大多数组织如何制定错误的数据策略 —以及如何修复它

揭秘数据策略

从数据、分析和人工智能中释放商业价值是一个热门话题。然而，大多数组织都在努力将现有数据货币化或创建新数据来推进他们的思维。我断言，在大多数情况下，这种困境源于缺乏有效的商业战略设计。

这是第三期系列揭秘数据策略– 组织似乎需要变得数据驱动并在当今的数字世界中保持竞争力。本文档作为数据策略的手册，通过消除误解、澄清定义、提出可操作的框架以及通过详细示例说明概念来提供深入且结构化的探索。

我是延斯是一位具有商业头脑的数据专家，在实施数据和人工智能用例方面拥有近二十年的经验。我帮助各行业的领导者设计战略并培育文化，以释放数据和算法的全部潜力。

• 数据战略不是业务战略的独立补充，也不是数据能力建设的计划。它更像是数据功能的策略，为内部数据客户提供服务。数据策略
• 不定义组织如何利用数据、分析和人工智能创造价值。这是业务战略的工作，与数据相关的战略选择是不可或缺的一部分。
• 对数据战略的误解加上薄弱的业务战略设计，阻碍了组织有效利用数据。这些问题阻碍了他们获得竞争优势的能力。
• Playing to Win 框架可以帮助组织实现以下两点：将数据创新嵌入到其业务战略中并设计有效的数据战略。

Data、分析和人工智能 (AI) 并不是商业中的新现象，但在过去几十年中变得越来越重要。

由于生成式人工智能最近在生成文本或图像等输出方面取得了令人印象深刻的进步，目前围绕生成式人工智能存在着很多令人兴奋的事情。当我提到人工智能时，我指的不仅仅是生成式人工智能，而是更广泛的从数据中生成价值的工具。这些工具包括商业智能、机器学习或数据科学，而且绝非新鲜事物。它们已经存在很多年了。

持久的兴趣从何而来？

1.1 数据可以为整个组织创造效益

虽然数据曾经主要与会计和控制团队相关，但现在它往往对整个组织的成功至关重要：

• 营销使用历史活动数据来识别推动成功的因素，从而更有效地分配营销支出。
• 采购部门使用数据和分析来分析供应商绩效，从而最大限度地减少中断并最大限度地节省成本。
• 销售人员使用历史数据来推荐最佳的销售点产品组合，从而增加收入。
• 客户服务中心利用数据来预测客户流失，从而采取主动行动来降低客户流失率并推动可持续的收入增长。
• 生产部门利用物联网数据来防止机器故障，降低维修成本并最大限度地减少生产停机时间。
• 产品开发团队通过数字应用丰富现有产品和服务，例如为客户提供增值见解、提高客户忠诚度并确保竞争优势。

许多组织已经成功地利用基于此类分析数据用例的数据价值。但数据和人工智能可以做更多的事情。

数据和人工智能提供了质疑公司和整个行业现状的手段。它们使组织能够通过提出有关客户、技术、竞争对手和行业的正确问题来获得知识优势，从而提供一种获得和维持竞争优势的方法。它是将数据洞察力与人类创造力相结合，积极设计未来[1a,2]。

鉴于巨大的利益潜力，越来越多的组织渴望利用数据作为工具也就不足为奇了。战略资产。公司努力成为数据驱动型——或者更确切地说数据启发[2]，这意味着他们的目标是：

• 持续利用现有数据的潜力来优化业务
• 使用数据和分析作为业务创新不可或缺的一部分，快速测试新的创意，从而推进思维并提高竞争力

那么问题在哪里呢？

1.2 组织努力成为数据驱动型组织

尽管前景广阔，但许多组织仍在努力利用数据、分析和人工智能的优势。在技​​术、算法和方法已经存在了几十年之后，这怎么可能呢？

变得数据驱动可能是一项复杂的工作

组织常常低估数据驱动的复杂性，尤其是当缺乏战略清晰度和数据素养阻碍进展时。

当组织开始努力利用数据作为资产时，很多事情都可能会出错。这是不是关于购买一些即插即用技术解决方案，正如许多供应商倾向于承诺的那样。

数据驱动通常被称为数据驱动的业务转型 [3]。它被称为转型是有充分理由的，因为它通常涉及影响整个企业的变化。

例如，需要从各种来源获取和集成数据，以便将其提供给业务。数据集成需要公司范围内的编排和治理。如果没有它，每个数据消费者都会重复繁琐且容易出错的数据准备工作，最终导致“数据狂野西部”，报告的真相不同，宝贵的资源被浪费。

利用数据和人工智能是企业的责任

数据的实际价值产生主要发生在业务部门和职能部门，即营销、采购、销售或制造。

许多人认为成为数据驱动纯粹是 IT 和分析方面的技术专业知识的问题，而忽视了它所需的文化和组织变革。但这是一个谬论。

业务用户自己需要具备数据素养。他们需要拥抱一个文化创新、识别和开发与其领域决策相关的数据用例。他们需要能够利用数据发现新见解，从而引发新问题并激发创新想法，从而创造竞争优势。

组织需要文化和技术能力来实现数据驱动，但很难识别、构建和维护这些能力。

最后，利用数据是团队的努力，需要跨学科的利益相关者和思考。

数据、分析和人工智能并不是解决一切问题的良药

数据、分析和人工智能目前被过度炒作，人们似乎忘记了每种算法的基础都是数学。然而，每种统计算法都有潜在的假设和固有的局限性。仅仅依赖数据进行决策与相反的极端（即根本不使用数据进行决策）一样糟糕。

著名的名言——预测是困难的，尤其是在面对未来时â，源自丹麦政治家卡尔·克里斯蒂安[4]，提示您需要谨慎应用任何预测：根据历史数据进行预测的一个基本假设是，生成数据的底层系统在未来不会发生根本性的变化。这种假设经常会被违反，有信誉的数据专业人士必须在以水晶球方式应用奇特算法之前仔细检查它。

数据、分析和人工智能被过度炒作这一事实可能会产生问题，因为利益相关者可能有不切实际的期望。需要通过在组织中建立数据素养和文化来解决这一问题，从而培养共同的语言和理解，从而实现现实的期望。

举个例子，需要谨慎对待分析的一个特定领域是战略设计过程，因为创新战略的目标往往是独特地塑造未来并对市场产生影响。如果这样的策略成功，那么基于过去样本数据的销售预测可能不具有代表性，并且没有价值从中得出有意义的结论。1a,1b]。

商业策略缺失是数据和人工智能频频失败的根本原因

正如我在第二篇文章[5]的揭秘数据策略文章系列我认为，组织在管理和利用数据作为资产方面遇到的大多数问题都源于缺乏战略基础。对于战略基础工作，我指的并不是数据战略的常用（错误）概念（正如我们在下面考虑的那样），而是真正的战略设计工作，它将与数据相关的战略选择自然地整合到业务战略中。

因此，本文利用著名的 Playing to Win 策略框架的工具，将这一想法扩展为一个随时可用的框架。

但首先，让我们仔细看看数据战略的前景以及围绕它的误解。

1.3 数据战略承诺

随着过去几十年商业智能和数据科学应用的增长，组织越来越难以管理两者的复杂性：数据管理以及分析解决方案的创建和扩展。需要更结构化的方法数据策略过去和现在仍然是解决这些问题的假定方法。

为了写这篇文章，我试图找到数据策略这个术语的起源。使用 Google 图书搜索我找到了这本书�分布式数据处理的设计和策略�[6]从1981年开始。

因此，数据策略这个术语并不新鲜，但自 2010 年以来，它引起了越来越多的兴趣，如谷歌趋势 [6]搜索，这与我在这段时间作为数据专业人员的个人经历相符。

战略通常被认为是一种聪明的东西，可以管理复杂性，而数据战略的前景在于解决组织在利用数据、分析和人工智能方面面临的挑战。

然而，有一个问题：尽管它日益突出，但数据策略— 定义仍然不明确，并且最常被误解，从而导致沟通不畅、举措不一致和期望得不到满足。

要理解为什么组织继续在数据战略方面苦苦挣扎，我们必须首先解决围绕该术语的误解。

作为作为一名数据专业人士和管理顾问，我了解到清晰的沟通对于成功至关重要。清晰的沟通始于共同语言，这就是数据策略问题开始的地方：对于它的实际含义或应该是什么还没有达成共识。数据专业人士以及业务专家在不同的环境中使用它，同时含义也不同。

目前对于“数据战略”的真正含义还没有达成普遍共识。

此外，最常用的数据策略定义与既定的策略定义不兼容。在我看来，这造成了一个严重的问题，因为它阻碍了数据专业人员与高管和业务利益相关者进行以目标为导向的讨论，因此成为组织应对数据挑战的障碍，阻止组织变得数据驱动。

