在产品管理中两种不同的驱动方法：数据驱动的工程方法 vs 直觉驱动的艺术方法---谷歌&苹果---两种产品文化的故事

Google and Apple are two of the most successful and admired technology companies, yet their approach to product development differs in some fundamental aspects. Google's is based heavily on data and experimentation and might be called the engineering approach. Apple's relies more on vision and intuition and might be called the liberal arts approach.

谷歌和苹果是两家最成功和最受尊敬的科技公司，但它们的产品开发方法在一些基本方面不同。谷歌的方法主要基于数据和实验，可以称为工程方法。苹果的则更依赖于远见和直觉，可以被称为艺术方法。

In a recent article by Marty Cagan, product management expert and author of “Inspired—How to Create Tech Products Customers Love”, he observes the following about difference in leading tech companies' product cultures:

产品管理专家、《启示录-如何创造客户喜爱的科技产品》一书的作者Marty Cagan在最近的一篇文章中，就领先科技公司的产品文化差异观察到以下几点：

I talk a lot about product culture and how important it is, but one inconvenient fact has always bothered me, which is that my favorite product companies – for example, Google, Apple and Amazon – all have such different cultures.

我经常谈论产品文化及其重要性，但有一个麻烦的事实一直困扰着我，那就是我最喜欢的产品公司——例如谷歌、苹果和亚马逊——都有如此不同的文化。

The difference between Apple and Google in particular is illustrated in the following tweet by Ken Kocienda, author of the book “Creative Selection: Inside Apple's Design Process During the Golden Age of Steve Jobs”.

《创造性选择：乔布斯黄金时代苹果设计过程的内幕》一书的作者Ken Kocienda在下面的推特中阐述了苹果和谷歌的区别。

We didn't use A/B tests to make the iPhone at Apple. pic.twitter.com/Sjo2Oa2Eiw— Ken Kocienda (@kocienda) April 3, 2019

我们在苹果制作iPhone时并没有使用A/B测试。

Of course, this quote perfectly matches what one would expect Apple's approach to product development to look like, and the way the author throws shade at Google's “41 shades of blue” test demonstrates a lot of disdain for that example of data-driven product decisions taken to the extreme.

当然，这句话完全符合人们对苹果产品开发方法的预期，而作者对谷歌的“41种蓝色的阴影”测试的蔑视表明，很多人对这种数据驱动的产品决策的例子不屑一顾。

These two companies, Apple and Google, epitomize the difference between quantitative and qualitative thinking in product development philosophies, cultures, and processes. Google's approach might be called the engineering approach, and Apple's the liberal arts approach. Obviously, both approaches can be successful: Apple and Google are two of the most successful, valuable, and admired companies in the world.

这两家公司，苹果和谷歌，集中体现了产品开发哲学、文化和过程中定量思维和定性思维之间的差异。谷歌的方法可以被称为工程方法，而苹果的方法被称为艺术方法。显然，这两种方法都可以成功：苹果和谷歌是世界上最成功、最有价值和最受尊敬的两家公司。

Google's data-driven “engineering” approach

谷歌数据驱动的“工程”方法

Google is first and foremost a data company. From its mission to “organize the world's information and make it universally accessible and useful” over its original product, the search engine, to the technologies such as BigTable or map/reduce that it has perfected for handling massive amounts of data, Google's focus and core competency is handling data (lots of it).

谷歌首先是一家数据公司。谷歌的使命是通过其原始产品搜索引擎“组织全世界的信息，使其普遍可访问和有用”，再到BigTable或map/reduce等为处理大量数据而完善的技术，谷歌的重点和核心竞争力都是处理数据（大量数据）。

With this focus on data comes a natural tendency to trust data for making product decisions. If you have such a treasure trove of data available at your fingertips, it makes sense to use it when developing or improving products: if you have enough users (and Google certainly does), you can A/B test almost anything, including shades of blue.

随着对数据的关注，人们自然倾向于在做产品决策时信任数据。如果您手边有这样一个数据宝库，那么在开发或改进产品时使用它是有意义的：如果您有足够多的用户（谷歌当然有），那么您几乎可以进行A/B测试，包括shades of blue。

It is also of course not a coincidence that with Google Analytics / Optimize, Google is offering the entry level A/B testing tool that a lot of companies use before moving on to more sophisticated ones.

谷歌Analytics / Optimize提供入门级A/B测试工具，许多公司在转向更复杂的工具之前都会使用它，这当然也不是巧合。

Beyond just being very data-driven and A/B test heavy, Google has also displayed a very low barrier to launching new products. Google might be the tech company that has launched and killed the highest number of products (some with fiercely loyal fans): Inbox, Google Reader, Google+,... The list goes on. Obviously, part of Google's approach is to launch quickly and kill quickly (presumably, based on usage data).

