作者：Trevor McCalmont

我爱数字！我要玩益智游戏、逻辑游戏或策略游戏直到世界的终结。我在大学时学的是数学，现在我成了一名数据分析师——我的工作就是成天与数字打交道！所以我居然会为一些度量指标感到烦恼，一定会让读者感到一点儿惊讶了。在手机游戏领域，我们总是用数字作报告，然而，那些数字往往传达不了任何可以评估的信息。

DAU/MAU

我们用日活跃用户（DAU）和月活跃用户（MAU）的比值来评估玩家在游戏中的沉浸感或游戏的粘性。在现在的手机游戏世界，DAU/MAU的比值最好超过0.2。如果今天登录游戏的人数大于过去30天里登录游戏的总人数的20%，并且这一情况持续了相当一段时间，那就说明玩家对这款游戏的沉浸感非常强。然而，成百上千的应用每天都在通过自然渠道或广告渠道开发新用户。所以问题就复杂了。

用我最喜欢的东西——数字来举例吧！比如在连续的两周内，游戏X有1千DAU和1万MAU。这款游戏的DAU/MAU的比值是0.1，表现不佳。游戏X的开发者决定开发2万名玩家，第二天他们有3万DAU和12万MAU。哇！现在DAU/MAU的比值是0.25。之前的游戏分析师错了！这款游戏的粘性明明高得不可思议！

等等，什么？游戏并没有什么变化。没有升级，没有新内容，相同的平衡性和进度。取决于游戏的留存率，这么高的DAU/MAU的比值只能维持几天或至多一两周。在这段时间内，天真的开发者或无知的投资者认为他们的游戏很好，不诚实的开发者可以说他们的游戏粘性非常强。

更理想的评估沉浸感的指标应该是登录次数/DAU。与DAU/MAU的比值相比，它体现的信息并无不同，都反映了用户在游戏中的沉浸感和与游戏的互动程度，但它所考虑到新用户的行为是不稳定的。

data-metrics(from profitinsight.com)

多余的数据点

当进行分析和事件追踪时，开发者必须问自己一个问题：“我想从中获知什么？”对于没有专职的数据分析师的小开发工作室，追踪得太多或太少的后果都会很严重。因为要从这么多或这么少的数据中找到有价值的信息，实在是件令人气馁的事。

几乎所有游戏开发者都希望通过分析来增加游戏的粘性、留存率和赢利。为了达到这些目标，你必须追踪的事件是：

菜单点击

你无法知道玩家点击菜单按键、空格键或静音键对提高留存率、沉浸感或赢利有何影响。一个弥补方案是，追踪玩家浏览商店的事件，但我认为这是没必要的。你的商店主页应该容易找到，任何页面上都应该有进入商店的按钮。通过赢利指标如转化率和IAP的每用户平均收益（ARPDAU），你可以知道玩家是否找得到商店和是否重视游戏货币。

消耗品的使用

至于消耗品，唯一重要的指标是玩家购买能量道具。无论玩家是购买消耗品囤着以后使用还是购买一些马上使用，都没关系，除非你打算通过预测性分析促进消费。

购买浏览、取消、失败

购买失败是一个QA问题，在游戏发布以前就应该解决。浏览和取消购买表明赢利指标较低，但更重要的是，这些并不能告诉你玩家的意图。那个玩家只是查看游戏中有什么道具？还是，他是需要更多钱的潜在购买者？

过分特殊的游戏机制

可以说，假设你制作一款赛车游戏，玩家会一直加档提速。那么，你将无法从那些执行“加速”事件的玩家中分类出付费玩家和非付费玩家，和将他们与少数下载后立即消费的玩家作比较。

关卡退出或重试

通过基本的漏洞分析可以很快确定某个关卡或任务是否存在平衡性问题。你必须检查的是各个关卡的完整性。退出关卡或重试关卡只会混乱你的事件追踪。

游戏排名

你可以轻易从任何应用商店获知该信息。这是另一个导致混乱的数据点。

为什么人人都爱K系数？

一方面，我认为K系数其实很愚蠢，因为它只反映用户分享游戏的频率。现在，人们对它的强调已经远远超过制作一款内容好、机制好的游戏了。只要人们总是制作免费游戏，玩家就会分享游戏。

另一方面，我认为手机游戏从业者应该先研究我们想评估的不同指标的意义。一旦我们掌握基本资料，我们当然就可以衡量游戏的“分享率”。

有些人把K系数看得很重要，但我不认为它对开发者开发一款免费游戏有什么实际帮助。归根到底，开发者要搞清楚的就是用这个指标对比什么，或这个指标有何用途。只有你理解了K系数再利用它，它的意义才大。

手机游戏在利用好内容创造有趣的体验方面，正在飞速进步。分析学、游戏数据和理解用户行为是游戏设计和创意的必要部分。不要害怕使用分析学，不要让你自己或你的团队陷入追踪多余数据的陷阱中。如果你不能证明你所追踪的指标，那么你就不应该追踪这些指标。（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）

Muddled Mobile Metrics

by Trevor McCalmont

I love numbers! I’ll play puzzle, logic, or strategy games until the cows come home. I studied math in college, and now I’m a data analyst – my job is to look at numbers all day! So it might come as a small surprise that there are metrics out there that annoy me. In the mobile space, we are reporting metrics that communicate very little information about what they are supposed to be measuring.

