Engagement is the most misunderstood and complex part of game analytics. The reason for this is that engagement is about fun and fun is something very difficult to infer. Average session length (measured in short units of time) and average session frequency (measured in a number of daily sessions per user) are often the metrics used to measure engagement. If we think about retention on a user count basis it is easy to see engagement as a session count and/or length basis. After all, the frequency at which players return and the amount of time they spend in the game should be good indicators of this, right?
As I see it, the difference between retention and engagement is not a matter of differentiating users from sessions. Retention is about returning to the game. Engagement is about interacting with it meaningfully. Sessions alone won’t tell you if players are having fun. Let me tell you a true story that will illustrate it perfectly. Although the story is a true one, data are illustrative.
The Tale of High Retention, Low Engagement Game
A couple of years ago I was looking at a game where retention was good but session length and frequency were… weird… There were two groups of sessions:
- Group A: 3 to 5 daily sessions per player averaging 6 minutes in length
- Group B: 1 daily sessions per player with several hours of length
The reason behind Group B behaviour was a timed soft currency offer. When the timer finished, the player could collect it, resetting the timer. Players in Group B would simply keep the game open to collect the soft currency while players in Group A would interact with the game as expected, so… actually playing it! Players on Group B wouldn’t even play the game most of the sessions. Although the retention mechanic (timed soft currency offer) was clearly working I had two questions to answer:
- Were the players that simply collected the soft currency (Group B) engaged?
- How could I measure engagement?
Why retention is not engagement
Let’s establish that, exceptions aside, if a player is engaged, she will return to the game. Therefor, if a player is engaged, a player is retained. The opposite might not be true as seen in the previous example, as extreme and rare as it might be. That wraps up question #1: the players in Group B were not engaged. What fun could there possibly be in pressing a button?
It is very important to establish this difference and not be fooled by high retention alone. If you read my previous posts, retention is king but it is only king because:
- Without players everything else is irrelevant
- It is easy to work with and quick to get actionable insights from
Engagement is the next in line to the crown. He is our prince charming and if we don’t raise him well our kingdom will be a poor one with angry peasants and a sad king. Drama aside, it is very important that we establish our engagment metrics and compare them to the other metrics. This leads us nicely to question #2.
How do we measure fun
Every step in the player lifecycle has well defined metrics. In the case of engagement every thing is on a per game basis with the exception of session frequency and session length. Each game has its own set of engagement metrics because each game has its own mechanics. Even games with similar mechanics have their own quirks unless it’s a pure reskin.
Ask yourself the following: Why do I expect my game to be fun? What will the players be doing in the game that it’s fun?
Let’s pick the Pedro’s drawing at the top to answer that. We are talking about a multiplayer racing game. The player races and depending on her finishing position she wins a given amount of credits. She can then upgrade her car or buy another car which will allow her to progress to more challenging races.
For starters we should be measuring this:
- daily races per player
- daily car upgrades per player
- daily car purchases per player
These are the three basic actions present in the game design loop. These are the starting point to understand directly from data how your players are having fun.
And then… complexity is born
For some games having the basic actions reported is enough but that is rarely the case. Many games have progression mechanics, balance and economy just to name a few. These games need a lot of dimensions added so we can analyse engagement as a function of those dimensions.
This is an exercise each producer, designer and analyst must do. It can get very complex, even for this simple hypothetical example of the racing game. Here are some dimensions that I could use to analyse engagement for this game:
- Adding the track to the race event would allow me to visualise a funnel of daily races per player per track which would give me a clear view on how players are evolving and where they are dropping.
- Adding the retention day to the race event (or joining queries, I’m not picky) would allow me to estimate how long the average player or segments of players take to progress through the game.
- Adding the information about finishing position, the car and its upgrades to the race event would allow me to find overlord strategies that unbalance the game.
- Adding the information of player wallets to the car upgrades and car purchases would allow me to identify player preferences and bottlenecks in the player progression that can have a severe impact on retention and engagement.
And I made up this list as I was writting it. It can get pretty complex really fast!
I’m not a fan of huge posts and this one is huge for my standards. I’m wrapping it up with the feeling I’ve barely scratched the surface of engagement. The reason being the complexity I mentioned in the first sentence of this blog post. Engagement measurement and analysis is unique to each game. What I wanted to achieve in this 101 was to establish that it is different from retention and that it is not just session metrics.
It is about fun! There’s no fun without meaningful interaction. Engagement is what happens when the players fall in love with the game. And love… well… like Facebook says, it’s complicated. In our case, complex.