缺乏对数据战略的共同理解阻碍了业务和数据专业人员之间以目标为导向的讨论，因此成为组织利用数据作为资产的障碍。

这种缺乏共识导致了对数据策略的广泛误解。

尽管许多有关数据策略的出版物有意无意地避免明确定义该术语（假设读者已经知道它的含义），但在提供定义时，它们常常会暴露出误解。让我们探讨一下我认为最常见的三个。

2.1 误解#1：构建数据和人工智能能力的计划

计划侧重于执行和时间表，而战略则侧重于在不确定性下做出深思熟虑的选择。将两者混为一谈会导致混乱和低效率。

Playing to Win 战略框架非常清楚计划和战略之间的区别[8]正如我们将在下面的 3.3 节中详细看到的。然而，当我用谷歌搜索这个词时 –数据策略— 第一页的几个结果 — 跳过广告 — 将其定义为计划或路线图。一些例子是：

1. —数据战略是一项长期计划，定义了管理组织信息资产所需的技术、流程、人员和规则。� [9]
2. —数据策略是一个明确的计划，旨在提高全公司数据管理的有效性。� [10]

从本质上讲，属于这一类别的数据战略定义是关于构建数据、分析和/或人工智能能力的计划。这些组织能力涵盖数据架构、数据治理、数据管理、BI、数据科学和数据文化等要素。

将数据战略定义为构建数据能力的计划反映了对战略真正含义的普遍误解。战略应首先定义您的组织真正需要哪些与数据相关的功能。

构建数据、分析和人工智能能力的计划不是你的战略，而是你的战略工作的输出。

创建这样的计划本身就是一项具有挑战性的工作，并且绝对需要它们的设计来利用数据作为公司资产。因此，它需要投入大量的时间，并且需要在广泛的跨学科领域拥有扎实的能力。我想说的是，不要将其称为数据策略，因为它就是一个计划。

2.2 误解#2：数据管理指南

另一个常见的定义将数据策略视为一种最佳实践指南，阐明如何获取、存储、管理和利用数据。

管理数据的方法是支持战略但又不同于战略的重要运营工具。将它们称为“战略”可能会掩盖深思熟虑的战略选择的重要性。

如何获取、存储和管理数据是重要的主题，也是在任何组织中创建可扩展的数据管理基础的关键任务。然而，方法和指导方针不应被称为战略。它们只是不同的东西。

2.3 误解#3：商业战略的补充

我必须承认，在我职业生涯的早期，我也陷入了这样的误解：数据战略可以作为业务战略的单独补充而存在。然而，数据策略永远不应该孤立，定义数据如何创造业务价值。这本质上是业务战略本身的作用，我们将在第 5 节中进一步探讨。

我喜欢使用乐高积木来形象化这一点：业务策略是一组相互关联的乐高积木的集成集，而错误的数据策略则表现为一个单独的实体，被臭名昭著的强行粘在上面。商业主— 与他可怕的 —克拉格Ø 胶水[11]，扼杀创造力并阻碍创新设计。

数据策略作为业务策略的补充是什么意思？每当数据策略定义为 时定义组织如何利用数据来实现其战略目标的东西— 这是一个误解 [1c]。A 明确定义的策略是单一的[1d]为每个不可分割的业务实体制定一项业务战略。企业战略的一系列选择形成了一个和谐的整体。试图强行推行某些东西，例如通过利用数据和人工智能来获胜的单独方式，根本行不通。所以：

拥有一个补充数据策略，通过说明您的组织如何使用数据、分析和人工智能来创造业务价值并获得竞争优势来修改您的业务战略，这并不是您的组织应该追求的目标。

相反，正如我在揭秘数据策略系列文章的第 2 部分 [5]，您的业务战略应明确您的组织是否以及如何使用数据、分析和人工智能来获得战略竞争优势。明确定义的业务战略已经包括数据如何创造价值[1c]。添加补充数据策略会削弱焦点并破坏集成。

第 5 节讨论了如何设计战略性地利用数据的业务战略的方法。

2.4 何去何从？

规避数据策略的三个误解中的前两个相对简单，因为它只是就一种共同语言达成一致。计划是计划，指导方针是指导方针，战略是战略。这些是不同的概念，数据和业务专业人员都必须在沟通中认识到这一点。

但第三个误解，即不应该有数据补充策略的事实又如何呢？

缺少的是数据策略的连贯定义，数据专业人员和业务专业人员都可以理解这一定义。

为了消除这种误解，我们需要做两件事：

1. 数据策略的连贯定义与通用业务策略定义相一致
2. 将数据相关选择集成到业务战略结构中的过程

本文的其余部分将解决这个问题。它提供：

• 数据策略的定义，与业务上下文中策略的定义兼容
• 明确利用数据作为资产的业务战略的作用
• 完成数据驱动所需的所有战略工作的流程

建议框架的一个重要副作用是它对数据领导力的积极影响，有助于数据文化和高管数据素养的发展。

本文其余部分的结构如下：

• 第三节提供了著名的“为赢而战”战略框架的摘要，该框架将作为数据驱动所需的所有战略基础工作的基础。
• 第 4 节提出了数据策略的定义，该定义与 Playing to Win 策略框架相一致并适合。它还提供了详细的端到端流程来执行组织实现数据驱动所需的所有战略工作。
• 第5节澄清，仅设计第 4 节中定义的数据策略对于旨在成为数据驱动的组织来说是不够的。您更需要将与数据相关的战略选择编织到业务战略的结构中，从而将战略设计与数据用例创新和开发结合起来。
• 第 6 节提供了为一家生产和销售薯片和坚果的咸味休闲食品公司设计数据策略的详细示例。本节面向希望深入探索数据策略设计过程的读者。读者可以跳过这一部分，而不会失去文章的整体流程。
• 第 7 条提供一些结论性意见。

3.1 为赢而战的战略框架磷

laying to Win 是由罗杰·马丁 [8]全球公认的商业思想家、首席执行官顾问、前摩立特顾问、罗特曼管理学院名誉教授。该框架是在 20 世纪 80 年代和 90 年代开发并不断完善的 [

1e]，并最终达到合着书[13]由罗杰·马丁 (Roger Martin) 和宝洁公司 (P&G) 前首席执行官 A. G. 雷富礼 (A. G. Lafley) 撰写。该书于2013年出版。

该框架成为宝洁公司的标准战略方法，并已成功应用于众多行业，成为广泛认可的战略方法。此外，罗杰·马丁（Roger Martin）通过广泛的研究不断详细阐述了“为赢而玩”框架。策略从业者见解系列 [1,1f]。

虽然存在其他策略框架，但我选择“为赢而战”，因为它被广泛认为是有效策略设计的标准。此外，它还配备了完整的资源生态系统，例如文献、流程、模板和培训，可用于设计任何类型的策略。

虽然 Playing to Win 为设计业务策略提供了一个强大的框架，但它还提供了一种如何将数据相关选择集成到业务策略中的方法。这将在第 5 节中详细探讨。

那么，战略一词的明确定义是什么呢？

3.2 战略实际上是什么

定义战略一词

这为赢而战战略设计框架将战略定义为一组综合的选择，这些选择形成了合理的理论使组织能够在竞争中获胜。

– 战略是一套综合的选择，它使企业在其行业中处于独特的地位，从而创造相对于竞争对手的可持续优势和卓越价值。 – [13]

该框架强调 –获胜— 强调战略的最终目标是竞争成功，而不仅仅是生存或参与市场。如果你只是为了玩而玩，那么你面临的风险是，至少有一个竞争对手知道如何获胜，并最终将你的公司赶出市场。

策略选择级联

战略设计团队需要做出的选择可以分为五组，它们共同构建了所谓的策略选择级联，这是 Playing to Win 策略框架的核心工具之一。

我喜欢用乐高积木来形象化这个级联：级联中的每个盒子代表一组战略选择——每个选择都由一个单独的乐高积木来象征。每个方框内的选择都是相互关联的，并且方框本身相互加强，从而创建一个集成且有凝聚力的策略。

该级联表明组织需要做出以下选择：

1. 获胜对组织意味着什么（胜利的愿望）
2. 服务哪些客户、提供哪些产品（去哪里玩）
3. 该组织如何旨在赢得市场上的这些客户（如何获胜）
4. 实现这一目标需要哪些关键活动和资源（能力）， 和
5. 什么是系统、流程、规范、文化和指标 需要确保这些能力的建立和维护（管理系统）

这些选择并不意味着是一个松散耦合的列表，而是需要很好地集成以形成一个引人注目的整体。这是你的策略。

显式策略与隐式策略

如果你仔细想想，每个组织要么以明确的战略形式明确、有意识地做出这五种选择，要么通过其行动隐含地做出这五种选择。例如。组织可以选择销售哪些产品、不销售哪些产品，或者服务哪些客户、不服务哪些客户。

因此，每个组织都有一个战略——无论是否明确阐述。这种隐性策略可以从公司采取的行动中推断出来[1克]。

这种隐含的“使用中的战略”的一个问题是，组织中可能无法就战略是什么达成共识。这有两个含义：

• 该组织很难改进，因为不清楚目前的状况。
• 基于此的行动可能无效，因为利益相关者会适得其反。

应用级联的用例e级联是一种在策略设计过程中不会只填写一次的工具，而是在整个策略设计过程中针对各种用例反复重新审视和使用。

例如，它用于：

• 重建竞争对手的战略通过观察他们做了什么：这是一个熟悉框架的好习惯，也是一种破冰船 [1 小时]开始策略设计过程。
• 记录您当前的策略：这是建立对现状的共同理解并使当前隐性选择明确的好方法（如果以前没有这样做过）。
• 记录新的战略可能性： 这设计策略的过程 [1i]包含一个创造性的元素，未来目标策略的许多可能性被生成并系统地评估。这些都使用策略选择级联进行记录和详细说明，每种策略对应一种可能性。

