This featured guest article by Christian Thurau on game metrics, from essential to advanced, and their value in game development. Christian has a PhD in data mining and machine learning, and he is the CTO of GameAnalytics, one of the latest Unity official Asset Store Service Partners.
Facts about Christian:
Favorite games: Old-school shooters, mostly Quake 1, Team Fortress, and Bioshock
Favorite GA tool: Heatmaps
Favorite metric: Playtime
Analytics has become a much-discussed topic in game development in recent years, not only for F2P but across the entire industry. While it is still creativity, intuition, and experience that matter most in creating successful games, statistics on games and players have become an integral part of game development.
Unfortunately, analytics have become somewhat notorious as tools for monetization, used to influence player behavior to maximize revenue. What is less known is that metrics are just as useful in increasing fun and engagement. Hopefully, with this article, I can shed some light on the true nature of game analytics and break through the shadow of corporate greed that clouds its horizon.
The first and most important question with any analytics tool is what exactly you should track. While you would need to see into the future to know every metric you’ll ever need, there are some clear must-have measures that will provide a very solid starting point. Generally, these basic metrics can be placed in three broad categories: Acquisition, Monetization, and Engagement.
You may have already figured it out from the introduction, but of these three categories, I believe engagement is the most important, mainly for two reasons:
Some engagement metrics sit at the base of player lifetime value calculations, a metric that is used exclusively to ensure user acquisition with a positive return on investment.
Engagement relates closely to the game’s “fun-ness.” It is a measure of the degree to which your game fulfills, so to speak, its destiny. There’s no reason to attempt to improve on other metrics if engagement is low.
Of course, the priority of these metrics might change according to individual game needs.
If you have never tackled analytics before, this might be a lot to take in. The good news is that the free GameAnalytics's SDK for Unity automatically takes care of all the basic tracking needs I mentioned. All you have to do is interpret the results.
For example, we estimate engagement by generating both retention and session metrics.
Day 1, day 3, day 7, day 14, day 28 retention: “Day X Retention” refers to the percentage of users who return to the game X days after installing it. This metric can be used to track how new builds or features of your game perform. The more often people return to the game, the more satisfied they are with the experience and the more likely they are to spend. On the other hand, low retention, depending on when it manifests itself, can indicate anything from a weak core loop to low production value or an endgame that offers insufficient content.
Daily active users (DAU), monthly active users (MAU), and the DAU/MAU ratio: The number of people who use your game on a given day and in a given month can give you key insights into its popularity, as well as allowing you to plan for growth and server loads.
Session length, average session length, and number of sessions per user: Sessions describe when and how long players engage with your game. Average session statistics can paint a clear picture of how your game fits in the player’s lifestyle—whether it’s played in short bursts on the bus or train, for example, or the player really likes to allocate a lot of time to get into the experience. Using session lengths, you can tweak the game experience to match the most popular (and natural) play styles. If you are using in-game advertisements, you can also look at sessions to estimate ad exposure and thus ad revenue.
Once you achieve solid engagement numbers, monetization and acquisition metrics come into play. GameAnalytics generates a number of fundamental metrics to allow you to estimate your income:
Revenue: the amount of currency, converted to USD, collected on a daily basis. You should always display revenues segmented by acquisition campaigns to identify your best-performing acquisition channels. Also, it is recommended that revenue from in-game purchases be distinguished from advertising revenue through custom event hierarchies.
Average revenue per daily active user (ARPDAU) and average revenue per paying user (ARPPU): how much users actually spend, both in absolute terms and compared to the total number of users. The bigger the ARPDAU, the better the chances your game will be self-sustaining.
Conversion rate: the percentage of users who make an in-game purchase for the first time on a given day. You can use conversion rates to assess the effectiveness of different special offers, for example.
Number of paying users and number of transactions: how many users and transactions occur in a given period. Flat numbers can sometimes be misleading; it’s helpful to look at the trend lines for the number of paying users versus the number of transactions in order to determine if you need to adjust the in-game product prices
Acquisition metrics have only recently been introduced into the GameAnalytics tool, so they are a bit on the basic side:
Paid vs organic users: comparison of the number of users who came to the game in a natural manner (word of mouth, friend invite, or stumbling upon it in the Game Store) to the number of users acquired via an advertising campaign. This basic metric can point to what is known as the K-factor, or the number of additional users that each user introduces to the game. This is very difficult to calculate without tracking all in-game invites, but the paid vs. organic users metric can at least give you a rough estimate. The K-factor is essential to determine whether your user acquisition campaigns pay off.
Number of installs by country, build, acquisition campaign, or other factor: the number of times your game is installed. The installs metric can show what iteration of your game is most popular and where the game performs best. The number of installs also gives you a rough overview of where your game is in its life cycle.
Remember, though, that you can segment all of the other predefined metrics by acquisition campaign parameters, as well. This ultimately allows you to make the best decisions when it comes to user acquisition.
The more you get the feel for using analytics as part of game development, the more you’ll find the need to track data that is unique to your game. This is easily done with GameAnalytics, which is centered around the concept of abstract events. An event is simply anything that could happen inside of your game; this can be a rocket that gets fired, an item that is picked up, or a banner that gets clicked. No matter what is important to your game, you can track it using GameAnalytics events.
Note, however, that while the old mantra of “there's no data like more data” is certainly true, selecting which custom events you should track is crucial. You should always design the desired metrics in advance; otherwise, you’ll risk wasting a lot of time to get your head around what is basically clutter.
As nothing beats solid statistics when it comes to optimizing in-game economies, you could start with tracking specific item purchases. Then, you can easily gather statistics on your best-selling virtual goods, find out what works and what doesn't, and ultimately get rid of useless items and optimize item pricing.
For even more advanced uses, combine user progression events with the funnel tool to open up a new world of tracking possibilities:
New user flow: Having a perfect first-time flow greatly improves the retention of your game. Data on where users drop out is essential for tweaking the learning curve.
First purchase: Identifying what item is being purchased first by players can help you understand what motivates players to spend money. This ultimately allows you to maximize conversion rates.
Missions and achievements completion: Since achievements and missions are optional in most games, their completion rate is a good indicator of what type of content is most popular among players.
Level progression: The players’ progression through the game is a good indicator of how much of the original game content has been consumed, allowing you to time your content updates perfectly.
There are a ton of other useful analyses you can do with the free GameAnalytics tool, and we are continuously working on new, exciting features to help you build better games.
Last but not least, I want to mention that there’s one GameAnalytics feature that is exclusively available inside Unity: 3D heatmaps. Heatmaps visually depict the frequency of events, directly over the game-level architecture. Let's assume that you are already tracking players’ death events. With the 3D heatmap visualization, you can determine exactly where in the game players are dying. You may notice bottlenecks, such as narrow bridges where players tend to bump into each other or where they are simply more exposed. If you also track killer positions, you may also notice exploits, points where players camp out to ambush their enemies. Heatmaps are not only a great tool for fine tuning game balance; they also help in verifying that your game is played as intended.
I’ve barely scratched the surface of heatmaps with this short example, but you can read a lot more about them on the GameAnalytics blog. There are some nice visual examples there, as well. Maybe we’ll be able to delve into the details in a future article.
You can read our blog for many kinds of insights into data analysis in games, from in-depth introductions to the basics to real-world case studies. We invite leading industry experts and researchers to contribute to our blog on a regular basis.