石头点评：

成也文化，败也文化。因为，不同的文化将导致不同的结果。

另外，文化通过行为而不是言语而改变。

 

大多数从事数据科学、人工智能和数字转型工作的人都痛苦地意识到，阻碍他们努力的往往是文化，而不是技术。许多人甚至知道大概的步骤（他们应该采取解决这个问题的措施），将注意力和资金投入到改变人们的心态和公司使用数据的方式上。但是，一旦公司和领导者开始了解如何做到这一点的实质细节，就很难知道实施这些步骤实际上是什么样子的。

要了解改变文化和鼓励数字思维需要什么，看看另一家公司实际上是如何做到的，这是很有帮助的。哪些策略有效，哪些是死胡同？员工有什么信息？你应该从哪里开始？

在本文中，我们开始通过总结科威特海湾银行新数据方案的头两年来解决这一差距，在该方案中，我们努力建立一种拥抱数据的文化。虽然两年的时间太短了，无法宣布这项工作完成，但数百人正在以不同的方式完成他们的工作，并以新的、令人兴奋的方式使用数据。

我们中的一位，阿尔瓦什，于2021年2月被聘为海湾银行的第一位首席数据官，作为战略计划的一部分启动全面数字化转型银行的业务，任务是提供数据驱动的客户体验。在这一任务范围内，她的工作是整理海湾银行的计划，建立一个小团队，并执行。虽然她是个成功的技术专家，但她知道自己必须成长为这个角色。所以，她雇了我们中的另一个，雷德曼，来建议。

在她准备开始新工作时，她考虑了一个常见的建议：获得一些快速的胜利，比如清理客户数据库、构建数据湖以改善访问权限，或者改进监管报告。但她的老板、副首席执行官拉古·梅农（Raghu Menon）是一位行业资深人士，当这个低垂的水果被证明是腐烂的时候，她看到了太多的数据项目无法启动。相反，他建议她首先“不要急于求成”（ “get the basics right.”）。

我们从梅农的见解中得到了两条信息。

首先，从数据质量开始。许多人将认为这是一个奇怪的选择，但在数据空间，特别是在数字转型和数据驱动的客户体验方面，没有什么比质量更基本的了。糟糕的数据是常态。它是个恶毒的杀手，给日常工作增加了巨大的成本，并使变现、分析和人工智能要困难得多。

第二，仔细思考我们将如何让每个人都参与进来，我们希望创造的文化，以及组织结构更加有效。具体来说，我们想把两件事说清楚：每个人都需要数据来完成他们的工作（例如，他们是数据客户），他们还创建下游使用的数据（例如，他们是数据创建者）。当人们进入这些角色时，他们会共同努力，寻找和消除不良数据的来源，质量也会迅速提高。在这一过程中，以这种方式攻击数据质量会导致人们参与并利用数据增强自己的能力。

在收费（charge off）之前，我们征求了各级许多长期员工的意见：员工会觉得攻击数据质量有吸引力吗？他们是否会发现新的角色作为数据客户和创建者角色来增强权能？他们的反馈告诉我们，虽然有些人需要说服，但很多人都会发现很多值得喜欢的地方，但我们需要提供一些简单的“入门”任务。这也鼓励了我们：员工们说，这些角色可以改变银行。

 

构建扩展数据团队

阿尔瓦什的小团队怎么能让整个银行1800人加入？为此，我们设计了一个“数据大使”计划，本质上是一个由领导努力为团队带来数据质量的人组成的网络。为了建设它，阿尔瓦什会见了该行管理委员会，解释梅农的指示，激励人们对质量的关注，并描述了她所寻求的人的概况。她还承诺提供培训和支持，整个银行都将在这一过程中学习。阿尔瓦什的“先人，后技术”方法引起了委员会的共鸣，委员会的13名成员提名了140名准大使。

尽管准大使如预测的那样是由高级领导人提名的，但许多人仍持怀疑态度。他们认为这个角色只是增加了工作。所以阿尔瓦什和她团队与人力资源部联合使工作有趣、有价值和乐趣。他们以三种方式这样做：

世界级的培训：大使们被告知，他们将学习和做一些对他们在整个职业生涯中都有很好的帮助的事情。Covid使交付变得困难，但培训——在五场会议中面对面进行——探讨了他们作为数据客户和创建者的角色和责任，向他们展示了如何进行第一次数据质量度量，并提供了一种找到和消除错误根本原因的方法。最后一次会议是一个动手实验室，重点是自助分析和数据可视化。每场会议都有一个在职任务，帮助大使们开始工作。