3.3 计划与策略：典型的误解

由于战略一词与智力和目的性相关，因此它在很多情况下用于许多事物和活动。它用于描述目标、使命和愿景声明、商业模式、口号或主题或任何类型的举措。如果你想让一个术语听起来更奇特，只需使用策略作为后缀即可。

典型清单战略神话提供于[1j]，包括最主要的误解：计划不是策略 [8]。罗杰·马丁（Roger Martin），Playing to Win框架的鼻祖，花费了大量的时间和精力来解释计划和策略之间的区别[8,1j,1k,1升,1米]。这些术语有何不同？

计划是关于确定性的。他们制定项目的时间表、可交付成果、预算和责任。战略是关于不确定性的。它包括对组织如何获胜的赌注的选择。

因此，当大多数人说战略时，他们实际上指的是一个计划或一系列举措。然而，这些是完全独立的事情，尽管它们是在一起的：规划遵循策略。一旦您的策略被设计并商定，您就知道激活您的策略所需的功能和系统。为了构建这些功能和系统，您需要制定计划。

这种计划-策略-混乱也适用于数据策略正如我们在 2.1 节中看到的。

知道什么是战略，它是如何制定的？

3.4 战略选择构建过程

要设计策略，需要的不仅仅是填写五个方框。Playing to Win 框架的第二个工具是一个既定的策略设计流程，称为战略选择构建过程。

这个过程实际上是一个通用工具，用于解决实际结果和期望结果之间的差距[1i]。它包括七个步骤。

第 1 步：问题定义

战略首先是解决问题的工具 [1n]，因此您首先要确定战略设计的业务需求。由于您当前的策略（隐式或显式）正在产生问题，因此您需要一组新的选择，以产生您想要的结果。

第二步：我们可以怎样做？

此步骤旨在生成一份声明，开启步骤 3 中构思阶段的讨论。

对于前两个步骤，重要的是要关注客户而不是内部元素。

第三步：可能性生成

这是设计过程的创意部分之一。这是关于创新不同的可能性，以及如何解决所识别的问题。对于每种可能性，制定了选择级联。

步骤4：必须正确的是什么？

在此步骤中，战略设计团队旨在揭示必须保持的这些条件，以使所考虑的可能性将成为一个很好的策略。需要为创建的每种可能性进行此练习。

团队问必须是真实的对于客户，公司和竞争对手，可以揭示每种战略可能性的基本假设。重要的是不要问什么是真的，由于这导致了无效的讨论，并且最佳策略创造了一个新的未来，也就是说，由于新战略的实施，今天不正确的事物可能会在将来成为现实。

步骤5：选择障碍

如果策略设计团队缺乏信心是正确的，或者将来可以实现，则步骤4中确定的任何条件都可能成为障碍。确定的障碍可以阻止设计团队在步骤7中选择相应的可能性作为最终策略。

如果已经确定了障碍，则需要进行测试。

步骤6：测试与转换

对于每个障碍，需要设计和进行测试，以确定情况当前是正确的，或者是否存在合理的转换路径以使其真实[1i]。这是过程中的第二个创意元素。

这些测试不是为了实现100％的信心，这是不可能的，而是要缩短战略可能性的几率，以使设计团队充满信心，可以将其作为最终策略。

步骤7：选择

在制定了所有战略可能性之后，已经确定和测试了潜在的障碍，战略设计团队应具有足够的信息来选择一种可能性，而不是其他可能性，这成为了新的战略。

请注意，战略选择结构过程包含设计思维[1O]分别在第3和第6步的不同思维的阶段[14]。

3.5组织需要哪些策略

一个在一个地理上出售一种产品的公司可能只需要一种策略[1天]。但是，如果业务要素在在哪里玩和如何获胜，每个元素都需要自己的策略。

对于每种策略，都需要应用上述设计过程，以获得一套独特的选择，从而导致单独的策略选择级联。因此，大型组织将有几个策略构建块，而挑战是使​​它们彼此合并并加强[1p,1q,1r]。

战略涉及在不确定性，约束和竞争中做出选择，这发生在公司的许多地方。1c]

业务策略

战略最核心的形式是经营策略，这是服务或产品的策略[1c]在一个地理区域[1秒]。在这种不可分割的战略水平上，组织与给定客户群的给定竞争对手竞争。

聚合策略

另一种策略形式是聚合策略[1q]，将一组策略结合在一起。这企业战略只是组织内部最高的聚合策略。汇总策略的任务是为下面的业务组合增加竞争价值，以便整体大于其部分的总和。

典型的嵌套聚合水平是公司，业务部门，类别和品牌。聚合策略的另一个例子是区域策略，例如给定地区或国家。

功能策略

第三种策略是功能策略。

功能是组织内部的专业领域，例如人力资源，IT，财务，市场营销和研发，可为更广泛的组织或业务部门提供支持服务。他们可以作为集中团队（例如公司职能）或特定业务领域的分散单位运行。功能的目的是提供关键的运营支持和专业知识，以帮助业务部门有效地执行其策略。

功能策略对于确保组织内的专业领域或团队为其内部客户带来价值，同时与组织的总体目标保持一致。每个功能都需要自己的策略[15]确定：

• 提供什么服务给内部客户
• 优先级的资源有限
• 如何在支持组织更广泛的目标的同时实现卓越运营

除了满足纯粹的运营需求外，某些功能还可以通过提供跨越多个业务或聚合策略的共享功能或系统来发挥战略性，综合的作用。这些共享功能和系统称为加固杆[1q]，专注于在整个组织中创建协同作用和结盟嵌套策略。我们将仔细研究第3.6节中的加固杆。

策略架构

在各个企业具有不同竞争对手和客户的大型组织中，存在一个嵌套和相互联系的生态系统。

例子

为了说明这个概念，我选择描绘一家咸零食食品公司的策略建设块，该公司生产和出售薯片和坚果。

有两个聚合策略层：一个针对德国和英国等地区，另一个用于总体企业集团级别。

在德国地区的汇总战略之下，有四种业务策略，每种策略形成了不可分割的单位，即竞争发生的基本水平。坚果品牌美食家和健康有不同的客户和竞争对手，因此需要单独的理论来获胜，就像私人标签坚果和薯片一样。

德国地区的聚合策略提供加固杆利用生产和人力资源功能提供的协同作用。每个功能都需要自己的策略。

数据策略在何处以及如何适合策略体系结构？在第4节中，我将数据策略定义为数据功能的策略，将其直接置于功能策略中。但是，仅此组织就不足以使组织成为数据驱动。正如我们将在第5节中探讨的那样，实现这一野心不仅需要精心设计的功能数据策略 - 它需要将与数据相关的选择嵌入到个人业务策略，其他功能策略和总体汇总策略中。

3.6加固杆

与增强混凝土结构的钢棒类似，加强杆是战略中的一个概念，它可以实现整体战略建筑。如第3.5节所述，汇总策略的主要任务是为其跨越企业的投资组合增加竞争价值。这通常是通过加强杆来实现的。

加强杆是跨越多个业务部门或组织级别的功能或系统，从而创造了增强价值和竞争优势的协同作用。他们通过将不同的层次（例如，公司，业务部门，地理）联系起来来加强整个组织，并使他们能够比单独统治更好。这样，加强杆为不可分割的策略提供了净收益，从而使其更有效。

增强杆的示例可以是在各个地区利用的研发或品牌建设的共享功能。

因此，加固杆是必不可少的功能或启用聚合策略的管理系统。汇总策略确定了为什么需要这些杆，并根据其对组织的战略价值对它们进行优先排序。但是，他们的意识通常会委派给函数。

总而言之，加强杆是关于在组织层面上建立具有战略影响的协同作用，减少重复和优化专业或稀缺资源的使用。它们使组织能够通过确保其嵌套策略无缝地协同工作来更有效地竞争。

3.7组织不需要哪些策略

每当您认为需要一种补充策略来修改或覆盖业务，汇总或功能策略时，这可能是一种误解。从定义上讲，策略是一组集成的选择，它们相互加强。通过与另一个策略叠加策略，您不会实现集成，而是相反，因为您会放松整合。

在 [1c]，罗杰·马丁（Roger Martin）提供了这种不希望的策略退化的例子。这些是内容策略，合作伙伴策略，增长战略，可持续性战略，数字战略以及数据策略。

