第九封信中英文对照：

To our shareholders:
致我们的股东：

Many of the important decisions we make at http://Amazon.com can be made with data. There is a right answer or a wrong answer, a better answer or a worse answer, and math tells us which is which. These are our favorite kinds of decisions.
Amazon做出的许多重要决策，都是基于数据。数学将告诉我们，哪一个是正确答案，哪一个是错误答案；哪一个是更好的答案，哪一个是较差的答案。我们喜欢这样子做决定。

Opening a new fulfillment center is an example. We use history from our existing fulfillment network to estimate seasonal peaks and to model alternatives for new capacity. We look at anticipated product mix, including product dimensions and weight, to decide how much space we need and whether we need a facility for smaller “sortable” items or for larger items that usually ship alone. To shorten delivery times and reduce outbound transportation costs, we analyze prospective locations based on proximity to customers, transportation hubs, and existing facilities. Quantitative analysis improves the customer’s experience and our cost structure.
设立新的出货中心就是一个例子。我们使用现有物流网络中的历史记录，来估算季节性高峰，并建立模型寻找解决需求的方案。我们会预估产品的组合，包括尺寸和重量，以决定我们需要多少空间。同时，评估我们是否要为体积较小的「可分类」物品，或通常单独运送的较大物品建立额外的设施。为了减少运送次数和出站运输成本，我们根据与客户、交通枢纽和现有设施的距离，来分析预期的设立地点。这样的定量分析，可以改善客户体验和我们的成本结构。

Similarly, most of our inventory purchase decisions can be numerically modeled and analyzed. We want products in stock and immediately available to customers, and we want minimal total inventory in order to keep associated holding costs, and thus prices, low. To achieve both, there is a right amount of inventory. We use historical purchase data to forecast customer demand for a product and expected variability in that demand. We use data on the historical performance of vendors to estimate replenishment times. We can determine where to stock the product within our fulfillment network based on inbound and outbound transportation costs, storage costs, and anticipated customer locations. With this approach, we keep over one million unique items under our own roof, immediately available for customers, while still turning inventory more than fourteen times per year.
同样的，我们大多数的购买库存的决策，都可以进行数值建模和分析。我们希望库存产品能够实时提供给客户，也希望库存量尽可能少，以维持较低的库存成本，进而提供客户较低的商品价格。要实现两者的平衡，需要有适量的库存。我们运用历史数据，来预测客户对商品的需求，以及需求的变化程度。我们运用供货商的历史数据，来预测补货时间。我们根据入站和入站的运输成本、库存成本和预期的客户所在位置，来决定我们的库存要放在哪个出货中心。如此一来，我们可以保留超过100万件库存商品，实时供货给客户，又可以实现每年14次以上的周转次数。

The above decisions require us to make some assumptions and judgments, but in such decisions, judgment and opinion come into play only as junior partners. The heavy lifting is done by the math.
上述决策需要先做出假设和判断，但在这样的决策中，个人的判断和意见只是基础要素，更重要的是数学构成的理性分析。

As you would expect, however, not all of our important decisions can be made in this enviable, math-based way. Sometimes we have little or no historical data to guide us and proactive experimentation is impossible, impractical, or tantamount to a decision to proceed. Though data, analysis, and math play a role, the prime ingredient in these decisions is judgment.
然而，正如你可能预期的那样，并非所有重要的决策，都可以透过这种基于数学的方式做出决定。有时，我们只有很少的历史数据可以指导我们，而进行相关实验是不切实际的，或者对于做出决策没有帮助。尽管数据、分析和数学起了重要作用，但这些决策的主要原因还是判断力。

As our shareholders know, we have made a decision to continuously and significantly lower prices for customers year after year as our efficiency and scale make it possible. This is an example of a very important decision that cannot be made in a math-based way. In fact, when we lower prices, we go against the math that we can do, which always says that the smart move is to raise prices. We have significant data related to price elasticity. With fair accuracy, we can predict that a price reduction of a certain percentage will result in an increase in units sold of a certain percentage. With rare exceptions, the volume increase in the short term is never enough to pay for the price decrease. However, our quantitative understanding of elasticity is short-term. We can estimate what a price reduction will do this week and this quarter. But we cannot numerically estimate the effect that consistently lowering prices will have on our business over five years or ten years or more. Our judgment is that relentlessly returning efficiency improvements and scale economies to customers in the form of lower prices creates a virtuous cycle that leads over the long term to a much larger dollar amount of free cash flow, and thereby to a much more valuable Amazon.com. We’ve made similar judgments around Free Super Saver Shipping and Amazon Prime, both of which are expensive in the short term and—we believe—important and valuable in the long term.
正如股东所知道的，由于我们的效率和规模做得到，我们决定要持续为消费者尽可能的降价。这是非常重要的决策，而且没办法根据理性的数学方法做出决定。事实上，我们持续降低价格的做法，和基于数学的分析结论是对立的。数学方法告诉我们，最聪明的作法是涨价。我们有与价格弹性相关的数据，可以在一定范围内，预测价格下降的百分比与销售额上升的百分比之间的关系。除了极少数的例外，销售额的短期增量，很难弥补价格下降带来的损失。然而，我们的定量分析，只能用于短期。我们可以估计价格这周和这个季度的降幅，但是我们没办法预估长期持续降价政策下，五年、十年以后会带来什么结果。我们的判断是，持续将效率进步的益处分享给消费者，藉此创造出庞大的经济规模，长期来说，这样的循环会给我们带来大量的现金流，因此可以创造一个更有价值的Amazon。我们对Free Super Saver Shipping服务、Prime服务都有类似的判断。这两项服务，在短期来说都是昂贵的投资，但我们认为，这两者长期来看都会是重要且有价值的。

