11
Creating a Data-Driven Culture in Business
Transforming organizations into data-driven ones requires more than just recipes that can be implemented to solve certain problems. There is a lot that we can learn from those who have already walked the path of changing organizations to become more efficient and successful when using data. In this chapter, we will dive into the experience of data leaders and learn about the following:
- How they started their careers in data
- What skills they had to master to become successful data leaders
- What technology and data mean and how they are used in the context of business
- How data-driven cultures can be implemented in organizations
Hopefully, their experience can guide both your transformation to become a data leader, as well as the organizational transformation required to become data-driven.
The invited leaders that will be sharing their perspectives are:
- Patrick Klink: Managing director and founder of ONBRDNG, a Dutch company that helps companies digitally transform, he is a seasoned data leader and pioneer. Patrick’s experience makes him one of the most sought-after digital transformation and growth experts in the Dutch market and abroad, as he helps companies transform from traditional businesses into all-around digital companies:
Before ONBRDNG, Patrick was the director of product and technology at RTL and an international leader in broadcast, content, and digital, with interests in 60 television channels. He also supported several companies as chief of product/data/tech or CTO. Patrick serves as a supervisory board member and provides venture capital for several scale-ups.
- Michael Curry: He is a general manager, chief product officer, strategy leader, and brand and product storyteller. With 30 years of experience in all aspects of enterprise business-to-business software, he is a technology product expert with skills and experience in product management, marketing, development, implementation, and sales. To complement this, he is also an expert in the financial services and healthcare industry, as well as an accomplished speaker, being a guest lecturer at Harvard and Penn State:
Michael has been at IBM for 17 years, working as vice president of IBM Watson Health. Leading the divestiture of Watson Health business to Francisco Partners, he helped simplify business from 30+ products to six major product areas and restructured development under a single development leader, reducing the development run rate and SaaS operations costs by double digits. He also led the shift to a public cloud, improved EBITDA by triple digits while improving NP, and led the divestiture of Watson Health business to Francisco Partners.
- Florian Prem: With more than 20 years of experience in data and analytics, Florian is the Chief Data Officer (CDO) at Deloitte Technology in Zurich, Switzerland:
Florian is part of the Deloitte Technology leadership team as the inaugural CDO. Deloitte Technology is the department of Deloitte Consulting, which is a globally integrated technology organization that spans more than 85 countries and has more than 10,000 professionals.
Before becoming the global CDO, Florian was the first CDO for Deloitte Switzerland with a mandate for data strategy, governance, risk and security, data management, data analysis and AI, automation, collaboration, and document and records management platforms.
- Micaela Kulesz: A behavioral data scientist with experience in AI, experimental economics, and consumer behavior. She is a lead data scientist, a leader in data and innovation, and a machine learning engineer (on the AI team). Now, she is the lead data scientist for data and innovation, with a focus on retail tech, developing solutions for the retail sector:
With a background in economics from the University of Buenos Aires, she received her Ph.D. at the University of Bremen in experimental and behavioral economics, applied game theory, and experiment metrics. After this, she worked as a researcher at the Swedish University of Agricultural Sciences, before transitioning to a data scientist position.
She chooses to work with honest, direct, and kind people and has a vision of data science as a pseudo-art that involves having creative conversations with colleagues and everyone in general.
- Jack Godau: Data Leader, currently Chief Technical Officer (CTO) at Doctorly in Berlin, Germany. He enjoys talking about startups, healthcare, recruiting, the future of work, and digital transformation:
Jack is an empathetic, cooperative, and passionate leader who has built on a strong career of being a technical expert with strong sales skills. He has worked with organizations of all types and sizes to manage their digital evolution by providing technical and strategic leadership, guidance, and coaching. Highly proficient in team building and in fostering an inclusive culture to enable local and remote teams to succeed, Jack has successfully built and grown strategic partnerships at an international level.
- Julio Rodriguez Martino: A dynamic leader with more than 15 years in establishing, developing, and directing high-performing, globally diverse teams. Julio is an analytics/data science/machine learning/artificial intelligence professional, with skills and experience in team management, as well as being a writer and mentor:
Julio has leveraged a strong educational background in physics to solve overly complex problems resulting in the development of innovative solutions. He is a team leader across various functions, including analytics, artificial intelligence, machine learning, data science, and engineering. Julio is a compelling communicator, well-versed in engaging key stakeholders, translating business requirements into technical requirements, and advising C-level executives as technical SMEs. With a strong aptitude for driving innovation and knowledge-sharing across cross-functional internal departments, Julio is a lifelong learner with a passion for research and development while remaining up to date with the latest developments within the field of AI and ML.
- Agustina Garcia Hernandez: Data and strategy director for Anheuser-Busch InBev. She is an economist and analytics pioneer, with a great strategic thinking mindset that brings business and technical data together. Her focus is to empower teams from diverse cultures and profiles so that they achieve impact results for the commonwealth. She has experience and skills having developed deep management and directing expertise in multinational companies. She has also led local projects as well as international ones, creating solid expertise in project management, which has given her the extra mile in her management career:
Agustina’s key strengths are leveraged by bringing business knowledge and data together; she has experience bringing teams of people from diverse cultures together, leading projects, and obtaining integrated results. She focuses on creating opportunities without losing the big picture. Empathetic, she understands the need and converts it into a solution. She bridges teams and stakeholders under the same goals with attractive oral and graphic presentations that inspire actions and results. Her processes align intuitive business knowledge with data-based insights to improve the decision-making process while connecting the dots from different areas of expertise, transforming complexity into simple and executable answers. Agustina leverages different methodologies such as focus groups, polls, segmentation, and machine learning/artificial intelligence to arrive at business solutions that optimize the usage of diverse disciplines.