正如我们在第2.3节中已经说过的那样，作为对业务策略的补充实体的数据策略是应避免的误解。

第4和第5节将讨论这对我们打算帮助组织成为数据驱动数据以利用数据作为资产来利用数据的含义。

3.8策略设计团队

一个较大的组织可以拥有一系列不同的策略，那么谁负责哪种策略设计？

通常，每个需要战略的实体的领导者最终负责做出相应的战略选择[1吨]。这意味着首席执行官负责公司战略，而IT领导者则负责功能性IT战略等。

策略很少是由一个人设计的，但这是一个团队的练习，从一开始就创建参与以激活策略很重要[16]。但是，这并不意味着涉及尽可能多的人。战略决策者必须具备在竞争，客户和公司的背景下评估战略选择的能力。这些能力应驻留在设计策略的实体领导团队中[1t]。

在战略设计团队中，通常存在以下角色：

• 赞助：负责总体选择，建议并最终批准战略
• 所有者：设计该战略的部门负责人对做出战略选择负有最终责任
• 核心团队成员：负责共同创造战略选择的人。对于公司战略，这是整个领导团队。
• 扩展团队成员：有观点的人，这对于制定策略很重要
• 主题专家：人们，拥有特定的专业知识

3.9总结策略基本面

如我们所见，赢得框架的游戏为设计策略提供了强大而结构化的方法。它提供了丰富的定义，工具和思维生态系统，有助于生成任何成功的策略。

策略不是一个计划，而是一组集成的选择，可以识别一个可以在那里赢得胜利的理论，这是由能力和相应的管理系统支持的，这些系统可以建立和维护这些功能[1天]。

战略选择级联有助于构建所需的选择。战略选择的结构过程有助于设计您的策略，从一个问题开始，而不是生成和评估不同可能性的问题，直到可以选择最终的可能性成为解决该问题的新策略。

既然我们对哪种策略及其设计的设计有了深刻的了解，那么重要的是要考虑与数据相关的战略选择如何适合此框架。第2.3节中突出显示的数据策略的误解只能通过将与数据相关的选择嵌入业务策略本身来解决。这样可以确保数据，分析和AI被视为战略野心的推动者，而不是独立或补充计划。

在下一节中，我们将探讨如何以与赢得框架无缝一致的方式定义数据策略。然后，在第5节中，我们将讨论如何直接从业务策略中得出数据的战略要求，从而创建一种凝聚力的方法，将数据用例创新整合到组织整体策略的结构中。

4.1定义数据策略

中号OST组织需要专门的功能，以提供与数据相关的服务或产品，主要是向内部和最终的外部数据客户提供。此功能需要如我们在第3.5节中看到的策略。这是您的数据策略。

数据策略是数据，分析和AI功能的策略。它概述了该功能如何通过提供与数据相关的产品和服务来为组织更广泛的业务战略服务。

您的数据策略阐明了获胜对数据功能的含义，您的内部数据客户是谁，数据功能提供的服务或产品，其如何与数据客户获胜以及赢得哪些功能和系统。

请注意，数据策略并未定义数据，分析和AI如何为组织创造竞争优势。这是您业务策略的任务！在第5节中，我表明，与数据相关的战略选择是业务战略的自然组成部分。反过来，业务策略提供了战略数据要求，数据功能的要求以及您设计数据策略的要求。

正如为业务服务的人力资源或IT策略一样，数据策略是交付数据要求的功能策略 - 不是对数据如何创造价值的独立定义。

未能定义清晰的数据策略会导致缺乏重点，而数据功能要么试图满足所有需求（奴役策略）或追求断开的数据计划（帝国战略）[15]。两种方法都破坏了功能获胜的能力。

根据此功能中提供的服务和产品的性质和重点，您可以称其为数据策略，分析策略，BI策略或AI策略。

这种功能的策略有助于数据，分析和AI团队的重点，在我看来，这是唯一应称为的实体数据策略。

4.2数据策略选择级联

当使用游戏来赢得数据策略的设计框架时，需要调整一些术语。这是第一部分[17]这个的神秘化数据策略文章系列。我们需要的调整是：

• 提供数据提供：这包括与数据，分析或AI相关的产品或服务。
• 客户数据客户：与数据提供的人一起服务。这些最有可能是公司内部利益相关者或团体，但有时也可能包括外部利益相关者。
• 公司数据服务提供商：创建数据提供的人。数据功能的工作人员。
• 竞赛：替代方案，除了数据服务提供商以外，数据客户可能会选择。这可以从根本无法提供的数据客户，服务或使用外部服务的范围内。
• 地理区域：业务部门，部门，团队，域或地理区域您的数据策略将重点关注。
• 渠道数据输送渠道：数据客户如何访问数据产品。
• 生产阶段 - 数据生命周期管理：数据生命周期的哪个阶段是内部或外包的。

翻译核心元素后，数据策略选择级联[17 号]可以定义如下：

1. 赢得愿望：您功能的获胜的定义。
2. 在哪里玩：您将在其上选择竞争的竞争环境。这通常包括五个维度：i）重点领域，ii）数据客户，iii）数据传输 渠道，iv）数据产品，v）数据生命周期管理。
3. 如何获胜：您的数据策略将创建战略优势。您将如何可持续地与数据客户获胜。
4. 必须具有的功能：它需要哪些关键活动和资源来实现如何获胜。
5. 启用管理系统：基础架构（系统，流程，治理，指标和文化），为了支持和维持您的数据策略而需要。

4.3数据策略设计过程

至于任何策略设计，我们都可以运用游戏的战略选择结构过程来赢得框架。我提出了五个其他步骤，以制定数据策略，以纳入业务策略要求，并将其与后来的计划明确区分，以构建功能和系统。

以下小节描述了数据策略设计的五个其他步骤。

4.4步骤I：确定数据和AI需求

从上面的定义来看，应该明确的是，数据策略不应回答数据，分析和人工智能如何创造业务价值和竞争优势。正如我在[5]，这应该是您业务，汇总和功能策略的一部分。这些在第3.5节中进行了讨论。对您的数据功能的战略数据需求来自这些策略，并符合运营和最终的未来数据需求。

从本质上讲，第一步是要清楚组织中现有的数据，分析和AI的要求。总共有三个来源来解决此类数据需求，为要构建的随后数据功能提供了自然的优先级。

PRIO 1：战略数据需求s

在制定战略体系结构时，随着个人策略设计团队的进步，组织战略层次的战略数据需求应变得清晰。固有的困难是策略设计不是线性过程，而是一种来回的迭代过程。这适用于单个策略的设计[1u]以及策略体系结构中不同级别的编排[1q]。

通常，对数据的战略需求发生在您的组织的不同级别，可能会产生对数据，分析和AI的需求，这是战略架构的增强杆（参见第3.6节）。正如我们在第6节中所看到的那样，这甚至可能是设计中央数据功能的动机。

另一种方式，战略数据要求可以自​​己表现出来的是数据用例的形式。我们将遇到此类用例的示例，该例子直接有助于组织的竞争优势。然后，这种用例可以阻止对新型AI功能的需求。大多数情况下，这要求业务由数据和AI的相应功能支持。

战略数据的另一个重要方面要求组织真正成为组织的愿望数据启发[2]。这是为了通过创建有关客户，技术，竞争对手和行业的见解来不断推动其思维，以获得竞争优势。这里可能会有一些递归，因为这种功能也将在业务策略设计过程的核心本身中使用，正如我们在第5节中所看到的。但是，如果组织完全承认数据的巨大潜力，那么这不仅限于在最佳实践数据用例的积压中提高运行性能表现，这当然是设计数据功能的必要条件。这将导致在数据策略中进行选择，以通过其业务策略设计来支持领导力，因为我们将在第5.2节和5.3节中进行详细介绍。

理想情况下，数据要求应由相应的战略所有者积极说明，并通过选择租赁到达公司数据负责人。

数据专业人员收集战略数据需求的最佳方法是数据策略的输入是成为实际战略设计团队的一部分，并积极帮助创新和评估战略数据用例，如我们将在第5.2和5.3节中讨论的那样。如果不可能，您应该分析组织的当前战略体系结构。如果正确设计，各个战略构件应直接提供战略数据要求。

不幸的是，在许多公司中，战略设计似乎是一项失落的艺术[1v]，这需要一些侦探工作来找出组织的当前战略架构。

所有战略层中确定的战略用例集合为数据，分析和AI提供了最重要的要求，因为相应的用例对于支持组织的策略和竞争优势至关重要。

PRIO 2：操作数据需求s

数据需求并不仅仅是由于战略选择作为战略的一部分而产生的，但也可能来自操作选择。

作为您战略的一部分，通常只有一些能力和系统的战略选择。提供详尽的组织能力和系统列表，而不是为赢得胜利理论而提供详尽的列表，这根本不是战略的任务。

因此，您的战略选择通常与其他选择相辅相成手术要求[1瓦]。这些是那些没有使您的组织与竞争对手区分开的操作选择，而是标准或最佳实践选择，也是其他人都在做的积极的事情。

具有数据相关性的操作选择的一个示例可能是报告和BI公司战略的功能。最好的做法是在组织中拥有BI解决方案，但是这些解决方案可能不会有助于您赢得的理论 - 它们只是手术的命令。