媒体：大使们得到了大量的宣传，内部通讯、社会频道和当地报纸强调了他们的工作。

品牌：数据团队与营销部门联系，为数据大使计划创建徽标，并通过提供品牌赠品，如数字笔记本，建立知名度。

即使是最怀疑的大使，在第一届会议结束时也看到了赋予个人权力的机会。他们发现，数据和分析不仅仅是为技术人员准备的，而且是他们自己可以做的事情。他们把这些信息带回了他们的团队。

 

让每个人就位（on Board）

下一个目标是其他人，特别关注那些在分行、问询中心和销售团队工作的人，因为银行客户体验如此之大都依赖于这些团队。我们设计了一个"数据101计划"，解释了他们作为数据创建者和客户的角色，并强调了数据质量对银行各级成功的影响。有趣的是，担任这些角色的人创造了银行最重要的数据，但永远不知道为什么。数据是离他们脑海最远的东西。最后，阿尔瓦什努力确保数据101计划现在包括在所有新员工入职中。

了解他们工作的更广泛影响，这比大多数银行的“只做销售”方法更令人兴奋。例如，直销代表Fahad AlRefaei在接受培训后找到阿尔瓦什，解释数据101计划是如何改变他态度的。在完成销售后开立新账户时，他现在会格外注意他个人不使用的数据，因为他知道银行内的数据客户需要这些数据。其他人提供了类似的反馈——一旦他们了解到质量数据有多重要，他们就认真对待了作为数据创建者的责任。他们感到被赋予了权力，并与银行的总体成功更好地联系在一起。数千个这样的小步骤使每个人都更容易为客户参与带来更多、更可信的数据。

 

创新走向前台/Innovation to the Fore

赋能是一件美丽的事情！正如我们所预期的那样，大使和银行的其他人开始合作，进行度量，瞄准数据清理，消除错误的根本原因。然后，在某种程度上，大使和正式雇员开始以新的方式使用培训中提供的方法和工具，自行创新。例如，两位大使联手改进反洗钱模式，增强分支机构的客户体验，同时降低风险和运营费用。今年早些时候，阿尔瓦什和她的团队组织了首届“创新锦标赛”在海湾银行。数百人参加了比赛——这是参与和赋权正在扎根的肯定迹象。

如上所述，两年时间就声称数据文化已经完全嵌入海湾银行还为时过早。很多事情还是会出错的。此外，阿尔瓦什和海湾银行有更大的雄心壮志，包括人工智能、共享语言、数据驱动的创新、数据供应链管理和货币化。这些努力中的许多都需要大数据、先进技术、拥有高级学位的专业人员，以及大使和其他人的支持。

 

经验教训总结

我们非常肯定，建立一个伟大的数据文化有很多途径。美国国务院采取了“浪涌哲学”，一次专注于一个部门，其他人可以利用与人工智能相关的兴奋点来做到这一点（原文：others may do so by taking advantage of the excitement around artificial intelligence. ）。不过，我们认为海湾银行的经验说明了一些重要的观点。

改变现有的文化是很困难的，如果你每一步都在与它作斗争，那就更困难了。因此，相反，寻找现有文化将包含的东西，并将推动您所希望的数据文化向前发展。例如，从事医疗保健工作的人可能会被“帮助人们过上更长、更健康的生活”。解释数据计划将如何推进这一任务，增加你的机会。

从第一天起就开始建设新文化是很重要的，即使这样做不是主要的任务。这与传统观念背道而驰，传统观念建议你需要快速获胜来帮助建立支持。但是，速赢的努力往往会走捷径，对人和文化进行粗暴对待，增加了这些项目失败的可能性。此外，成功的速赢可能会导致公司错误地得出结论，他们不需要担心人和文化，这将导致未来的失败。相反，目标是“重大胜利”，充分拥抱业务成果、组织、人员和文化。

其次，要改变一种文化，你需要让每个人都参与进来。在海湾银行，我们联系了管理委员会、人力资源、营销和企业沟通部等，并及时收到了所有人的支持。我们通过面对面提供培训并为每个小组量身定制培训，强调了数据的重要性。事实上，Data 101有20多个版本。此外，文化通过行为而不是言语而改变。所以我们明确了我们希望人们做什么，而不仅仅是我们希望他们如何思考或感受。我们培训中提供的任务帮助人们开始了努力。

第三，就像我们所做的那样，给予数据质量作为起点。虽然许多人认为数据质量是所有数据中最不性感的话题，但它是让每个人都参与进来的好方法，而且是基础性的。您无法在坏数据之上构建一个好用的数据程序。

最后，建设文化需要坚持和勇气。预计会有一些糟糕的日子，但请记住将来一定有更大的回报。

 

英文原文：

What Does It Actually Take to Build a Data-Driven Culture?

by Mai B. AlOwaish and Thomas C. Redman

May 23, 2023

Summary.