- Bob Wuisman: With a background in human resources and business, Bob has leveraged his experience in business processes to become the data operations director at Ebiquity Digital Innovation Center in Utrecht, Netherlands:
Building on his years of experience as a business process consultant, Bob has successfully built business intelligence environments in various businesses. With a holistic vision and process-driven mindset, Bob thrives on building teams and sustainably growing data-driven operations, and with skills and experience in professional competencies including people, process, and project management.
- Wim van der Meer: With a focus on sustainability and water management, Wim is the CDO of Waterschap Vallei & Veluwe in Apeldoorn, Netherlands:
Wim has been part of the Water Board Vallei en Veluwe as program manager for digital transformation. He was responsible for the development of a data-driven network organization. As project leader implementer, Wim has overseen the Environment Act Permit, supervision, and enforcement, as well as having been a water board inspector. His background is in water management, law, and agile organization.
Let us start by looking at the path that these leaders made to start working with data.
Starting to work with data
Starting the journey to become a data leader requires improving your skill set to incorporate certain capabilities that allow you to not only understand descriptive visualizations, basic statistical concepts, as well as tech and data concepts but also create the skills to lead teams and understand business requirements.
This section consists of answers that our data leaders gave when asked about which skills and capabilities were required to become successful data leaders.
Julio Rodriguez Martino
I have a degree in science, and I focused my scientific career on experimental physics. Having experience in data analysis, statistics, and problem-solving made data science a natural choice when moving to the industry.
- Which are the areas in which you needed to work the most to get there?
Machine learning and Python. I had little experience in both.
- If you were to start your journey to become a data leader again, how would you tackle it?
I would not change a single thing.
Michael Curry
Early in my career, I took a product role in a data integration company. In this role, I saw firsthand how much of an impact a strategic focus on data has on businesses. The companies that excelled in integrating, curating, and analyzing data were the companies that performed the best in their market. They were better at serving customers, better at selling, better at building products, and better at managing their investments. From this experience, I realized that data would be a critical tool for every step of my career.
- Which skills did you have to develop to become a data leader?
Data cataloging and curating, data analytics, data visualization, and data governance.
- How can business decision-makers and analysts prepare themselves to use data science and analytics?
Business decision-makers need to define the problem spaces that will yield the biggest returns on data. Where can specific insights help the most to impact the business? Is it better to help your sales organization to better target new customers, or to streamline product development? Where are the data blind spots that can be removed to help improve decisions and operations? These are the types of questions that business decision-makers need to answer so that they can define the outcomes they are hoping to achieve.
Micaela Kulesz
I have a background in applied quantitative methods, with a Ph.D. in economics. It was a natural step to get into data science, and then into data engineering to advance toward innovation.
- Can you tell us why you became interested in working with data?
Data is factual. Loose interpretations are feasible, but not scalable. “What are the facts?” is the first question to ask ourselves. And only after comes what we read from the facts. This is a general attitude that drives prosperous societies.
There are words hidden in the data, and you can dig as much as time exists. Data is magic.
- Which skills did you have to develop to become a data leader?
Dedication and discipline are the keys to becoming an expert. To become a leader, however, I had to develop patience and focus. Data teams need a purpose, as does any team. Yet as data can acquire important volumes, it is easy to get lost in it: the leader needs to ensure focus at every point in time.
- Which are the areas in which you needed to work the most to get there?
I would say my “multi-purpose” programming skills. I could program very well in R, STATA, and bash (sh, zsh), but to grow within the data science field, a multi-purpose language helps enormously. It helps to be able to communicate with people from other teams and backgrounds, but mostly to let our ideas grow with the field.
Once the discussion opens the door to data – thus, facts – innovative ideas will come in, and little by little the change will start. This is the best preparation: discussing and listening.
- If you were to start your journey to become a data leader again, how would you tackle it?
I would participate more in meetups and hackathons.
Bob Wuisman
I was good at Excel and as a business analyst, I was involved in an implementation program for a new CRM system (Microsoft Dynamics AX). No one had thought about the reports that had to be built to keep track of the operations. I raised my hand, pointed out this missing part of the program, and got the project. A month later, I realized I did not know what I signed up for but loved it and started to Google what a SQL server is.
- Which are the areas in which you needed to work the most to get there?
Learning SQL Server, QlikView, data warehouse concepts, ETL, and data in general. Everything that was tech and data-related.
- If you were to start your journey to become a data leader again, how would you tackle it?
From the domain perspective – if your role is extracting business value from data.
- Can you tell us why you became interested in working with data?
The insights I got from data enabled me to get a clear picture of overly complex topics, such as end-to-end supply chains, from the first coffee drink to the invoice.
- Which skills did you have to develop to become a data leader?
Mostly technical skills and concepts, because I already had a strong organizational background in business processes, culture, architecture, and more. I learned SQL to extract data, transform it, and load it into the databases, as well as to query and combine data. Data modeling helped me learn how data can be joined and provided me with data visualization tools such as QlikView and PowerBI. The basics of Python, the main language these days, and understanding the full implications of each decision and code on each downstream step also helped.
- How can business decision-makers and analysts prepare themselves to use data science and analytics?
It is not a goal per se. First, figure out why you want to use data science and analytics and what it will bring you. Having data science is no shortcut to success.
Wim Van Der
- Can you tell us why you became interested in working with data?
As a governmental organization in the water domain, we face huge challenges when it comes to climate change and water quality, especially in a country such as The Netherlands, where we live one-third under sea level. We need data and technology to help us understand the complexity and to model future solutions.
- Which skills did you have to develop to become a data leader?
Basic technical knowledge. Change management. Ethics. The dos and don’ts.
- How can business decision-makers and analysts prepare themselves to use data science and analytics?
Start training people that know your business and make them aware of the possibilities that data and technology have. From there, start expanding with, for example, DevOps teams.
Florian Prem
It was the best way to solve business topics and get insights/find root causes in my professional work.
- How did you get into data science and engineering?
