your game, you want to tweak their cost. And you want to tweak it now.
When the player launches the game, you can download values from the distant server. That way, you can balance your game on the fly! As most players are always connected to the Internet, we can keep improving the gameplay experience from day to day. This is also a great way to iterate really fast during beta phases. Although that technique is somewhat complex to implement, it is a great possibility to keep in mind.
Hopefully, this article gave you an overview of what analytics have to offer to a game designer.
To sum it up, Game analytics are not just about sales. They are about getting a better understanding of our games and of our users. They empower us designers to fix gameplay issues and improve the game’s immersion. They empower us to boost our game’s success. And by extension the game’s ability to spread and generate revenue. This is at least one terrain where designers and marketers can appreciate to work hand-in-hand. Ultimately, a game studio needs both visibility and money to live.
To harness the power of game analytics, as a game designer:
- Implement them from the pre-production phase
- Focus on understanding your users’ behaviors
- Take interest in all sorts of metrics!
Pushing the quality of the game is essential for any studio to thrive. It used to be a blind balancing act.
Now, we have game analytics.
This tutorial was originally published on the GameAnalytics blog