除了非常依赖数据和A/B测试外，谷歌在发布新产品时也显示出了非常低的门槛。谷歌可能是推出和淘汰产品数量最多的科技公司（其中一些产品拥有狂热的忠实粉丝）：收件箱、谷歌阅读器、谷歌+……这样的例子不胜枚举。显然，谷歌方法的一部分是快速启动和快速终止（根据使用数据推测）。

Google is also famous for its management framework “Objectives and Key Results” (OKRs), which was conceived at Intel but perfected at Google. Management through OKRs involves setting priorities in the form of “objectives” (for example, “Improve new user onboarding”), and defining measurable “key results” that determine whether progress toward the objective was made (for example, “increase completion rate of the signup flow from 80% to 85%”). Given the focus on measurable results, OKRs more naturally favor a data-driven or data-informed way of product development.

谷歌还以其管理框架“目标和关键结果”（OKRs）而闻名，该框架是在英特尔提出的，但在谷歌得到了完善。通过OKR进行的管理包括以“目标”的形式设定优先级（例如，“提高新用户的入门水平”），并定义可衡量的“关键结果”，以确定是否实现了目标（例如，“将注册流的完成率从80%提高到85%”）。考虑到对可测量结果的关注，OKR更自然地倾向于数据驱动或数据为准的产品开发方式。

More recently, one of Google's biggest core competencies is machine learning (again, driven by Google's massive amounts of data). Machine learning makes (qualitatively) reasoning through product decisions harder: you can't think through and actively design all “rules” of the model, or even validate what outputs the model will produce in all possible circumstances. On the other hand, it makes building data-driven feedback loops even easier: they can feed back right into the model without any human intervention—the product improves itself.

近来，谷歌最大的核心竞争力之一是机器学习（同样，由谷歌的海量数据驱动）。机器学习使得通过产品决策进行（定性的）推理变得更加困难：你无法思考并积极地设计模型的所有“规则”，甚至无法验证模型在所有可能的情况下将产生什么输出。另一方面，它使构建数据驱动的反馈循环更加容易：它们可以直接反馈到模型中，而不需要任何人为干预——产品会自我改进。

Google can't possibly “design” the search results (including rich previews etc.) for every possible search query, but it can design a machine learning algorithm that gets better with every user interaction: the user behavior on the page, including which search result the user clicks, can inform what results future users will be shown for a similar query. This is data-driven product development driven to the extreme. Therefore, the application of machine learning fits extremely well with Google's product development approach and even reinforces it.

谷歌不可能为每一个可能的搜索查询“设计”搜索结果（包括丰富的预览等），但它可以设计一种机器学习算法，随着每次用户交互而变得更好：用户在页面上的行为，包括用户单击的搜索结果，可以告知未来用户在类似查询中将显示什么结果。这是数据驱动的产品开发的极致驱动力。因此，机器学习的应用非常适合谷歌的产品开发方法，甚至可以加强它。

In summary, Google's products, core competencies, and management techniques are all very much aligned around an extremely data-driven approach to product development.

总之，谷歌的产品、核心竞争力和管理技术都非常紧密地围绕着一种数据驱动的产品开发方法。

Apple's intuition-driven “liberal arts” approach

苹果的直觉驱动的“艺术”方法

Apple is at its core a hardware company. Of course, Apple also makes a lot of software, but most of it is very tightly coupled to the hardware and/or intended to make the hardware more valuable. Hardware development by nature is less iterative and requires a greater amount of “perfection” once the product ships. Software can be patched after release, even more so in Google's case where a lot of the software is run on Google's servers, but a flaw in the design of the hardware can't be ironed out after launch. Apple brings this scrutiny in the design process to its software development efforts too.

苹果的核心是一家硬件公司。当然，苹果也生产很多软件，但大多数软件都与硬件紧密相连，或旨在提高硬件的价值。硬件开发本质上迭代较少，一旦产品发布，就需要更多的“完美”。软件可以在发布后进行修补，谷歌的情况更是如此，因为它的很多软件都运行在谷歌的服务器上，但硬件设计中的缺陷在发布后无法解决。苹果公司在软件开发过程中也会对设计过程进行审查。

Apple's hardware products are also “luxury products”, or at least high-end products. If you want the most affordable or value-for-money phone, computer, media player, etc. you are not going to buy the Apple version of the product. This means that Apple can command a price premium, but it also means that customer expectations are that the products work flawlessly, or at least attain a higher level of perfection than competitor products.