By applying hands-on cross-domain knowledge (for example, within finance doing analytics more than 20 years ago), with a background in law, IT and management on compliance/governance, and delivering a series of digital transformations, AI/ML, data management, and IT modernization programs.
- Which areas did you need to work in the most to get there?
I had to do constant work to stay up to date with technology and regulatory matters.
- Which skills did you have to develop to become a data leader?
Data governance, compliance, regulatory knowledge; data management and analytics/AI and ML/technology skills; program, project, and product management skills, and change management skills and soft skills.
- How can business decision-makers and analysts prepare themselves to use data science and analytics?
Become more tech-savvy and learn the importance of data/using data within their operations, decision-making, products, and processes.
Most of today’s data leaders transition from distinct roles into positions that require them not only to create systems that gather and display data but also to manage highly specialized data teams. In their answers, we can see how each one of them, in their journey to become leaders, was able to reinvent themselves and learn the skills required to thrive in these leadership positions.
In the next section, we will get their perspective on how data is being leveraged today in organizations and what their focus has been.
Using data in organizations
Each of the interviewed leaders currently holds a position where they need to either transform their organizations into data-driven ones or improve how they currently extract value from them.
Let us see their perspective on the current state of data in companies and organizations.
Florian Prem
End-to-end in their business and operating model with a customer, operations, and employee focus, but most still only in silos.
- How important is it for companies to become data-driven?
Today, very; tomorrow, it is a survival topic.
- What is the benefit that businesses get from storing large volumes of data?
There is no direct benefit from storing large volumes of data. Value comes only from insights and contexts that become assets.
- In your experience, are businesses aware and open to exploring new uses for their data?
Some are, and most are just starting – the leaders have built their business and operation models around it.
There is not one answer for this besides: do not start with the infrastructure/technology – the technology is defined by the business needs.
- Which business areas more often need to process big volumes of data?
It depends on your industry – operations and finance and customers/products.
- What is the frequency at which this data needs to be updated?
It depends on the data and the insights/use case you want to achieve – from real-time to daily/weekly/monthly/yearly or per case (for example, address change).
Micaela Kulesz
So far, this is only a current cost and a potential benefit. Unfortunately, data science resources are very scarce, and developments in the field are driven by non-structured data, which, in turn, attracts the few resources there.
- In your experience, are businesses aware and open to exploring new uses for their data?
Not enough to drive a consistent change.
- How do companies deal with the complexity of creating data teams and maintaining complex infrastructure?
This depends on the size and the focus of the company.
- What infrastructure is needed for companies?
It depends on the scale and the industry.
- Which business areas more often need to process big volumes of data?
They all need to; always.
- What is the frequency at which this data needs to be updated?
It depends on how they use it.
- Which areas in these companies can benefit the most from using data science and engineering?
Innovation and discovery. This area can work in parallel with the rest of the firm. A blue/green development focused on increasing the data-driven culture.
- How can data be used to drive revenue in the context of economic contraction?
To increase the margins of profitability. In the meantime, there is no need for much effort to maintain and even increase profits. In times of crisis, we need to work within the limits. Here, data is key.
- How can data be used to reduce costs in operations in the context of rising costs in the supply chain?
I would like to improve the performance of the current models. Yet to build new ones, I do not think volume is necessary.
- How are companies using data to improve their operations?
They are starting to try small products that improve user experience.
- How important is it for companies to become data-driven?
Crucial. I think this will define their place in the market in the next 5 years.
- How are companies using data to understand customers’ behavior patterns?
A/B tests are very widespread and are the main tool for approaching behavior. However, as I mentioned previously, when it comes to structural data, there is much room for improvement.
Wim Van Der
- How are companies using data to improve their operations?
In this phase, mostly to make processes more efficient, reduce energy use, get better output, create new software, and create models for climate scenarios.
- How important is it for companies to become data-driven?
It is not. It is important to stay human-driven, but humans need to learn how data can help.
Julio Rodriguez Martino
Quality is more important than quantity. A lot of bad data will not help solve any problem.
- In your experience, are businesses aware and open to exploring new uses for their data?
They are aware and open, but not always able to do it. Businesses often need help to get started.
- How do companies deal with the complexity of creating data teams and maintaining complex infrastructure?
I have seen diverse ways of dealing with data teams at various times. Many years ago, companies were hiring data scientists without knowing what their value was. Later, they realized it was easier to hire third-party teams to do specific projects. When it was clear that the knowledge that was created was not kept in the company, there was a shift toward a hybrid approach: small in-company teams helped by external professionals.
- What infrastructure do companies need?
Having access to cloud computing, I do not think there is a need for specific infrastructure in most cases.
Any business unit dealing with transactional data. Other areas usually manage smaller amounts of data.
- What is the frequency at which this data needs to be updated?
It depends on the type of analysis. In some cases, the frequency will be low; in other cases, almost real-time frequency might be needed.
Michael Curry
I have seen data used effectively in all aspects of operational management. In my career, I have used data to drive transformation, pinpointing areas that most needed improvement. I have used data to inform on areas of new opportunity in new markets or even within existing customer bases. I have also used data to predict where products were likely to see increased demand over time.
- How important is it for companies to become data-driven?
The more data-driven an organization is, the more successful they tend to be. This is a straight-line relationship that has been proven in study after study. Working on assumptions and intuition rather than data will usually lead to wasted time and money. Today, with the power of machine learning, we can get more use out of data than ever before. The companies that stay at the forefront of these investments are the ones that will outperform their peers.
Data is challenging to obtain and often has quality or timeliness issues. In addition, people often struggle to understand what they are looking at in the data. As such, it is often easier to base decisions on assumptions and intuition. The last conversation they had with a customer, for example, can oversteer many decisions. Creating a culture where the default is to turn to a deeper analysis of data in decision-making requires that there be a strong focus on making the necessary data available, easily understandable, and trustworthy. These require large investments that many companies are not willing to make.