与战略数据需求一样，操作数据需求可以表现出来的另一种方式是数据用例的形式。这些操作数据用例提供点优化，例如现有过程的效率提高。

我通常会通过一系列采访发现操作数据需求，这些访谈与客户的业务和数据利益相关者进行。

PRIO 3：未来数据需求s

为了构建未来的数据策略，组织应考虑数据需求的第三类。

通常，组织的不同部分已经有了想法，甚至计划如何利用数据和AI将来为其领域。因此，重要的是要确定这些未来的数据需求，因为这些可能对数据或AI函数的策略有影响。

可以通过与业务利益相关者的访谈或更积极的方式通过进行数据用例创新研讨会来以被动的方式收集此类数据需求。

一旦所有数据需求都清楚，您就需要在潜入数据策略设计过程之前对当前的疼痛点有深入的了解，因为这些可能会影响问题的表述。

4.5步骤II：收集挑战

在阐明数据功能的要求之后，您需要确定利用组织中数据的潜在挑战。这些挑战将构成在战略选择结构过程的步骤1中选择问题定义的基础。

任务是确定组织中数据值生成的最紧迫的缺陷。通常，我将其作为对业务利益相关者的访谈的一部分，当在步骤I中确定数据需求时。出于文档目的，我喜欢借用著名的右半价值命题画布[18]。

此步骤的结果是对大多数紧迫数据相关的挑战的深刻理解，这些挑战与数据的要求一起提供了一种适合差距分析，作为战略选择结构过程的步骤1的基础。

由于数据策略设计过程的第三步是我们已经在第3.4节中讨论的标准战略选择结构过程，因此我将在此处跳过重复，直接跳到步骤IV。

4.6步骤IV：操作功能和系统

在第三步完成后，数据策略很容易设计，需要激活[16]。这是通过设计和构建这些功能赢得所需的功能和系统来完成的。为了构建有效的数据功能，必须清楚地了解所需的战略和运营能力和系统。

战略能力和系统通常是数据功能所需的一小部分功能和系统。这些是在您的数据策略中定义的。运营能力是那些基本活动，可以有效且可靠地执行战略选择，但并未直接与您的数据客户获胜。

战略和运营能力和系统之间的区别提供了自然的优先级，在构建这些要素时，在其中投资大部分时间和资源。

如我们所见，操作数据的需求已在步骤I中确定。操作能力和系统通常是直接从中得出的。例如，这可能包括有效的报告或BI等功能。请注意，战略设计过程的第三步，这是战略选择结构过程的测试和转换步骤，可能需要提供其他操作功能和系统。

因此，步骤IV是通过概述数据功能成功的哪些操作功能和系统来完成图片。为此，使用能力图可能会有所帮助。数据，分析和AI的功能图是数据专业人员的众所周知的工具（参见上面的图2），通常用作一种清单[3]构建数据功能时。

在数据策略的背景下，用于操作启用管理系统的典型要素是：

• 数据平台
• 数据目录
• 商业智能工具
• 决策框架
• 开发数据文化的系统
• 数据成熟度模型
• 数据治理框架
• 技能矩阵
• …

一旦您对所有战略和运营功能以及数据功能都需要成功的概述，那么该是时候评估现状并识别差距了。

4.7步骤V：成熟度评估

在定义了战略和运营数据功能和系统之后，必须通过进行数据成熟度评估来评估您的当前状态。

许多咨询和数据专业人员倾向于从数据策略设计过程的早期开始这种活动开始。但是，我相信这种方法被误导了。尽管概述该组织当前拥有的功能肯定是有价值的，但在战略设计过程开始时进行了详细评估，几乎没有价值。您首先需要确定所需的特定功能和系统，以及每种特定的成熟度。这意味着必须定义一组功能和系统及其各自的目标状态。但是，只有在设计数据策略之后才有可能。

此外，我对重点是针对竞争对手或行业平均的板凳标记的数据成熟度评估表示怀疑在其他地方争论[19]。尽管这种比较似乎很有见地，但它们通常为您的组织提供有限的可行价值，因为它们不考虑您业务的独特战略数据需求和背景。这种成熟度评估却假装是一种标准化的简单解决方案，它并不符合成为数据驱动的复杂问题。

但是，需要对您的功能和系统进行彻底的当前国家评估，这是步骤VI的基础，在该基础上，创建了构建和维护每个功能和系统的计划。

此外，一些组织成功利用数据成熟度模型作为衡量其在数据驱动的业务转型计划中的进展并将其作为组织中各利益相关者的交流工具的工具。在这种情况下，比简单的交通信号灯更复杂的指标可能是有利的。

4.8步骤VI：计划

在为每个功能和系统进行拟合差距分析之后，现在是时候制定计划和维护数据功能所需的功能的计划了。建立与数据相关的功能和系统的计划是大多数公司对数据，分析或AI的热情。正如我们在第2.1节中看到的那样，这通常被错误地宣布为数据策略。

建立每个功能和系统通常是一个项目，需要所有者和相应的项目团队。所有者最终负责设计和构建元素。

单个项目通常是用于建立数据驱动或数据启发的组织的总体转型路线图的一部分。

实施相应的项目逐渐构建操作模型[20]对于数据，分析和AI [21]，从而激活您的数据策略。

在使用第6节中的实践示例进行更深入的深入研究数据策略设计过程之前，让我们考虑如何将业务策略设计和数据用例设计组合在一起。确定您的战略数据业务需求是必需的，这是上述第一步的一部分。我们将在下一部分中发现这一点。

在第4节，我们看到它是不是数据策略的任务定义为功能策略，以定义组织如何使用数据，分析或AI创建价值或竞争优势。这是我们在第3.5节中定义的业务，汇总和/或功能策略的任务。这些策略定义了如何获胜。有时，组织可以借助数据，分析和AI获胜。

在本节中，我们首先了解数据战略选择如何是业务策略设计的自然组成部分。有了数据策略选择，我的意思是依靠或涉及数据使用的组织理论的那些选择。

然后，我认为，数据用例发现，创新和验证应被视为业务，聚合或功能策略设计的战略选择结构过程的组成部分。This naturally bridges business strategy and data use case design, which is often perceived as one of the greatest challenges for organizations to become data-driven.

5.1 Data Choices as Part of Your Business Strategy

When you make your choices of Where to Play, How to Win and what capabilities and systems you need to win as part of your business strategy design, some of these choices might rely on leveraging data, analytics and AI.Strategic choices for data and AI capabilities are a natural part of your business strategy.我提供了some examples for this in [5]。

Therefore, organizations should consider data and AI as something, which needs to be weaved into the very fabric of their business strategy.

Strategic choices with data or AI relevance are simply part of your business, aggregation, or functional strategy.

For illustration purposes, think of your strategic choices as Lego pieces, from which you build your Strategic Choice Cascade, and data-related strategic choices are just an integral part of it.

Let’s have a look at an example, where AI is a capability as part of the strategy.

例子

I used this example, which is borrowed from [1x], already before inpart two of this series of articles [5]。It is about a salty snack producer, which uses a direct-store-delivery system, where the product is directly delivered to convenience stores and placed on the shelf by the delivery driver.

This direct-store-delivery system is labor-intensive and hence expensive, but it differentiates the company from its competition. It is a strategic choice for the regional aggregation strategy, those realization might bechartered [1r]to the Sales department.

Building data and AI solutions to predict store inventory and to generate optimal product orders for each point-of-sale helps reduce costs for the direct-store-delivery system. Hence, the choice to posses a data and AI capability, which enables the organization to predict store inventory, would be a strategic choice of the function Sales strategy, which directly supports the competitive advantage of the organization.

This data-related strategic choice provides a clearstrategic data demand[5], i.e. a strategic requirements of the organization for the usage of data and AI. It is an integrated part of the company’s strategy and reinforces other strategic choices regarding differentiation from the competition.

The example illustrates that strategy — in this case it was the strategy of the Sales function — should define how data and AI create strategic value in the organization and not a supplementary data strategy ‘misconception’.

Data use case innovation should therefore be an integral part of business strategy design. The following two subsections show that the strategy design process of the Playing to Win framework provides a natural environment to integrate data use case innovation.

5.2 Data for Problem Formulation and Possibility Generation

Data use cases are the heart of data value creation, but organizations often struggle to identify their individual set of feasible and impactful use cases.

When innovating data use cases, it is often not differentiated whether the purpose is optimization of existing processes or the creation of new businesses opportunities. However, it is essential to differentiate between operational use cases, which address routine needs or improve efficiency within existing systems, and strategic use cases, which directly support competitive advantage and enable new ways of operating. The latter should be part of business strategy design.

There exist well-known and proven formats and techniques for use case discovery and innovation [2,22 号]。These often incorporate elements of Design Thinking and are called Data Thinking [23], due to the inherent iterative approach required.

In fact, these formats and techniques can — and should — be directly integrated into the business strategy design process rather than being stand-alone exercises. This aligns nicely with the Strategic Choice Structuring Process, which is not surprising, as it also leveragesDesign Thinking [1o]。Thus, the Playing to Win framework allows to naturally combine data use case innovation and business strategy design, bridging the gap between business and data folks.