苹果的硬件产品也是“奢侈品”，至少是高端产品。如果你想要最实惠或最物有所值的手机、电脑、媒体播放器等，你就不会买苹果版本的产品。这意味着苹果可以获得价格溢价，但这也意味着消费者的期望是，产品运行起来完美无瑕，或至少达到比竞争对手产品更高的完美水平。

In “Creative Selection”, Ken Kocienda summarizes his view of the Apple way of software product development:

Ken Kocienda在“Creative Selection”中总结了他对苹果软件产品开发方式的看法：

A small group of passionate, talented, imaginative, ingenious, ever-curious people built a work culture based on applying their inspiration and collaboration with diligence, craft, decisiveness, taste, and empathy and, through a lengthy progression of demo-feedback sessions, repeatedly tuned and optimized heuristics and algorithms, persisted through doubts and setbacks, selected the most promising bits of progress at every step, all with the goal of creating the best products possible.

一小群热情、有才华、有想象力、有独创性、充满好奇心的人建立了一种工作文化，基于他们的灵感和勤奋、工艺、果断、品味和同情心，通过漫长的演示反馈会议，反复调整和优化启发和算法，在怀疑和挫折中坚持，在每一步都选择最有前途的进展，所有的目标都是创造最好的产品。

Interestingly, like Google, Apple also uses an iterative, evolutionary approach to product development. However, the evolution is not achieved by building small increments, releasing them as experiments, and gathering data about how successful they were in moving some KPI. Rather, the evolution is internal, through a process of demo sessions with leadership which provides both feedback and the vision for where the product should eventually end up.

有趣的是，像谷歌一样，苹果也使用迭代的、渐进的方法进行产品开发。然而，通过构建小增量，将其作为实验发布，并收集关于移动某些KPI的成功程度的数据，并不能实现演进。相反，演进是内部的，通过一个演示会议的过程，领导提供反馈和愿景的产品最终应该在哪里。

This internal evolution is not driven by data, nor even by direct feedback from users and customers. Instead, it is driven by empathy and taste, two very qualitative means of making decisions. In Ken Kocienda's words:

这种内部演进不是由数据驱动的，甚至也不是由用户和客户的直接反馈驱动的。相反，它是由同理心和品味驱动的，这是两种相当定性的决策手段。用Ken Kocienda的话来说：

[Empathy means] trying to see the world from other people's perspectives and creating work that fits into their lives and adapts to their needs. […] Taste is developing a refined sense of judgment and finding the balance that produces a pleasing and integrated whole.

(同理心意味着)试着从别人的角度看世界，创造出符合他们生活和需求的作品。品味是发展一种精致的判断力，并找到一种平衡，从而产生令人愉悦的整体。

Apple's feedback loops, and therefore the entire way of product development, is hence more qualitative, and perhaps even more subjective.

因此，苹果的反馈循环，以及整个产品开发方式，更加定性，甚至可能更加主观。

While you perhaps need a visionary leader like Steve Jobs to maximize the impact of this qualitatively driven liberal arts approach to product development, the evolutionary approach of developing demos and gathering feedback on them has worked well for Apple, and allowed them to consistently ship products that are of higher quality and more “tasteful” than their competitors.

虽然你可能需要一个像史蒂夫·乔布斯这样有远见的领导者来最大化这种定性驱动的艺术方法对产品开发的影响，但开发演示并收集反馈的进化方法对苹果来说很有效，并允许他们持续发布比竞争对手更高质量和更“有品位”的产品。

One aspect of this difference in product cultures that I find particularly interesting is how this manifests itself in the development ecosystems of these two companies—after all, they own the vast majority of the mobile platform market. I witnessed one anecdote illustrating this in a workshop about in-app subscriptions, hosted by Apple. One participant brought up the question how pricing A/B tests should be run on iOS—and another (non-Apple) participant responded by saying that the best way of running pricing A/B tests was to run them on Android. Given the background of these two different product cultures, the reason becomes clearer: Apple does not really believe in A/B testing as a necessary product development tool, but Google does, so it's not a surprise that Google's developer tools more easily enable A/B testing than Apple's.

我发现这种产品文化差异的一个特别有趣的方面是，这两家公司的开发生态系统是如何表现出来的——毕竟，他们拥有绝大多数的移动平台市场。在苹果主办的一个关于应用内订阅的研讨会上，我目睹了一件能说明这一点的轶事。一位与会者提出了如何在ios上运行A/B定价测试的问题，而另一位（非苹果）与会者回应称，运行A/B定价测试的最佳方式是在Android上运行。考虑到这两种不同的产品文化背景，原因就变得更清楚了：苹果并不真的认为A/B测试是必要的产品开发工具，但谷歌却这么认为，所以谷歌的开发工具比苹果的更容易实现A/B测试也就不足为奇了。