- Digital companies have been born with data analytics and ML embedded at the core of their business. Are more traditional companies lagging in the adoption of these technologies in their day-to-day operations?
Certainly, some are. However, even traditional companies have been making large investments in data, analytics, and ML. Purely digital companies have an advantage in that often, their entire supply chains are digitized, making it easier to access the data needed for decision-making, but more traditional companies often have more directly relevant data, larger existing customer bases, and more history to work with, so effective investments in data analytics and AI can sometimes yield superior results. The issue is really whether traditional companies can overcome the cultural inertia to become data-driven.
- Which areas in these companies can benefit the most from the use of data science and engineering?
There are not any areas that cannot benefit from data science and engineering. The most progress, however, has happened on the marketing and sales side of businesses. Using analytics to understand customer behavior and better pinpoint offers to customers has become very commonplace. A lot of recent data science investment has focused more on the production side of businesses, and this could yield even bigger returns – helping pharmaceutical companies speed up drug discovery, helping agricultural companies to improve crop yield, and helping chemical companies to produce novel new compounds faster. Using data to reduce long, expensive investments and improve their yield has the potential to dramatically improve productivity across all industries.
Bob Wuisman
None. It only costs money. It is about the quality and value of the company.
- In your experience, are businesses aware and open to exploring new uses for their data?
Yes, but there is always the political complication when you make actual results transparent with data analysis and point out someone polished their results a bit too much. This happens everywhere – internally and externally.
- What infrastructure do companies need?
Elite and high-performance teams are all on the cloud or multiple cloud platforms; check out the Dora research done by Google.
- Which business areas often need to process big volumes of data?
All industries with data streaming components.
- What is the frequency at which this data needs to be updated?
No general rule can be applied to this. Sometimes, every nanosecond; sometimes, once a year.
Jack Godau
I am on a slightly different track than you. It starts with the vision, the organizational goals, and strong alignment on those. Once that exists, only then should the effort be made to determine the information required to make decisions and only after that to discover the data sources and pull the required information.
The analogy I would use is mining (not data but digging in the ground for stuff). We do not simply pull down forests and mine entire mountains to “see what we can find.” We set out to specifically extract the resources needed at the minimum cost and effort – this is something that “data teams” miss, so a huge effort is poured into “let’s keep it all,” “let’s run some analysis, and see what it tells us.” That is not useful or viable from a business point of view.
There is only a benefit if there is a purpose. Hoarding data – especially things that have no value – can be a costly mistake.
Data should not be confused with information.
You need to be truly clear about what goals you are pursuing organizationally and then determine what information you need; then, you can determine the data sources for such information. This is, of course, much harder than just “hiring a data team and letting them work it out.”
Now that we have a perspective on how teams work in implementing successful data-driven strategies in companies, in the next section, we will focus on why these companies will find it relevant to implement successful data strategies.
Benefits of being data-driven
Transforming organizations into data-driven ones is not an easy task. It requires a clear focus on the objectives we want to achieve, as this transformation requires a lot of resources and creates a lot of inconveniences when changing the current process to make use of data for informed decision-making.
Wim Van Der
It depends on the company and its identity; I do not think there is one solution that suits all.
- How can data be used to drive revenue in the context of economic contraction?
Economic revenue should no longer be the core drive for companies. Value in a wider perspective is. So, do not stare at huge piles of economic revenue; also look at ecological revenue.
- How can data be used to reduce costs in operations in the context of rising costs in the supply chain?
Start with a DevOps team to analyze your business processes and give them an open task to make it more efficient.
- How are companies using data to understand customers’ behavior patterns?
Process mining is still a successful tool to make customer journeys.
Michael Curry
- How can data be used to drive revenue in the context of economic contraction?
Even in slowing market conditions, money is being spent by companies, governments, and individuals. Data is the secret to uncovering where and why that money is being spent so that businesses can more effectively compete for it. Understanding how priorities are changing as economic conditions change, for example, is something that data can uncover. Data can also be used to help fine-tune pricing and packaging to optimize the changing needs of buyers.
- How can data be used to reduce costs in operations in the context of rising costs in the supply chain?
Optimizing supply chains has long been a focus of data analytics. Better predictions of demand and better allocation of resources can help minimize costs without sacrificing revenue.
- How are companies using data to understand customers’ behavior patterns?
Customer behavior analytics have been some of the largest investments that companies have made over the past decade. This has been accelerated by the natural data footprints that people leave behind in a digital setting. These footprints can give a much deeper understanding of the behaviors and motivations of individuals than could be achieved in purely analog settings. Therefore, digital tracking has become so ubiquitous (along with the backlash to it). Home insurance companies, for example, are interested in knowing when people might be house hunting in a new ZIP code so that they can target offers to them.
- More traditional companies are used to outsourcing their market research to consultancy companies. Is the use of social media, web analytics, and search engine data replacing these approaches?
To a considerable extent, this is becoming true. The tools that used to be available to only a few very skilled data analysts (and thus concentrated by consultancies) are now much more available to a much broader population of people. In addition, data analytics has become a much more common skill that a much larger percentage of businesspeople are learning, and the data that is used to enrich internal data is now more accessible to even casual business analysts. As a result, the reliance on expensive external consultants has been reduced.
Jack Godau
- In your experience, are businesses aware and open to exploring new uses for their data?
No.
Zero percent – we need data to make decisions, BUT the business must drive the vision and strategy. Purely working just from the data and “bubbling up” is not a viable strategy. Sure, Amazon uses its data to help sell more products – but its vision is to be the best online store. Data enables this but data by itself does not do it. The strategy and vision must be there first and must be well articulated.
Bob Wuisman
- How do companies deal with the complexity of creating data teams and maintaining complex infrastructure?