Integrating data use case innovation into the business strategy design process contributes to bridging the gap between business and data folks

This way, the business strategy design team can leverage data and AI as a tool to create strategic innovations by combining human creativity with data insights. This applies to Step 1 – problem formulation – as well as to Step 3 – possibility generation. Let’s have s look at an example.

例子

Consider the salty snack food company as in the previous examples. The salty snack food company noticed declining appeal among health-conscious consumers. An analysis of sales data, combined with social media sentiment analysis, highlighted a key insight: customers increasingly sought personalized snack options.受到启发by this insight, the strategy design team moved beyond merely optimizing existing offerings and stated the following problem with the correspondinghow might we问题：

1 Problem Definition

Customers are increasingly seeking personalized snack options, but our current product offerings are standardized, leading to declining market share among health-conscious consumers.

2 How Might We Question

How might we better meet the growing demand for personalized snack options among health-conscious consumers?

By combining the data-driven insights with creative formats, the team developed the following innovative possibilities:

3.a Possibility ‘Snack Kits’:Offer build-your-own snack kits in stores where customers can manually select ingredients to create their own mix, catering to their unique preferences.

3.b Possibility ‘Influencers’:Collaborate with health and wellness influencers to design branded snack bundles targeted at their followers, promoting personalization through co-creation with trusted figures.

3.c Possibility ‘Direct-to-Consumer Platform’:Design a direct-to-consumer platform where customers can input their dietary preferences and receive personalized snack bundles using AI-driven recommendations based on insights from customer data.

These possibilities weren’t simply derived from data — they were inspired by data, leveraging insights as a foundation for creative solutions that addressed both customer desires and business goals.

Furthermore, with possibility 3.c the design team created a new data use case. This illustrates that within the strategy design process, data and AI can be used to:

1. Inspire the problem formulation
2. Create new data use cases when innovating new possibilities

The implication for this is, that:

Subject matter experts for data, analytics and AI should be a natural part of any strategy design.

This connects nicely back to Section 4.4 and serves as a general motivation for including ‘business strategy design support’ as a Where to Play choice within the data offering of the data strategy.

Not only the innovation of new data use cases can be incorporated into the business strategy design, but also the subsequent iterative and incremental development.

5.3 Data Use Case Validation as Part of Testing

When organizations discover or innovate data and AI use cases, it is not always clear whether the desired solution will work as intended. There are uncertain assumptions that need to be identified, evaluated and eventually validated often using proof-of-concepts or minimum viable products. This is where design thinking meets data use case design.

This process perfectly aligns with Step 6 — Testing and Transformation — of the Strategic Choice Structuring Process outlined in Section 3.4.

The Strategic Choice Structuring Process naturally allows integrating iterative and incremental data use case development.

For possibility 3.c (Direct-to-Consumer Platform) of the previous example, this might look as follows:

Example (continued)

For each strategic possibility generated, the assumptions for the possibility being a great strategy need to be identified and evaluated. This is the task of Step 4 — What Would Have to Be True?— of the Strategic Choice Structuring Process.

Step 4 is about identifyingcritical conditionsfor customers, competition and company. To illustrate the concept, some of the critical conditions might be:

为了顾客, it would have to be true, that health-conscious consumers are willing to use a digital platform to personalize their snack options.

为了竞争对手, it would have to be true, that they are either unable or unwilling to replicate the direct-to-consumer platform quickly. This is the so called can’t/won’t test [1x,1年,1z]。

为了公司, it would have to be true that we can consistently derive meaningful and actionable recommendations from historical consumer data to personalize snack bundles effectively.

If the strategy design team would not be reasonably confident about that any of the identified critical conditions is currently true or could be made true, it would declare it as a障碍, which might stop the team from choosing possibility 3.c as final strategy. If a barrier has been identified, the team would design and conduct a test, in order to learn more and to gain (or lose) confidence that the condition could be made true. This is Step 6 of the Strategic Choice Structuring Process.

For the last of the three critical conditions above (company), a meaningful test could be to design a minimum viable product, where a data science team builds an algorithm using data available. The recommendations are then probed with a set of potential customers.

Such a test would validate, if the company has sufficient data to make meaningful predictions and if the data quality of that data as well as the prediction accuracy of the used algorithms are good enough to derive meaningful recommendations.

The example illustrates, that:

The data use case innovation and development process applied by data professionals, can be seamlessly integrated into the generic strategy design process of the Playing to Win framework.

Note that by doing so, business experts and data experts work together closely to design and probe new strategic possibilities, which addresses a common problem in organizations: the existing gaps between business and data teams.

5.4 Concluding Remarks on Business Strategy & Data

To conclude this section, we first link the strategic data demands originating from the business strategy design process back to the data strategy design process described in Section 4.3.

Strategic Data Demands as Inputs for Data Strategy

We saw that a part of Step I of the data strategy design process is to determine the strategic data demands of the organization, which then become the input in form of requirements for building a data function.

From this one may conclude that:

A solid design of business, aggregation and functional strategies is a necessary condition for success with data, analytics and AI in an organization.

And this is where in my opinionthe root cause of why organizations fail with data & AI lies [5]: With solidstrategy design becoming a lost art [1v], many organizations lack a sound strategic architecture and strategy design process. The lack of strategic competencies has many negative effects for an organization. One such effect is that organizations struggle to use data as an asset.

By leveraging the Playing to Win strategy framework to create a solid strategy architecture, organizations can embed data-related strategic innovation and resulting choices into their business strategy, generating clear strategic demands that inform and guide the design of the data function’s strategy, ensuring alignment and focus.

数据领先

As we concluded that leveraging strategic value from data, analytics and AI is the task of business strategy design, it is clear that the responsibility lies ultimately with the respective strategy owners, i.e. the corresponding leaders.

This emphasizes once more that leaders of all kinds and levels need to embrace adata culture, ensuring they understand, are willing to, are skilled to, and are required by their superiors to exploit possibilities to use data and AI as strategic levers. They need to become so called ‘data explorers’ as defined in [2]。

Leaders of all kinds need to become data explorers

Data, analytics and AI are therefore a score skill in a digital world. That does not solely apply to data professionals such as data scientist or data engineers, but is also true for business leaders.

The ROI question for programs to become data-driven

Transformations of any kind are not an end in themselves. A transformation program to become data-driven should always arise from business strategy. Such programs are the plans and initiatives designed to activate the strategy, ensuring the realization of the organization’s Winning Aspiration, Where to Play, and How to Win choices.

When it comes to committing resources to these transformation programs, I often encounter executives and leaders raising questions about the return on investment (ROI) — an understandable concern. However, the evaluation of anticipated costs should occur earlier during the business strategy design process, not during data strategy design or even at the implementation stage.

Ideally, the need for specific capabilities or systems should naturally emerge from a well-defined strategy. When ROI calculations dominate the discussion, it may indicate that the leadership perceives the proposed investments as unaligned with immediate strategic priorities​.

During the business strategy design process, the design team should ask, “What would have to be true regarding the cost of building the required capabilities and systems?” If uncertainty exists about the costs, this critical condition becomes a barrier, prompting the need for testing. Such tests must evaluate both the strategic benefit and the associated costs, ensuring that data investments are assessed within the broader context of organizational priorities.

This again highlights the importance of involving data teams in the business strategy design process and demonstrates that strategy and execution are not separate phases but instead form a seamless transition to activation as soon as capabilities and systems are designed.

The next section provides a detailed walk-through of the developed framework, applying it to a practical example that builds upon the earlier case of a salty snack company.

在Section 4.3, we introduced an amended version of the Strategic Choice Structuring Process for designing a data strategy. In Section 5, we demonstrated how data-related strategic choices are integral to the business strategy, forming a key requirement for designing the data function.

Now we have all jigsaw pieces together to do the strategy work which is required in an organization to leverage data as an asset using analytics and/or AI.

This section provides a detailed example, illustrating how a data strategy can be developed. For this, we use again the salty snack food company, that produces and sells potato chips and nuts, for which we design a strategy for a group-wide data function. The section is meant to be a deep-dive and can be skipped without losing too much context.

Recall, that we touched upon this example already in Section 3.5 and Sections 5.1, 5.2 and 5.3. We will now go through the steps of the data strategy design process depicted in Figure 11.

6.1 Step I: Identify Data & AI Demands

Strategic demands

As part of the group strategy design, the salty snack food company chose to add competitive value to the country businesses by building a new reinforcing rod. The possibility chosen entailed to centralize all major IT systems (Enterprise Resource Planning, Customer Relationship Management, Manufacturing Execution System), leading to enabling management systems that span across the entire organization.

Thus, the multinational salty snack food company decided to undergo a major IT harmonization program with the goal to centralize all systems and corresponding business processes across all countries.

Based on this, the group strategy design team also chose to build another reinforcing rod for data, analytics and AI, which should be realized through a newly group-wide data function. For instance, this entailed the harmonization of all reporting solutions for the individual countries.

From several regional Sales function strategies, another strategic demand was to provide solutions for store inventory prediction, as we saw in Section 5.1.

These strategic data demands were directly tied to the organization’s strategy, ensuring that the data function supports the competitive advantage of the organization. Strategic demands took precedence as they directly support the organization’s theory of winning, while operational and future demands addressed foundational and exploratory needs. Let’s have a look at the latter two.