Building a data driven culture is hard. To capture what it takes to succeed, the authors look at the first two years of a new data program at Kuwait’s Gulf Bank in which they worked to build a culture that embraced data, and offer a few lessons. First, it is important to start building the new culture from day one, even as doing so is not the primary mandate. Second, to change a culture, you need to get everyone involved. Third, give data quality strong consideration as the place to start. Finally, building this new culture takes courage and persistence.

Most people who work on data science, AI, and digital transformation are painfully aware that it is often culture, not technology, that stymies their efforts. Many even know thehigh-level steps they’re supposed to take to fix this problem — invest attention and money into changing people’s mindsets and how the company uses data. But once companies and leaders get into the nitty-gritty details of how to do this, it can be hard to know what implementing those steps actually looks like.

To understand what it takes to change a culture and encourage a digital mindset, it’s helpful to see how another company isactually doing it. Which strategies worked and which were dead ends? What messaging landed with staff? Where should you actually start?

In this article, we begin to address this gap by summarizing the first two years of a new data program at Kuwait’s Gulf Bank in which we worked to build a culture that embraced data. While two years is far too short a time to claim the job complete, hundreds of people are doing their jobs differently and using data in new, exciting ways.

One of us, AlOwaish, was hired as Gulf Bank’s first chief data officer in February, 2021, as part of the strategic plan tolaunch a complete digital transformation of the bank’s operations, with a mandate to provide data-driven customer experiences. Within that mandate her job was to sort out Gulf Bank’s plans, build a small team, and execute. Though a successful technologist, she knew she would have to grow into the role. So, she hired the other of us, Redman, to advise.

As she prepared to start her new job, she considered a common piece of advice: Score some quick wins, such as cleaning up a customer database, building a data lake to improve access, or improving regulatory reporting. But her boss, deputy CEO Raghu Menon, was an industry veteran and had seen too many data programs fail to launch when the low-hanging fruit turned out to be rotten. Instead, he counseled her to first “get the basics right.”

We took two messages from Menon’s insights. First, start with data quality. Many will view this as an odd choice, but in the data space, and especially for digital transformation and data-driven customer experience, nothing is more basic than quality.Bad data is the norm. And it is a vicious killer, adding enormous costs to day-in, day-out work, and making monetization, analytics, and artificial intelligence far more difficult.

The second message was to think carefully about how we would get everyone involved, the culture we wished to create, and the organizational structures needed to be effective. Specifically, we wanted to drive home two things: that everyone needs data to do their job (e.g., they are data customers), and that they also create data used downstream (e.g., they are data creators). When people step into these roles, they work together to find and eliminate the sources of bad data and quality improves quickly. Along the way, attacking data quality in this manner leads people toengage and empower themselves with data.

Before charging off, we sought input from many long-term employees at all levels: Would employees find attacking data quality appealing? Would they find new roles as data customer and creator roles empowering? Their feedback told us that while some people would need convincing, plenty would find a lot to like, but we needed to provide some easy “getting started” assignments. It also encouraged us: done well, employees said, these roles could transform the bank.

Building the Extended Data Team

How could AlOwaish’s small team get the entire bank of 1,800 people on board? To do so, we designed a “data ambassadors” program, essentially a network of people who would lead efforts to bring data quality to their teams. To build it, AlOwaish met the bank’s management committee to explain Menon’s charge, motivate the focus on quality, and describe the profile of people she sought. She also promised to provide training and support and that the entire bank would learn along the way. AlOwaish’s “people, then technology” approach resonated with the committee, and its 13 members nominated 140 ambassadors-to-be.

Even as ambassadors-to-be had been nominated by senior leaders, as predicted, many were skeptical. They saw the role as nothing but added work. So AlOwaish and herteam joined up with human resources to make the work interesting, rewarding and fun. They did so in three ways:

•  

World-class training: The ambassadors were told     that they would learn and do things that would serve them well throughout     their careers. Covid made delivery difficult, but the training — delivered     face to face in five sessions — explored their roles and responsibilities     as data customers and creators, showed them how to make their first     data-quality measurement, and provided a method to find and eliminate the     root causes of error. The final session was a hands-on lab focused on     self-serve analytics and data visualization. Each session featured an     on-the-job assignment to help ambassadors get started.