This highly depends on C-level data literacy and company culture. Not unimportant is the number of attempts that they made to transform into a data-driven organization. On the first try, there is a lot more freedom and delegated responsibility than on the next try.
Florian Prem
Customer, people, and products – across the board with every company being different and having different pain points.
- How can data be used to drive revenue in the context of economic contraction?
Be better than your competitors, know your customers, and apply cost savings.
- How can data be used to reduce costs in operations in the context of rising costs in the supply chain?
It depends on your industry and supply chain – not one answer is possible.
- How are companies using data to understand customers’ behavior patterns?
The best way IMHO is data-driven end-to-end customer journeys.
We now have a perspective of how these leaders would think that organizations can benefit from implementing data-driven strategies. In the next section, we will dive into the challenges that come with these transformations.
Challenges of data-driven strategies
There are no out-of-the-box strategies that we can implement to champion a data-driven transformation in a company, so each of these strategies needs to be tailor-made according to the needs of these organizations.
We asked the data leaders about the usual challenges they faced when transforming these organizations into fast-paced, data-driven companies.
Bob Wuisman
70 percent of the time, such an attempt fails. Culture and politics. Siloed departments that do not want to collaborate and a power-oriented culture. The lack of sufficient technology. Data quality is not safeguarded, and technology is centralized instead of organization performance.
A project is led by a CIO, CTO, or CFO. A CDO or a similar position should lead and report to the CEO or COO.
- What are the main technical challenges of creating a data-driven culture in business?
Become stable, reliable, and trustworthy. Deliver what you promise and point out the advantages of the needed investments.
Florian Prem
- How do companies deal with the complexity of creating data teams and maintaining complex infrastructure?
Not very well – they’re better off hiring and empowering a tech and business-savvy CDO and starting digital transformation programs with exec endorsement.
What is the best way for business decision-makers to collaborate with highly specialized data teams to serve the company’s needs?
Combine the data teams into a CDO/corporate data office and let the CDO and their team lead deal with it – different skills than what are normally used in data teams are required for that.
- What are the main challenges of creating a data-driven culture in business?
Politics, people, processes, and solutions, and lack of exec endorsement/driving.
- What are the main technical challenges of creating a data-driven culture in business?
People but also existing solutions/technical debt/silos.
- Digital companies have been born with data analytics and ML embedded at the core of their business. Are more traditional companies lagging in the adoption of these technologies in their day-to-day operations?
Yes, data must be incorporated into their operating and business model.
- More traditional companies are used to outsourcing their market research to consultancy companies. Is the use of social media, web analytics, and search engine data replacing these approaches?
No, but end-to-end analytic platforms might make it easier soon – you still need the competencies and tools to understand the data and market insights, and so on.
Wim Van Der
Start bottom up and create small successes. Make people tell their success stories. Do not start with the newest technology; people need to learn first and get in the right mindset.
- Digital companies have been born with data analytics and ML embedded at the core of their business. Are more traditional companies lagging in the adoption of these technologies in their day-to-day operations?
The question remains: what are your business values, and what is the identity of what you strive for? After that, start with technology. The ones that seem to be ahead now, might be behind soon – not on technological levels but ethical levels.
Jack Godau
The question is wrong – nobody in the real world wants a data-driven culture. Here is the great misalignment between data and business. Take healthcare, for example – data mining for the right information can be a huge benefit to the treatment of patients. But in the real world, everyone understands that it should be the focus of the organization to treat the patient, not to have the data. Making information from data easily accessible and visible to enable treatment is a good thing, but it is not a data-driven culture.
Micaela Kulesz
Large companies need processes, and processes take time to incorporate at scale. Small firms are more flexible, but also lack the budget to invest in innovative technologies. Quite a challenge! But, in general, the answer is “people.”
- What are the main challenges of creating a data-driven culture in business?
Employees fear losing jobs to automation, human rights are being challenged by algorithmic fairness, and companies fear losing rights over the use of their data. There is a lot of fear. Exposing the company to more data-driven action will help to overcome this fear.
- Digital companies have been born with data analytics and ML embedded at the core of their business. Are more traditional companies lagging in the adoption of these technologies in their day-to-day operations?
Larger non-digital companies are the ones most lagging. On the one hand, their scale makes changes intrinsically complex. On the other hand, until now, the data had a specific usage and destiny, whose sense and purpose are starting to be challenged.
- More traditional companies are used to outsourcing their market research to consultancy companies. Is the use of social media, web analytics, and search engine data replacing these approaches?
I do not think so: it increases their competitive landscape and forces them to innovate.
Julio Rodriguez Martino
Having access to cloud computing. I do not think there is a need for specific infrastructure in most cases.
- What are the main technical challenges of creating a data-driven culture in business?
I do not think the challenge is technical. The most important challenge is to change the way the whole organization thinks about the data. It is extremely important to show everyone the value of the data so that they know how important it is to generate quality data and keep it safe.
Having a vision of the challenges that we will face when championing data initiatives helps us plan a strategy that will foster the adoption and continuity of our data strategy.
In the next section, our questions will focus on how we can create data teams that support the transformation of our organizations.
Creating effective data teams
Data initiatives require the implementation of data teams with specialized skills, as well as a multidisciplinary approach as the technology that supports the data strategy sits on top of the commercial strategy.
We asked the data leaders their perspective on data teams; here are their answers.
Florian Prem
- What does a good data team look like?
It depends on the mandate of the CDO. I normally organize into governance, data analytics, and data value creation/product. Then, I build cross-domain teams for agile product delivery (IT, business, and data).
- What are the most important topics to consider when leading a data team?
Understand your team, understand your business, and understand your technology.
- What are the best practices for business decision-makers to collaborate with highly specialized data teams?
See what I stated previously, hire a tech and business-savvy CDO, and build enterprise-wide/multiple linked data offices.