Operational demands

A critical operational data demand for the new data function was to ensure that the harmonized systems and centralized reporting solutions were designed to seamlessly replace the existing local systems at the rollout time, without interrupting day-to-day operations. During the transition phase, the existing local data teams would be responsible for maintaining the continuity of current operational dashboards and reporting solutions (e.g. for sales tracking, inventory and production management).

Future data demands

Whilst preparing for the forthcoming data strategy design process, the design team discovered several future data demands through a series of interviews with the individual regional units.

One such future data demand was the planned digital manufacturing program in several producing countries, which aimed to optimize production and quality by intensively leveraging digital and data potentials. The program’s roadmap included potential data use cases, such as predictive process tuning for production machinery.

Strategy Design Team

The team for the designing the data strategy consisted of representatives from both the corporate data function and regional stakeholders.

At the corporate level, the team included the group data leader, who owned the data strategy as she was in charge of the group data function. Other members of the core team were a data governance specialist and an analytics manager overseeing centralized reporting frameworks. As the group CEO chartered the choice of building the reinforcing rod, he acted as sponsor for the data strategy design.

From the regional perspective, the core team involved the German division’s data manager, which oversees the German data, analytics and AI demands and solutions. Furthermore, a representative from the manufacturing function was part of the core team to ensure alignment with local needs, such as digital manufacturing initiatives.

In addition, external subject matter experts for strategy design, data and culture where leveraged. This ensured a sound methodical approach and allowed to address change-related challenges form the very start.

The extended team consisted of several analytics power users and stakeholders with key requirements, who were involved on demand.

This composition allowed the team to balance corporate priorities with the specific demands of regional teams.

6.2 Step II: Collect Challenges

During the preparation phase, interviews with stakeholders across the group and regional units uncovered the most pressing challenges related to data, analytics and AI. These challenges needed to be addressed during data strategy design, in order to ensure the successful implementation of the centralized data function while minimizing disruptions and maximizing the value of the group data function.

The following key challenges were identified:

1 Uncertainty about future state

• Regional units voiced concerns about the clarity of the target operating model for the group data function, especially regarding their role and autonomy in managing and leveraging data after the rollout.
• Stakeholders highlighted a lack of understanding of how the harmonized systems and reporting solutions will meet both group-level and region-specific requirements.

2 Resistance to centralization

• Several regions raised concerns about losing flexibility and responsiveness in addressing their unique data business needs, particularly in scenarios where ad-hoc reporting or rapid adjustments were required.
• Fear of a one-size-fits-all solution failing to meet local business nuances was cited as a potential hurdle to adoption.

3 Skills and capability gaps

• Regional teams highlighted a need for training and support to adapt to the new centralized systems, particularly for self-service analytics and AI use cases.
• A skill and capability gap of the local business was identified in leveraging predictive analytics use cases, such as those outlined in the digital manufacturing program.

4 Interim business continuity risks

• Maintaining uninterrupted operational reporting during the transition phase remains a top priority for local teams, as any disruptions could impact critical business decisions in areas like sales, production, and inventory management.
• Stakeholders highlighted challenges in meeting local interim demands for data and analyses. They expressed uncertainty about which new local solutions would be worth developing within the existing legacy environment prior to the roll-out and which solutions could be deprioritized.

5 Alignment with Future Programs

• The design team noted that the success of future programs, such as the digital manufacturing initiative, would depend on establishing a robust data infrastructure and architecture that is flexible enough to support emerging data use cases like predictive process tuning.

6.3 Step III: Designing a Data Strategy

The strategy design process followed the Strategic Choice Structuring Process depicted in Figure 7, which itself consists of 7 steps.

第 1 步：问题定义

The data strategy design team began by defining the primary problem they aimed to solve. The team defined the problem as:

“We need to harmonize data, analytics, and AI services across all regions based on the newly centralized IT landscape, while meeting diverse regional needs, maintaining business continuity, and enabling future data-driven opportunities.”

Step 2: How Might We?

这how might wequestion was designed to inspire innovative possibilities for the data function’s strategy while keeping the focus on delivering value and addressing the core challenges.

“How might we create value for our regional units through harmonized data, analytics, and AI solutions, while addressing their unique needs and aligning with our centralized IT landscape and governance?”

Step 3: Possibilities Generation

After defining the problem and framing the “How Might We” question, the team generated strategic possibilities.

For illustration purposes, I have limited the possibilities of the example to two. These possibilities contrast distinct approaches to structuring and operating the function, focusing on the degree of centralization and regional autonomy. Typically, teams may develop up to six possibilities and may contrast other aspects such as the level of relying on external professional service companies or how offensive or defensive the strategy should be [14]。

Possibility ‘fully centralized data function’

This possibility envisioned a fully centralized governance and operational model for the data function. All major data, analytics, and AI activities, including data integration, governance, demand management, and solution delivery, would be managed by the central team. Regional units act as data consumers, relying on standardized reporting frameworks, dashboards, and analytics solutions provided uniformly by the central team to ensure consistency, scalability, and efficiency.

Possibility ‘federated data function with regional customization’

This possibility proposed a federated model that combines centralized governance with significant autonomy for regional units. Regional teams would play an active role in demand management, analytics development, and leveraging data science/AI for local initiatives, particularly in programs like digital manufacturing.

Choice Cascades

To detail both possibilities, the team completed the Data Strategy Choice Cascade for each. This helped the team to build a detailed joint understanding of each possibility.

Fleshing out the Must-Have Capabilities and Enabling Management Systems was for the design team already the first step to activate the data strategy as it was a kind of reality check for each possibility.

Step 4: What Would Have to Be True?

In order not to blow up the content of this section, I will only focus on possibility B for steps 4–6 of the Strategic Choice Structuring Process, as it will be sufficient to illustrate the concept.

The critical conditions were collected for customer, company and competition.

Recall the translations for company and competition we defined in Section 4.2 for this exercise.

顾客

1. Regional data customers value the ability to create their own data solutions to meet their country-specific needs
2. Regional data customers have the competencies to tailor analytics and AI solutions to their specific needs through locally available tools and experimentation.
3. Regional business units trust that centralized governance frameworks enable, rather than hinder, their ability to innovate and adapt to local needs.

公司

1. The central data team is able to provide effective governance frameworks and tools that empower regions to innovate while maintaining alignment with group-wide standards.
2. Regional teams successfully develop and share innovative use cases and best practices that are evaluated by local teams and adopted across other regions where sensible.
3. Resources are allocated effectively to balance support for both central governance and regional customization, without overburdening either side.
4. The company can implement and maintain the federated model in a cost-effective way, ensuring resource allocation supports both central and local capabilities.

竞赛

1. The federated model outperforms do-it-yourself workarounds (e.g. with Excel) or external consultancies in delivering faster, more integrated, and cost-effective analytics and AI solutions.

Step 5: Barriers to Choice

From these critical conditions, the data strategy design team was least confident about two:

客户3：“Regional business units trust that centralized governance frameworks enable, rather than hinder, their ability to innovate and adapt to local needs.”

The reason for the low confidence was, that regional business units have historically relied on local data experts or external consultants for data solutions. It was unclear whether the data consumer would accept the new central authority.

Company-4：“The company can implement and maintain the federated model in a cost-effective way, ensuring resource allocation supports both central and local capabilities.”

The reason for the low confidence was, that the team had at the time not enough knowledge about the costs of both, data platform and tools as well as staff costs as the overall data operating model was not detailed to a sufficient extent.

Step 6: Testing & Transformation

To address these uncertainties, the data strategy design team developed a series of tests to validate the two critical conditions.

To address the identified barriers, the data strategy design team devised two focused tests:

Test 1 (trust in new data governance model)

The design team created a rough sketch of the future data governance operating model, that made the implications for local data consumers more tangible. The draft was shared with local key stakeholders and their feedback was incorporated within a next design iteration of the data governance operating model.

Test 2 (cost efficient implementation)

The design team created a first draft of the data operating model detailing the technological solution to estimate the future total cost of ownership for technology. They also detailed the newly required expert roles to estimate the future staff cost of the data function to be built.

These tests provided actionable insights into the assumptions and validated the possibility’s potential to succeed.

Step 7: Choice

In the context of this example, the data strategy design team decided to pursue the federated model, as it offered the highest likelihood of winning with the data customers by better balancing the need for regional innovation and customization while achieving the group’s centralized governance and alignment goals.

6.4 Steps IV-VI: Activating Strategy

Once possibility B was selected as the new data strategy, the next task was to activate it by addressing the implementation requirements. The steps IV to VI ensured that the strategic choices made during the design process were translated into operational realities, enabling the data function to deliver on its strategic demands. By activating the data strategy in alignment with the business strategy, the organization ensured that data capabilities directly support its competitive goals.

Step IV: Determining operational capabilities & systems

After the strategic capabilities and systems had been defined in Step III as part of the data strategy, the design team identified the operational capabilities and enabling management systems required to implement the federated model effectively.

Key operational capabilities included efficient report generation. These were complemented by operational systems such as BI and ETL tools as well as systems to monitor data quality.

