作者：Mark Robinson

过去，我们发行游戏，阅读评论，摒除其中不足，然后从新开始。现在情况不再如此。游戏会不断发展，用户持续告诉你自己的体验看法，他们总是呈现无穷尽的数据。玩家表达自己的意见，陈述自己的预期体验方式。

但你是否真的有听取他们的意见？你是否真的理解玩家行为？这些数据是成真的童话美梦，还是只是持续不断的噩梦？

现在游戏会将玩家的关卡活动数据输入分析数据库。通常复杂游戏会有100种不同类型的活动。各类活动，如“选择武器”、“邀请好友”或“开始任务”都和数据参数相关，这样我们就能够详细追踪玩家的历史记录和状态。若游戏有100万MAU，每位玩家都会生成多种活动记录，那么游戏每分钟的玩法体验将涉及众多数据。

那么我们该如何诠释这些数据，真正把握玩家的动机和潜能？

答案显然不在于控制台参数。理论上看，只要我们能够瞄准正确的参数，那我们就能够找到“下金蛋的鹅”，充分把握玩家。但遗憾的是，此参数并不存在。在现实生活中，参数总是有些滞后——它们呈现已发生的情况（游戏邦注：或好或坏），但它们不会告诉你下步该怎么做。

那么我们应该从哪里着手？下述5个基本法则能够帮助我们生成有效分析数据，深入把握玩家。

magnify(from gemini-systems.com)

1）收集适量的活动数据——既不要过多，也不要过少（以免无法深入把握分析数据）。

要确定活动数据的收集方式，我们需要做出若干重要决策。操作错误就会遇到绊脚石，所以我们应该给各数据收集设定合理理由。问问自己：“此数据项能够让我们传递什么涵义？”不要笼统收集数据，认为其将来也许会派上用场。例如，在FPS游戏中，你定不想要记录玩家所进行的所有射击，但你会希望收集重新填充的弹药。我们应花时间思考如何巧妙收集活动数据。

2）分解目标

现在我们来解决留存率问题！但留存率问题涉及多种类型，我们需要进行区分，方能最终解决此问题。你解决的是早期留存率问题（游戏邦注：换而言之就是辨别还未融入游戏就在初期关卡中离开的玩家）？还是游戏中的“叛变”——之前的忠实玩家出于某些原因变成非固定玩家；还是后期“叛变”，也许玩家玩完关卡，需要新功能？

各独立目标都有自己的特定关注焦点和对应解决方案。

3）保持多变量

此法则的意思是，你需要综合把握数据，而不是逐一查看。通过运用统计和可视化技术，我们可以综合把握玩家的各种行为，分析他们的玩法。整体把握玩家的体验周期既颇具趣味，又能够换来商业回报。这样你就能够清楚把握玩家行为，从中获得回报。

4）耐心等待创收时机

其魅力在于持续联系玩家——在游戏中频繁发送邀请和奖励。但我们清楚投资步伐缓慢的玩家通常会花大把时间融入游戏，所以当他们需要掏钱时，他们依然会继续体验。我们不应追求初期游戏体验的收益，应保持耐心，在玩家完全融入其中的时候，才发送购买道具的对话框。

5）以玩家为中心

以玩家为中心，而不是游戏，将此默念10遍。

若你坚持上述5条法则，游戏将能够蓬勃发展。（本文为游戏邦/gamerboom.com编译，拒绝任何不保留版权的转载，如需转载请联系：游戏邦）

The Fairytale Rules of Exceptional Analytics

by Mark Robinson

Once upon a time, you shipped a game, read the review and shaved off the beard then started all over again.  Not anymore. Now the game never ends and the world is trying to tell you what they think about your game as it spews out endless data. Players are proclaiming what they like, what they don’t like and exactly how they want to experience your games.

But are you really listening?  Do you truly understand your player behaviours? And is all this data a fairytale dream come true or a recurring nightmare?

Games are configured to write player level event data into analytics databases.  Typically a relatively complex game may have 100 different types of events defined.  Each of these events, such as ‘Select Weapon’, ‘Invite a Friend’ or ‘Start Mission’ will have associated data parameters so that player histories and status can be tracked in detail. A game with 1,000,000 MAUs, each generating multiple event records for each and every minute of gameplay soon means you are awash with data.

So how can we interpret this and really start to understand our players’ motivations and potential?

The answer surely does not lay with dashboard metrics.   The perception has been that if only we could think of the right metric, then we can unleash the golden goose and unlock a universal understanding of players.  But, sadly this metric does not exist. The reality is that metrics look backwards – they tell you what has happened (good or bad) but they don’t tell you what to do next.

So where to start? There are five fundamental rules for producing effective and actionable analytics that will help you start to unlock a deeper understanding of your players.

1) The Goldilocks Rule: collect the right amount of event data – not too much to become overwhelmed, but not too little to limit the depth of analysis.

There are some critical decisions to make when configuring event collection.  Getting this wrong could be career limiting so have a good reason for each event. Ask yourself:  ‘what value will this data item allow us to deliver?’ And definitely don’t default to collect everything just because it might be useful sometime, somewhere …For example on a FPS, you probably don’t want to record every shot taken but you may want to collect each ammo reload.  Being smart about event collection is worth the effort.

2) The Three Little Pigs Rule: break down the objective

Let’s solve retention! OK fine, but there are different types of retention issues and we need to differentiate them to really blow down this particular house. Are you solving early game play retention, in other words identifying players that leave in the early levels before engaging with the game; or is it mid-game defection – where previously engaged players have, for some reason, become sporadic players; or is your problem late defection when perhaps players are running out of levels and new features?

Each has separate objectives with a different focus, approach and solution.

3) The Seven Dwarfs Rule: Be multi-variate

This textbook term simply means you to need to look at all the data together not metric by metric.  By using statistical and visualisation techniques the combination of each player’s behaviour can be analysed to characterise their game play.  Being holistic about each player’s playing lifecycle is both fascinating and commercially rewarding.  This is how you unlock a clear understanding of player behaviours and drive revenues.

4) The Tortoise and The Hare Rule: Bide your time to monitize

The temptation is to contact every player about everything – spamming offers and rewards around the game.  However we know that players who are slow to monetize have often spent considerable time engaging with the game so when they do spend they then continue to spend.  Instead of trying to chase revenues in early game play, be patient and develop a dialog with players by only targeting purchase offers when they are truly engaged.

5) The Sleeping Beauty Rule: be player centric

Think players not game. Repeat this ten times.

If you abide by these five rules then your game will live happily ever after.（Source：gamasutra）