Michael Curry
- What are the best practices for business decision-makers to collaborate with highly specialized data teams?
The most important best practice is to establish a data organization. The goal of this organization should be to define and execute the curation process for data across the business, the alignment of data to business needs, and the governance of how data is being used.
Micaela Kulesz
- What does a good data team look like?
Three or four members: a lead data scientist, a junior/semi-senior data scientist, a data engineer from an IT background, and a product manager if possible. This team composition is very dynamic and can tackle many – if not most – business problems. I find it important to emphasize that at least one of the members must come from an IT background and another one from a business one.
- What are the most important topics to consider when leading a data team?
- Ensure dynamism within the team.
- Understanding is a privilege. Research about the latest developments is ca. 15% of the time of the team, and it is mostly carried out by juniors/semi-seniors.
- Seniors are responsible for providing a clear context for the projects, and they must be able to code.
- Spend the necessary time to discuss and make the problem to tackle clear, and don’t depart from it.
- What is the best way for business decision-makers to collaborate with highly specialized data teams to serve the company’s needs?
Business decision-makers have a noticeably clear focus and optimization functions. Here, the data team can provide data services to the business and help them to increase their focus, or not. If they do, they become an operative service of the firm with no life on its own. However, as data teams are new and thus finding their identity within the firm as well, placing themselves as “data operatives” or “data providers” is not what they want. If they accept themselves as “data servers,” I think it optimal that the business takes the time to explain the problems and the pains until they are truly clear, and the data team must bring specific solutions for these pain points. No more, no less.
Bob Wuisman
- What does a good data team look like?
A good data team is a team that has a clear view of the company strategy and knows how to contribute to that in a fast way. Everything they do has a clear business objective, even if it is at the infrastructure level. Having adopted a DevOps methodology is key.
- What are the most important topics to consider when leading a data team?
Have a transparent organized backlog that is supported, appreciated, and recognized by senior managers and executives. Make sure engineers have no constraints in doing their work. Ideally, they can work without any external support. Results need to be celebrated and “errors” must be taken as learning opportunities.
- What are the best practices for business decision-makers to collaborate with highly specialized data teams?
Adopt DevOps as well as have a clear strategy, objectives, KPIs, processes, and IT to capture the right data. Become familiar with data concepts and dive into business administration. Do not oversell or over-ask.
Jack Godau
- What is the best way for business decision-makers to collaborate with highly specialized data teams to serve the company’s needs?
Having clarity of vision and needs, transporting this to the teams, and making sure they understand the organizational goals. Regular feedback loops, working with the teams to understand their outcomes, and validating those against real-world cases. After that, all the standard stuff – classic enablement of the teams, no micromanagement, providing good tools, and more should be undertaken.
Julio Rodriguez Martino
- What does a good data team look like?
A good mix of different profiles, coming from diverse backgrounds. They should be eager to learn and teach by helping each other.
- What are the most important topics to consider when leading a data team?
Leading by example. The leader must understand firsthand the challenges each member faces: learning new topics and working with colleagues with different backgrounds and experiences. The leader should also be aware of the details of each project, at least up to their knowledge of the subject.
- What is the best way for business decision-makers to collaborate with highly specialized data teams to serve the company’s needs?
To help identify business problems that can be solved with data and will return value. In addition, to be ready to work together with the data team to understand and explain the results of the analysis.
Wim van der
- What are the best practices for business decision-makers to collaborate with highly specialized data teams?
The extended reality, blockchain, and digital twins, but this depends on the business. However, DevOps teams are known for success when positioned well.
Creating and supporting efficient data teams to develop a successful data strategy in an organization is of crucial importance and requires more than simply hard skills.
In the next section, we will ask the data leaders their perspectives on what the future looks like for these organizations in terms of data.
Visualizing the future of data
Data has become a central pillar for companies that want to uncover the possibilities of value for the business, which is why modern technologies will shape the future of organizations.
We asked the data leaders what their perspectives are on the future in terms of technologies.
Michael Curry
Data integration, data quality, and data governance continue to be expensive and overly manual, and yet they remain the biggest barriers to companies being able to get full leverage from their data. I expect to see dramatic improvements in these areas, using new AI-based approaches to reduce manual effort.
Florian Prem
Explainable AI, device-to-device comms/IoT -> predictive analytics, automation, end-to-end analytics platforms, and new NLP and image/video processing models – once these technologies are available as products and platforms, they will be more broadly used.
Bob Wuisman
The next big step will be quantum computing. Until that time, there are more and faster cloud platforms that have increased niche services.
Micaela Kulesz
In data science, these come in line with increasing the understanding and modeling of big structural data.
In engineering, this involves improving the algorithms; we already know how to perform fast and easy tasks.
Julio Rodriguez Martino
I guess that no-code and AutoML technologies will become increasingly popular. Only sophisticated analyses will require coding in the future.
Shaping the future of data in organizations will require being nimble to adapt to new challenges and technologies. Hopefully, the perspective of our data leaders has provided you with an idea of which technologies will shape the future of data.
The next section will dive specifically into the experience of data leaders who have specialized in championing digital and data transformations in several organizations.
Implementing a data-driven culture
Data has gained center stage in the table of discussion for company leaders and managers. The opportunity to improve operational performance and engage in an optimized manner with the customers to find new sources of revenue are promising goals to achieve by implementing a data strategy, but these transformations do not come without hassle.
For this section, we have interviewed Patrick Flink, a data leader who has a company that focuses on transforming organizations into digital organizations, and Agustina Hernandez, who has built global data teams from scratch in some of the biggest companies in the world.
Agustina Hernandez
With a background in econometry, Agustina had previous experience as an econometric modeler and decided to follow her passion early on in her career, even if it meant having a reduction in pay. Her other passions are coaching and photography. She worked on several projects, creating insights for top FMCG companies, and leading the teams that would create insights on topics such as brand equity, media ROI, assortment, and more. One of her main focuses has been to understand the client’s needs in depth to achieve the objectives of business revenue and EBITA.
