Starting to use game analytics tools is high on the ‘gamedev new year resolutions list’ for many. We’ve heard it’s good, and some companies swear by it - but how can we start? How do we begin learning about game analytics without getting overwhelmed and shutting down?
This article is for a beginner audience re: game analytics. We’ll go into what it is, why it’s useful, and a non-intimidating way to start understanding and using it. Before you ask, all this information (and game analytics generally) is useful for premium games too - not just F2P, and not just mobile titles.
We define game analytics as the process of dealing with any kind of relevant game information or data, and putting it to good use. The ‘good use’ part is up to you - do you want to understand how players are enjoying your game design? Or maybe you want to investigate why your tutorial is discouraging so many players. You decide what you need to know, or the problem you want to investigate and go from there.
The science of game data analytics is a relatively new and exciting field, with a lot of recent research and fresh opportunities for the modern game developer. Introducing data-informed design considerations, testing, and more detailed ongoing observations of game data has been driven by a need for better knowledge about the players.
Games are no longer static, but often designed to live, breathe and grow - with updates, fixes, DLC, added features and more becoming common practice to delight players. Some would call game data analytics a new form of business intelligence, others a fundamental new approach to design and the developer-player relationship. No matter the labels used, game analytics is a rich new tool for developers in a rapidly changing world.
By using game analytics tools efficiently and effectively, game developers can test and uncover player experience issues (and work to fix them!) before a big global release, or rolling out the next update etc. The problem is that game analytics can be really confusing, overwhelming or feel really complicated to work with.
Persevering is worth it, and we’re working to break it down further to take the ‘blood magic’ reputation out of the game analytics world. Our goal is for every developer to feel comfortable learning and improving their player experiences by using a little bit of data science and creativity.
Here are some examples of common player experience problems that game analytics can help uncover:
Game analytics can give developers (you) unique insight into how your game is (or will) perform with players. By strategically using data, developers can look at game analytics to determine how their game is performing among players in several key areas of interest, as a way of determining ‘overall health’ of the current title.
Pick your weapon! For beginners, we suggest integrating with a free analytics tool that uses some default metrics. This may include a tool like Unity Analytics or Google Analytics. It seems that Facebook Analytics is also offering a very beginner-friendly toolset, and we will create an article in future to help you compare the different analytics tools out there - stay tuned! (Ps. We’ll email it to our mailing list first, so you can sign up if you don’t want to miss the comparison sheet.)
Before you set up your game analytics stack you should think about what key performance indicators (KPIs) you’ll want to use to track your game’s ‘health’.
As an example, retention related metrics are a must for games that monetize through both IAP AND advertisements, while monetization related metrics are important for games focusing mostly on IAP. There are other categories of metrics such as acquisition, engagement and virality which may be ideal for you, but as a start point we recommend the most fundamental metrics to track as a beginner are retention and monetization.
To use retention as an example, short term retention often indicates issues with early onboarding user experience:
Consider Day 1, 3 and 7 retention. Even if you have a strong Day 1 retention, pay particular attention to whether Day 7 comparing to Day 1 is too low. Rule of thumb is to seek around 50% of Day 1 Retention. If you are seeing below 50% Day 1 retention, that usually means players cannot see enough goals and variety of play experiences to continue playing the game.
Longer term retention drop-off often indicates issues with layers of gameplay, as well as the difficulty curve for the long term play experience. To evaluate this, look at the following areas:
Day 14 and 30 retention.
Monetisation reflects whether the items have the right appeal to the players. Often you need to pay attention to the placement of the items (ie. don’t hide the shop in your game too well, if you use one), the timing of the sell, and the layers of the items. To evaluate this, look at the following areas:
ARPPU and conversion rate (measured daily)
LTV (We will discuss how this can be calculated more easily in future feature articles!)
The first step in setting up a data analytics stack for your game is to ensure you track the right data points in your game. No amount of analytics can make up for poor quality source data, so it’s very important to structure your data points correctly from the beginning. The best time to do this is after your main game features are relatively stable within your game build, so you don’t change your data points constantly. You can wait to do this before the final testing prior to launch.
Data points in games are called events. They’re basically bunch of actions players take in game that are significant to you as the developer. While you could track every kind of data possible, your time is precious - selecting efficient and effective events to track is important to avoid distractions. It’s no use tracking heaps of data you don’t know how to analyse, as you may get drown in it. Analytics start with asking questions - Think about what questions YOU want answered. If you care about monetization, then you must include events such as IAP purchase events. The events you want to track will be very different for different business models, and certain game designs.
This article lists out some key events to track in general, which apply to a wide range of game developers - once you’ve launched (or soft launched) you can always add more events to answer a particular question.
Ok, so all the above is a really basic description of how to start looking and using some data analytics, no matter what kind of digital game you’re building. Like going to the gym, analytics is best used regularly, a little bit at a time. Read over our introduction whenever you need to refamiliarize yourself, and don’t be hard on yourself if it takes a few times to form game analytics habits.
The purpose of having some data is not for the sake of having it, but for you to be able to get into the measure, learn and build loop above. This is how game developers can innovate and consistently improve the player experience. Your game data can help you highlight how your players are experiencing your game, and shine a light on potential gaps you can address in your design - so it’s worth learning about! It could be as simple as an obstacle that’s placed at the wrong place that makes an early level too hard for your player. By helping you understand your player experience, you should be able to learn from mistakes quickly and improve your game for your players.
A typical analytics cycle (habits!) should look like the following:
Look at your health metrics. Have they changed in the right direction since your last update? Are there notable differences to the usual values? Using Unity Analytics Data Explorer you should find most default data points such as retention and ARPDAU. (Note you should segment platforms, eg. iOS and Android when you look at overall health data.)
This should point you to an area that requires your further attention.
Perform an analysis. The most fundamental ones are funnel analysis, which looks at the dropoffs at each progression point. For example, step 3 in your tutorial has more than usual dropoff rate. This level might require your attention. You should construct a tutorial funnel and a general progression funnel with significant gameplay milestones in Unity Analytics’ Funnel Analyzer.
Come up with a hypothesis. If level 3 in my tutorial has a larger than usual dropoff rate, what could be causing that? You could use more detailed events (e.g. death events in a level) to find out where players die. Or you could use player testing to monitor how real players play.
Release an update to affirm or deny the hypothesis. If you have some good ideas why players die, then it’s time for you to experiment with this. You could release an update to test this or use A/B testing to test several variants of it.
Then repeat the whole cycle again. Once you know that data is an honest agent reflecting how well you’re changing your game, it would become your best game development helper. Imagine combining your creativity with the power of data to guide you! It’s both a scary and exciting thing. Don’t be scared off from using it, just take it slow and experiment.
We are building a free dashboard so game developers can always get a health check for their games, with tips on how to improve their player experience and how to use their data effectively. Sign up to join our beta at www.secondsight.io or email us for more tips on game analytics. :)