As another example, in 2000 we invited third parties to compete directly against us on our “prime retail real estate”—our product detail pages. Launching a single detail page for both Amazon retail and third-party items seemed risky. Well-meaning people internally and externally worried it would cannibalize Amazon’s retail business, and—as is often the case with consumer-focused innovations—there was no way to prove in advance that it would work. Our buyers pointed out that inviting third parties onto http://Amazon.com would make inventory forecasting more difficult and that we could get “stuck” with excess inventory if we “lost the detail page” to one of our third-party sellers. However, our judgment was simple. If a third party could offer a better price or better availability on a particular item, then we wanted our customer to get easy access to that offer. Over time, thirdparty sales have become a successful and significant part of our business. Third-party units have grown from 6% of total units sold in 2000 to 28% in 2005, even as retail revenues have grown three-fold.
另一个例子是，2000年，我们邀请了第三方卖家进驻Amazon，在我们最重要的资产，也就是产品详情页上面与我们竞争。让第三方卖家和Amazon使用同一款产品详情页，看起来是一件很冒险的事。有些好心人认为，这会蚕食我们的零售业务。而且，以消费者为中心的创新，通常很难在事前就证明它会起作用。我们的买家指出，邀请第三方卖家进驻Amazon，而且第三方卖家表现比我们更好时，会使我们的库存预测难以进行，Amazon将会陷入库存过多的泥淖中。但是，我们的判断其实很单纯，如果第三方卖家提供比我们更好的商品、更低的价格，那么我们希望消费者可以轻易买到第三方卖家的东西。长期来说，第三方卖家将会成为Amazon业务中不可或缺的一部分。在零售收入提升3倍的背景下，第三方卖家的销售额占比从2000年的6％，提升到2005年的28％。

Math-based decisions command wide agreement, whereas judgment-based decisions are rightly debated and often controversial, at least until put into practice and demonstrated. Any institution unwilling to endure controversy must limit itself to decisions of the first type. In our view, doing so would not only limit controversy —it would also significantly limit innovation and long-term value creation.
基于数学的决策，通常能够形成广泛的共识。相对来说，基于价值观判断的决策经常是彼此矛盾的，在公诸于世之前只经过适度的辩论，所以无法像基于数学的决策取得广泛共识。任何一间不愿意接受矛盾的公司，都会把自己限制在极度安全的决策上，也就是基于数学的决策。而我们认为，这么做不只是减少矛盾，也在极大程度上也减少了创新与创造。

The foundation of our decision-making philosophy was laid out in our 1997 letter to shareholders, a copy of which is attached:
我们在1997年发布的第一封致股东信，阐述我们的决策哲学。这封信的要点节录：
• We will continue to focus relentlessly on our customers.
• 我们将会持续把注意力放在客户身上。

• We will continue to make investment decisions in light of long-term market leadership considerations rather than short-term profitability considerations or short-term Wall Street reactions.
• 我们将继续根据长期的市场领导地位，而不是短期的获利能力，或华尔街的短期反应，来进行投资决策。

• We will continue to measure our programs and the effectiveness of our investments analytically, to jettison those that do not provide acceptable returns, and to step up our investment in those that work best. We will continue to learn from both our successes and our failures.
• 我们将继续评估和分析计划和投资的有效性，抛弃那些无法提供可接受报酬的计划，加大投资成效最好的计划。无论是成功或失败，我们将继续从中学习。

• We will make bold rather than timid investment decisions where we see a sufficient probability of gaining market leadership advantages. Some of these investments will pay off, others will not, and we will have learned another valuable lesson in either case.
• 若有足够高的可能性可获得市场领导优势，我们选择做出大胆，而不是保守的投资决定。这些投资中，有一部分会有回报，有一部分不会。无论是哪种情况，我们都会学到宝贵的经验。

You can count on us to combine a strong quantitative and analytical culture with a willingness to make bold decisions. As we do so, we’ll start with the customer and work backwards. In our judgment, that is the best way to create shareholder value.
你可以期待我们将强大的定量分析能力和愿意冒险的价值观相结合。当我们这么做时，我们将一如往常的聚焦于客户。根据我们的判断，这是创造股东价值的最佳途径。

Jeffrey P. Bezos
Founder and Chief Executive Officer
杰夫·贝索斯
Amazon创始人暨CEO

我的观点

注意这是2005年，深度学习还没突破。当时所有的机器学习算法都属于传统算法。

贝索斯强调，他们很早就开始在使用数据分析来做决策。注意，这里是数据分析。然后根据分析结果做决策，也算是数据驱动。

他举了几个例子，很多属于分析，还有一些属于模型预测问题。这在当时，应该算不错的思路，用来减少成本，从而达到降价的目的。

他还提到，很多决策还是靠判断力，因为缺少数据，没办法进行相关的实验或者并没有帮助。其实，现在数据驱动已经是共识了，特别是精益创业理论出来之后。

他认为做数据分析来做决策，只能用于短期，不能用于长期。这个其实也看情况，大部分的情况都是做辅助决策，模型只是服务于长期的目标而已。

在创业阶段，采用数据驱动是非常重要的。至少你需要让用户使用才能产生数据，而反馈的数据不管大小都是有用的。重要的是，怎么去利用这个数据来对业务进行调整。