After working for several client accounts, she joined a top cosmetic company as head of Global Data Analytics and was tasked with building the data team for this company, to deliver data-driven solutions and create a cultural change in the organization to be more focused on data.
Her leadership strategy can be regarded as setting in the table of discussion all the different stakeholders of a data team, regardless of their domain and backgrounds. She fosters communication within the team without imposing a specific view, rather than working to get to know the client’s requirements to focus on the task to deliver, making sure that the team has all the resources to continue while focusing on improving this communication.
The vision she has of a data team is that it is a team in charge of changing realities and that companies can improve their operations by understanding that creating a data-driven culture is necessary to improve skills and have a top-down approach. Most of the time, this requires a team that gets to be attached to the current organization and pushes these changes through a lot of work, and focuses on changing behaviors. For Agustina, the eruption of data in companies can be seen as a technology that will drive any kind of change, magically solving and improving all metrics. The data strategy sits on top of the current strategy to be able to increase the reach of the knowledge that we can use to make better decisions.
- What are the approaches to effectively implementing a data-driven transformation in an organization?
There are several types of approaches to achieving an effective transformation. The creation of lighthouse groups, which are aspirational teams, examples of the transformation that they want to achieve, to guide the transformation is not enough because later, these teams collide with a reality different from the conditions in which they work. This is because these teams have more resources than the others, and it differs from the real situation in which the rest of the company finds itself.
The Lighthouse model is opposed to the one set up by a parallel organization, which is in charge of transforming the company’s processes. It strongly implements them by establishing transformation objectives and clear incentives such as variable bonuses in the middle. This generates internal conflicts but is maintained for the time that a new culture lasts. This also implies that resources are required and that this transformation hurts as if it were organizational adolescence.
Some organizations have large margins that allow them to duplicate resources by creating transformation teams for each of the departments, which then end up joining those teams and departments that are seeking to improve. The problem with this is that it duplicates resources but not all the equipment in those resources to carry out this transformation.
- What is the perspective of management on these transformations?
Middle management is the hardest to convince because by adopting a new strategy, they end up doing two things at once. While there are middle managers excited about change, everything is defined with hands and resources, and carrying out the transformation requires diverting change processes and diverting resources from operations, which affects the performance metrics by which these middle managers are judged. Implementing an effective data strategy collides with reality when those who should be using the data do not change their processes to incorporate these metrics or insights into their decision-making and daily operations. These individuals are middle managers who are closer to operations and do not have the time or the skills to lead the transformation or drive the use of data.
- What benefits can these organizations expect from becoming data-driven?
The data is nothing if the processes are not changed; if you do not think about equipment and processes, the data that’s collected will not have a use or an impact. Data-driven does not have fewer resources. This is a misconception that arises from the spirit of IT, which always seeks to improve its operating costs. The data-driven strategy must focus on what generates revenue in the company. The CDO, who depends on IT or operations, aims to save costs and that does not work. Data is neither less cost nor less time – it involves finding value opportunities. An example is that of promotions or price reductions, which, if applied without sufficient data on customers and products, generate that subsidy for those who were already about to buy. If we used data, we could calculate promotion elasticities and customize them to give each of the buyer segments what they want most by making them buy more than they would have but not generating trade spend savings. Once the organization matures and understands the true role of the data strategy, that data strategy becomes the responsibility of either the sales department, the strategy department, or the general manager. Alternatively, you will need to do a C-level.
This data strategy must have storytelling that supports this strategy and communicates what each of the organization’s participants’ roles is. This strategy must be implemented from the top down because if, on the contrary, it arises from each team or department, atomization is generated, which is why an alignment is necessary that effectively shows and communicates the benefits obtained step by step.
- What are the ingredients for a successful data-driven strategy?
The ingredients of the data strategy are clear purposes that are tangible objectives, a culture that aligns everyone behind those objectives, and the resources and perseverance necessary to carry out this transformation within the organization.
- How do people fit in the data-driven strategy?
In the case of people, it is important to think about the skills and abilities of each of them. Some people are good at analysis, some people are good at bringing strategies to reality, and there are a few who do both things well. Typically, data teams have strategy development capabilities and domain knowledge to carry out the data strategy. This is because hard knowledge is not enough and data is complementary to a reality that needs to be known, so there is a need to have field knowledge to complement the use of data to make effective decisions. This field knowledge lives in these middle ranks close to operations, who must learn from the data scientists doing the parts of the team with a CDO that coordinates these developments throughout all the teams. This does not work if there is a single centralized department that provides data to the entire organization.
- What is the role of technology in this transformation?
In the case of technology, this is an enabler that allows the strategy to be carried out, but this technology is built on the data strategy, which, in turn, is built on a commercial strategy, not the other way around. Solutions of minimum complexity should always be sought to solve a data problem that solves a business problem. This is carried out with a team of data engineers, MLOps, DataOps, and others, always controlling the quality of the development to ensure the reliability of information and the rules and abilities of systems.
Patrick Flink
Patrick is an expert in transformative strategy implementation. He places enormous focus on extracting actionable value from data with a minimal approach that can deliver value fast. His company, ONBRDNG, is a Dutch company that specializes in digital transformation, with an accent not only on strategy for being data-driven but also on execution.
With a background in business and management, he had a multidisciplinary approach to learning the tools and skills necessary to extract value from data, prioritizing actionable outcomes. He has been a CTO and CDO and has worked with media companies in the Netherlands to develop innovative products and solutions.
Patrick is both pragmatic and creative. He is a results-driven strategist who enjoys communicating with every level of an organization and rebuilding companies from traditional businesses into all-around digital companies.
- How did you prepare yourself to have the skills to be a data leader?