•  
•  

Media: Ambassadors received lots of     publicity, as internal newsletters, social channels, and local newspapers     highlighted their work.

•  
•  

Branding: The data team reached out to     marketing to create a logo for the data ambassadors program and build     awareness by providing branded giveaways, such as a digital notebook.

•  

Even the most skeptical ambassadors saw opportunities for personal empowerment by the end of the first session. They came to see that data and analytics weren’t just for techies, but something they could do on their own. And they carried these messages back to their teams.

Getting Everyone on Board

The next target was everyone else, with special focus on those working in branches, the call center, and on sales teams, on which so much bank customer experience depended. We designed a “Data 101 program” that explained their roles as data creators and customers, and highlighted the impact of data quality on the bank’s success at all levels. Interestingly, people in these roles create much of the bank’s most important data, but never knew why. Data was the furthest thing from their minds. Finally, AlOwaish worked to ensure that Data 101 is now included in all new employee onboarding.

Understanding the broader reach of their work made it more exciting than the “just make the sale” approach in most banks. For example, Fahad AlRefaei, a direct sales representative, sought AlOwaish out after a training to explain how Data 101 changed his attitude. When opening a new account after closing a sale, he now he pays extra attention to the data he doesn’t personally use, because he knows that data customers within the bank need it. Others provided similar feedback — once they learned how important quality data was, they took their responsibilities as data creators seriously. They felt empowered and better connected to the bank’s overall success. Thousands of such small steps make it easier for everyone to bring more, and more trusted, data to customer engagements.

Innovation to the Fore

Empowerment is a beautiful thing! As we expected, ambassadors and others across the bank began working together, making measurements, targeting data cleanups, and eliminating root causes of error. Then, somewhat organically, ambassadors and regular employees began using methods and tools provided in the training in new ways, to innovate on their own. For example, two ambassadors joined forces to improve anti-money laundering models, enhancing the customer experience in the branch, while simultaneously reducing risk and operational expense. Earlier this year, AlOwaish and her team organized theinaugural “innovation tournament” at Gulf Bank. Hundreds competed — a sure sign that engagement and empowerment are taking root.

As noted above, two years is too soon to claim that a data culture has become fully embedded at Gulf Bank. Much can still go wrong. Moreover, AlOwaish and Gulf Bank have larger ambitions, including artificial intelligence, shared language, data-driven innovation, data-supply-chain management, and monetization. Many of these efforts will require big data, advanced technologies, professionals with advanced degrees, andsupport from ambassadors and others.

Lessons Learned

We are quite certain that there are many paths to building a great data culture.The U.S. Department of State has adopted a “surge philosophy,” focusing on one department at a time and others may do so by taking advantage of the excitement around artificial intelligence. Still, we think Gulf Bank’s experiences illustrate some important points.

It is hard to change an existing culture and harder still if you’re fighting it every step of the way. So instead, look for things the existing culture will embrace and will move the data culture you desire forward. For example, people working in health care may be bought into “helping people lead longer, healthier lives.” Explaining how a data program will advance that mission increases your chances.

It is important to start building the new culture from day one, even as doing so is not the primary mandate. This runs counter to conventional wisdom, which advises that you need quick wins to help build support. But quick win efforts often take shortcuts, running roughshod over people and culture and increasing the likelihood that these projects fail. Further, successful quick wins may lead companies to falsely conclude they need not worry about people and culture, setting themselves up for future failures. Instead, aim for “significant wins,” that fully embrace business results, structure, people, and culture.

Second, to change a culture, you need to get everyone involved. At Gulf Bank, we sought out the management committee, human resources, marketing and corporate communications and received timely contributions from all. We emphasized data’s importance by delivering the training face-to-face and tailoring it to each group. Indeed, there were more than 20 versions of Data 101. Further, cultures change through deeds, not words. So we were explicit about what we wanted people to do, not just how we wanted them to think or feel. The assignments provided in our training helped people jumpstart the efforts.

Third, give data quality strong consideration as the place to start, as we did. While many view data quality as the least sexy topic in all of data, it is a great way to get everyone involved and it is foundational. You cannot build a great data program atop bad data.

Finally, building a culture takes persistence and courage. Expect to have some bad days, but keep the larger prize fully in mind.