DAU/MAU

The ratio of daily active users (DAUs) to monthly active users (MAUs) is supposed to quantify the engagement or stickiness of a game. In mobile right now, a good DAU/MAU ratio is over 0.2. If the number of people who have logged in today is over 20% of the number of people who have logged in over the past 30 days, and this is consistently true over an extended period of time, that user base is definitely engaged with the game. However, every day thousands of apps are acquiring new users both organically and through advertising. This is where it gets tricky.

Let’s use an example with my favorite thing: numbers! Say Game X has 10,000 DAUs and 100,000 MAUs consistently for two weeks. This game is underperforming with a DAU/MAU ratio of 0.1. The developers of Game X decide to acquire 20,000 users and the next day they have 30,000 DAUs and 120,000 MAUs. BOOM! DAU/MAU ratio is now 0.25, that game analyst was wrong! This game has incredible engagement!

Wait, what? Nothing in the game changed. No update, no new content, same old balancing and progression. Depending on the retention of the game, they could maintain this higher DAU/MAU ratio for just a couple days or maybe even one to two weeks. During that time, a na?ve developer or innocent investor will think their game is great, and a devious developer can talk about their game with strong engagement.

A better measure of engagement is Sessions/DAU. It communicates much of the same information with respect to a user engaging and interacting with a game, and new user behavior is not volatile.

Unnecessary Data Points

When setting up analytics and event tracking, it is important to ask, “What am I trying to learn from this?” For small studios that don’t have a dedicated data analyst, tracking too much or tracking inefficiently is a recipe for disaster. It will discourage anyone from going back in and sifting through all the noise to find any actionable insights.

At a very high level, almost all game developers want to use analytics to increase engagement, retention, and monetization. With these end goals in mind, some events you do not need to track are:

Menu Clicks

There is no way that knowing a player tapped on menu buttons, pause, or mute would yield any insights toward improving retention, engagement, or monetization. One potential counterpoint is tracking when a user visits the store, but I even see this as unnecessary. Your storefront should be easy to navigate to, one click away from any screen, without being in the user’s face. Monetization metrics like Conversion Rate and Average Revenue Per DAU (ARPDAU) from In-App Purchases, will let you know if users can find the store and if they value the currency.

Use of Consumables

As far as consumables go, the only important element is that users buy the power-up. It does not matter if someone buys consumables and hoards them for later use or they buy a few and use them right away, unless you are using predictive analytics to push a promotion when someone runs out.

Viewed, Cancelled, Failed Purchase

Failed purchase is a QA issue and should be fixed before your game goes live in its respective marketplace. Viewed and cancelled purchases will be reflected in weak monetization metrics, but more importantly these do not tell you the user’s intent. Is that user just checking out what items exist in the game? Is that a potential buyer who needs more currency?

Overly Specific Game Mechanics

It’s a safe assumption that if you made a racing game, people will be shifting up. Everyone. The payers and the non-payers. You will not be able to segment those who have executed the event “Shift_Up” and compare them to the few players that immediately bounced after download.

Level Quit or Retry

If there is a problem with the balancing of a certain level or quest, a basic funnel will quickly identify this. All that is necessary to track is each level complete. Quitting a level or retrying a level will only muddle your event tracking.

Rate Game

You can easily obtain this information from any app marketplace. This is another example of a data point that creates clutter.

Why Does Everyone Love K-Factor?

Part of me thinks I should say K-Factor is silly because it only reflects how often users share the game, and right now it is way more important to focus on making a great game with great content and compelling monetization hooks. Once people are consistently making great free-to-play games, users will share those.

Part of me thinks we in the mobile space should be looking ahead to all the different things we want to measure, and once we have mastered the basics, sure we might want to measure the “sharability” of a game.

K-Factor is important to some, but I can’t see how it can help developers succeed with free-to-play games. It comes down to what this metric is being compared against or what is it being used for. If you currently use and understand K-Factor, more power to you.

The mobile space as a whole is making great strides towards creating fun experiences with great content. Analytics, game data, and understanding user behavior are an incredible complement to game design and creativity. Don’t be afraid to start using an analytics product, and don’t overwhelm yourself or your team by tracking unnecessary data points. If you can’t prove something with the metrics you are tracking, you should not track those metrics.(source:gamasutra)