And so it begins. Acquisition is that special moment where someone opens our game for the first time. As magic as it is (and it is!) there’s quite a bit going on. Acquisition is important in a couple of different areas: the first, more business oriented, is marketing and user acquisition. The second is early knowledge of the player. Let’s begin.
User Acquisition and the different acquisition channels
UA is a world on its own. It influences and is influenced by The Player Lifecycle but it is not part of it. The Player Lifecycle is about user behaviour. UA is about marketing. However, UA serves many purposes. From getting enough users to A/B on soft launch to ROI-based campaigns, there are many use cases. The users acquired through UA have behaviours and we should segment and analyse these segments to support UA.
To understand how UA works and why it interests us, there are some basic things we need to know. For starters, there are three acquisition channels:
- Paid Acquisition is when we spend money to acquire users. This is all things marketing or UA.
- Viral Acquisition is when new users are acquired by the actions of existing users. More on this in the upcoming Virality 101.
- Organic Acquisition is everything else. From players that search for our game in the stores to the players that hear about the game from a friend, or saw it in the store charts. If we can’t attribute a source, it is considered organic.
These channel types are not closed silos. Each install brings with it a rate of viral and organic installs. This rate influences the cost of paid acquisition and the Lifetime Value (LTV) of the cohort if we account the indirect Average Revenue per User (ARPU). It is, therefore, relevant that we correctly attribute each user to the acquisition source.
Let me give you an example with a simple (and quite irrealistic) scenario. You acquired 100 users at $1 per install. Your ARPU is $0.75. Bad news, your paid acquisition will cost you $0.25 per player. However, those 100 users gave you top #5 position in the free to download charts. Your models tell you that with that position you have 50 more new organic players on average. At the same ARPU, your acquisition cost of $100 has a total projected revenue of $112.5.
This should give you an idea of how important it is to attribute a source to a user. Don’t be fooled by the example. It is an oversimplification for the sake of illustration.
Knowledge: what do we want to know?
As far as The Player Lifecycle goes, there are three fundamental things that we want to know about a new user:
- Who is she?
- When did she start playing?
- How did she find the game?
Who she is is quite a profound question. First, we want an anonymised unique identifier. Server side games should have an easy time, single player games not really. Second, we want to gather additional dimensions for each user in order to segment users, like geographic and demographic data. The “urgency” to gather these data depends on what we want to do with it. As an example, having a randomised controlled trial system that supports new user segments needs to “know” new user segments much faster than a reporting system that will probably update overnight.
There are some idiosyncrasies to When she started playing but overall it is easy. It is the date (or timestamp) where the player first opened the game. How we do it has some consequences. We can keep it clean: define a specific event (session start for instance) and get a timestamp of that event. This might bring some problems, as an example, if there’s a telemetry issue, we might have events that are older than the first timestamp. The other, dirtier and more computational intensive way in my humble opinion, is to get the first timestamp of the user on any event.
How did she find the game is tricky. This is relevant for UA explained earlier in the post. We must guarantee that we are able to cross reference the ID used by a tracking partner with an ID we receive from the game. If the user ID that we use to uniquely identify the user is the device ID also used by the tracking partner, we are good to go. The catch, in this case, is that the same user can have several devices (an iPhone and an iPad for instance). Not only we will be over counting users but we may be undervaluing cohort monetisation which is vital for UA. On the other hand if we are using a user ID and not a device ID we will have unique ID counts but extra work is needed to be sure that we get the correct attribution data, by crossing the device ID received from the user with the device ID received from the tracking partner. We may also have to deal with having multiple attribution sources for the same user.
In case something needs clarification check the Glossary.
Why does this matter?
As I mentioned, there are two interests regarding Acquisition.
The first one is marketing, meaning, UA. Discoverability is one of the biggest problems that publishers face so understanding how the player is acquired, the ROI of campaigns and the interaction between the types of acquisition channels is vital on many levels.
The second one is knowledge. To be able to create segments of players is very powerful and to do it we need dimensions. Many of these dimensions are only known later so depending on the technical needs, early or even real-time data might be needed.
In the end, thinking about Acquisition is about identifying needs and making decisions around what we need to know and how we’ll need to use it. It is fairly easy to grasp as a concept, but it can be quite tricky to implement.