Step V: Data Maturity Assessment

To assess readiness for the data strategy, the team conducted a comprehensive data maturity assessment for all capabilities and systems required to activate the strategy and across all regions.

They used a traffic light evaluation metric to assess current capabilities and systems in areas such as data modeling, reporting, and local experimentation. Capabilities and systems with “green” ratings were deemed ready to adopt the data strategy, while those flagged as “yellow” or “red” required targeted interventions in form of dedicated projects to build and maintain the new capabilities and systems.

The assessment also revealed regions where experimentation and self-service adoption were underdeveloped, providing a roadmap for prioritizing maturity-building efforts.

Step VI: A Plan to Build Data, Analytics and AI Capabilities

Based on the maturity assessment findings, the team developed a phased plans to build and sustain the required capabilities and systems with priority assigned to those elements of strategic nature.

Each region was assigned specific milestones tailored to its maturity level, with clear timelines and success metrics. To ensure continuous improvement, feedback loops were integrated into the plan, enabling the central team to refine tools and frameworks based on regional experiences and evolving needs.时间

his article aimed to demystify data strategy by addressing common misconceptions that prevent organizations from fully leveraging data, analytics, and AI.Contrary to popular belief, a data strategy is not a plan for building data capabilities, a set of guidelines for managing data, or a supplementary element to business strategy. Instead, it is a functional strategy — the strategy for the data, analytics, and AI function, which provides data services to internal customers.

The data strategy is designed on the basis of the organization’s overarching business strategy. And it is business strategy, not data strategy, which defines how the organization leverages data to achieve competitive advantage.

By leveraging the Playing to Win framework, I demonstrated that data-related strategic choices must be embedded within the organization’s business strategy, forming the input for a robust data strategy. By integrating data use case innovation into the business strategy design process and deriving clear strategic demands, organizations ensure that their data strategy serves as a functional enabler, delivering capabilities that directly support competitive advantage.

Using the practical example of a salty snack food company, I illustrated how organizations can apply this approach to bridge the gap between theory and practice. The structured process not only aligns the data function with business needs but also allows data to become a key driver of competitive advantage.

Organizations that continue to treat data strategy as a standalone supplement or a list of technical initiatives will fail to realize its true potential. Instead, by weaving data-related strategic choices into the very fabric of their business strategy, they can build a cohesive, purpose-driven strategy architecture that enables them to win.

The promise of data, analytics, and AI lies not in their technical complexity but in their strategic and cultural alignment. By embedding data innovation into business strategy design, organizations can move beyond the data hype and leverage data as a true strategic asset.

I welcome your feedback, suggestions, and questions. Do you agree, disagree, or have additional thoughts? I look forward to engaging in discussions with you in the comments here on Medium.

Playing to Win/ Practitioner Insights(2024), website with list of articles[1a] Roger Martin,

What Strategy Questions are You Asking?(2023), Medium article of the‘Playing to Win Practitioner Insights’系列[1b] Roger Martin,

Overcoming the Pervasive Analytical Blunder of Strategists(2021), Medium article of the‘Playing to Win Practitioner Insights’系列[1c] Roger Martin,

Strategy, Strategy Everywhere(2023), Medium article of the‘Playing to Win Practitioner Insights’系列[1d] Roger Martin,

Strategy is Singular(2023), Medium article of the‘Playing to Win Practitioner Insights’系列[1e] Roger Martin,

The Origins of Playing To Win(2023), Medium article of the‘Playing to Win Practitioner Insights’系列[1f] Roger Martin,

(Playing to Win) x 5 (2024), Medium article of the‘Playing to Win Practitioner Insights’系列[1g] Roger Martin,

Strategy is what you DO, not what you SAY(2020), Medium article of the‘Playing to Win Practitioner Insights’系列[1h] Roger Martin,

The Best Strategy Icebreaker(2024), Medium article of the‘Playing to Win Practitioner Insights’系列[1i] Roger Martin,

战略选择构建过程(2024), Medium article of the‘Playing to Win Practitioner Insights’系列[1j] Roger Martin,

The Five Deadliest Strategy Myths, Medium article of the‘Playing to Win Practitioner Insights’系列[1k] Roger Martin,

Why Planning Over Strategy?(2022), Medium article of the‘Playing to Win Practitioner Insights’系列[1l] Roger Martin,

Strategy vs. Planning: Complements not Substitutes(2024), Medium article of the‘Playing to Win Practitioner Insights’系列[1m] Roger Martin,

From Strategy to Planning(2021), Medium article of the‘Playing to Win Practitioner Insights’系列[1n] Roger Martin,

Strategy as Problem-Solving(2020), Medium article of the‘Playing to Win Practitioner Insights’系列[1o] Roger Martin,

Strategy & Design Thinking(2020), Medium article of the‘Playing to Win Practitioner Insights’系列[1p] Roger Martin,

The Secret to Knitting Strategy Together Corporate-Wide(2021), Medium article of the‘Playing to Win Practitioner Insights’系列[1q] Roger Martin,

Corporate vs. Business Unit Strategy(2022), Medium article of the‘Playing to Win Practitioner Insights’系列[1r] Roger Martin,

Strategic Choice Chartering(2020), Medium article of the‘Playing to Win Practitioner Insights’系列[1s] Roger Martin,

Understanding the True Building Blocks of Corporate Strategy(2022), Medium article of the‘Playing to Win Practitioner Insights’系列[1t] Roger Martin,

Who Should Do Strategy?(2023), Medium article of the‘Playing to Win Practitioner Insights’系列[1u] Roger Martin,

克服一体化战略挑战(2024), Medium article of the‘Playing to Win Practitioner Insights’系列[1v] Roger Martin,

The Lost Art of Strategy(2021), Medium article of the‘Playing to Win Practitioner Insights’系列[1w] Roger Martin,

Is or Is Not The Opposite Stupid on its Face?(2024), Medium article of the‘Playing to Win Practitioner Insights’系列[1x] Roger Martin,

Distinguishing How-to-Win from Capabilities in Your Strategy Choice(2021), Medium article of the‘Playing to Win Practitioner Insights’系列[1y] Roger Martin,

Can Your Strategy Pass Its Most Important Test?(2021), Medium article of the‘Playing to Win Practitioner Insights’系列[1z] Roger Martin,

Untangling Value Proposition & Competitive Advantage(2022), Medium article of the‘Playing to Win Practitioner Insights’系列[2] Sebastian Wernicke,

Data Inspired(2024), book in German language published by Vahlen[3] Caroline Carruthers and Peter Jackson,

Data Driven Business Transformation (2019), book published by Wiley[4] Quote Investigator,

Quote Origin: It’s Difficult to Make Predictions, Especially About the Future, website accessed on Jan. 4th 2025[5] Jens Linden,

The Root Cause of Why Organizations Fail With Data & AI(2024), Medium article published inToward Data Science[6] James Martin,

Design and strategy for distributed data processing (1981), book published by Prentice-Hall[7] Google Trends,

Worldwide interest of the term ‘data strategy’(2024), website[8] Roger Martin,

A Plan Is Not a Strategy(2022), video[9] AWS,

What is a data strategy?(2024), website accessed on Dec. 22nd 2024[10] SAP,

What is a data strategy?(2024), website article accessed on Dec. 22nd 2024[11] Wikipedia,

乐高电影(2014), WIKI entry[12]

Roger Martin’s website(2024)[13] A. G. Lafley and Roger L. Martin,

为赢而战(2013), book published by Harvard Business Review Press[14] Michael Goitein,

How a leading strategy expert combined Design Thinking and scientific inquiry to allow anyone to create winning strategies, blog entry accessed on Dec. 25th 2024[15] Roger Martin, Jennifer Riel,

The One Thing You Need to Know About Managing Functions(2019), article published in Harvard Business Review[16] Jennifer Riel,

How to Make Your Strategy Real(2024), IDEOU Blog entry[17] Jens Linden,

The Data Strategy Choice Cascade(2024), Medium article published in走向数据科学[18] Alex Osterwalder,

Achieve product-market fit with our brand-new Value Proposition Canvas(2012), website accessed on Dec. 30th 2024[19] Jens Linden,

Unser Datenreifegrad ist „Reactive“ oder“Managed“ — und nu?!(2023), LinkedIn blog entry in German language[20] Andrew Campbell, Mikel Gutierrez, Mark Lancelott,

Operating Model Canvas(2017), Book published by Van Haren[21] Jens Linden,

Data & AI Operating Model Canvas(2023), LinkedIn blog entry in German language[22] Datentreiber,

The Datentreiber Method: Data & AI Business Design(2024), website accessed on Dec. 30th 2024[23] Wikipedia,

数据思维(2024), website entry on accessed on Dec. 30th 2024[24] Leandro DalleMule,

Thomas H. Davenport What’s Your Data Strategy?(2017), article published in Harvard Business Review除非另有说明，所有图片均由作者提供。

All Lego illustrations have been created with

MECABRICKS。I used ChatGPT 4o as sparring partner, to generate some of the example ideas.

I used ChatGPT 4o as sparring partner, to generate some of the example ideas.