Yes, well, I am a lifelong learner. So, of course, I did some stuff myself but also spent a lot of time with our developers and analysts. I spend a lot of time just sitting next to them. So, it is always good to be multidisciplinary; then, you can see the value of working with other professionals from different domains. Often, this involves asking a lot of questions, looking from different perspectives, and so on. So, yes, that is where I started. We also did a lot of stuff surrounding media, the benefit of which is that media companies are always focused on how we can engage our customers and our consumers increasingly. The latest features, functionality, and products are also presented here because everybody knows that they are incredibly open to new stuff. They generate a lot of data in terms of behavior, what they like, what they do not like, and so on. You can use this information to be able to produce a proper strategy.
- Which kind of methodologies do you use to uncover value in the data?
We have an approach that is moving toward a methodology. But in my experience, most of the time, everybody has the same examples. And they do not take everything important to be successful in that area into account and forget to dive into the whole story. And then it is not only about collecting all kinds of data and then getting better because of it. It is just to see and understand how they do it – more than what they do because you know that can differ. And what we like about our approach is that we make it small, and quite simple. We use data to tell us what is happening, and then we see what we can do, and everybody says data-driven or data decision-making or whatever. But it is much more on the human aspect and creativity. And if you also look at the most, you know, data-driven companies, of course, they use data, but it is also still about creativity, intuition, and betting while being very honest. I have seen too many companies that hire a lot of data engineers, data and data scientists, and more while spending a lot of money on infrastructure. We use data to see if our bet is right and if we are moving the needle. We use a lot of it, and without customers, we work a lot on data sprints, also to make it focused.
We follow the journey of a customer, from beginning to end. All of this is possible, yet everybody’s busy with so much stuff, so we try to make it simple. That’s why we do a lot of stuff in data sprints for a couple of weeks – nobody tells you that a lot of times, to use data and get value from data, you don’t need a lot of people.
- What kind of challenges have you faced when working with data teams in a company?
One of the challenges is that companies that hire too many people. I experienced this myself; I asked a data manager who is responsible for customer insights to say we need a churn prediction model, and then a couple of weeks later, or months, we ask how it is going. They say that it’s not clear what the business requirements are, but that we can do facial recognition. However, that is not what we asked for. We can build a recommendation engine, but we already have a recommendation, so this is not where we need it now. So, it is not our biggest issue to have more people. There are a lot of companies with so many people working on a topic that it’s not beneficial or actionable, and we do not get real value from it either.
- What is your approach to creating successful data projects?
What we are doing is trying to make it exceedingly small. So, start with a number. Say, for example, we are going to work data-driven. Okay; what is the answer? What is our first assumption? Let us see what we are, what we are discovering, and what we are experiencing – this 90 percent is data cleaning. We are structuring data and getting the right labels, the right data, and the right definitions. But nobody talks about it. So, we might try to make it simple, and then really get real about it. This concept of data sprints is a view of the landscape of where we work on it. Is it sometimes very fragmented; for example, or it is a monolithic system or a landscape? Then, we ask, what is the business? What are we promising to the customer? How are we earning our money? Then, we try to get a view of the end-to-end journey of a customer. We start from here. What are our main challenges? What is the key question? Is it growth? Is it churn? Is it process automation? This is also related to the decision. Do we want to automate all kinds of stuff? Then, we must start with the hypothesis and work from there. We must always try to get some value out of it in 6 weeks – 6 weeks only. We evaluate the infrastructure and the business model, and we look at the channels in the customer’s journey. Then, we must set ourselves a goal based on those 6 weeks, with a small team. This team should be multidisciplinary and part of the company. It is your team. Well, of course, we do it for the customers. And it is always, of course, a hybrid multi-disciplinary team.
The focus of the team depends, of course, on the top management. If it is concerned with digital process automation, the objectives will be of a certain kind. If it is a factory, for example, or an e-commerce company, it will focus on conversion. It is, of course, a little bit of a different question when it comes to experience or where the focus is, depending on the company. So, that sometimes changes a little bit, but we always assist and work with the customer. We use main and we have our data platform also, so it is our customers are difficult with five legacy systems. Then, we have our data platform, where we do the landscape analysis on the data in certain systems. Then, we can say, “give us the elements from this system so that we can create some sort of initial view.” We use a lot of the global stack as well as some tooling, such as scripting. There are also experts on that platform. However, our data platform is used to ensure that we can accelerate.
- What should be the goals of a data-driven transformation?
So, on average, I think that if you talk about data-driven, you should talk much more to your customers. Then, you should talk about what you see in the data, as well as what everyone else sees. After, you should talk about your revenue, costs, and the results you wish to obtain. In the end, a lot of companies are only talking about those components. If you are data-driven, then you can already predict what your result will be because you already see it in the behavior of customers. You will see what the outcome will be somewhere further along the line than the next quarter. I think that, in the end, you will always end up committing to delivering on your promise, which is important to your customers.
Transforming companies into data-driven organizations requires a profound cultural change that needs to be implemented with a top-down approach, championed by leaders who can create a narrative to support the data strategy and adapt to the needs and objectives of the organization. This transformation is not just about technology, but mostly about people and processes. I am thankful for the perspective of both Agustina and Patrick and hope that their experiences can help you create effective cultural transformations.
Summary
In this chapter, we explored the distinct parts that constitute a data transformation for companies that want to become data-driven by leveraging the perspective and experience of data leaders with extensive knowledge and experience in leading digital and data transformations.
These approaches allow not only companies to transform but also leaders and managers to adapt to the new skills and processes required to implement a successful data strategy. By doing so, they have a clear goal of how the company will benefit from these efforts, how to create and lead data teams, as well as how to consider the implications of enabling digital and data transformation.
Overall, the objective of this book has been to help business leaders, managers, and enthusiasts implement ready-to-use recipes that seek to undercover insights about consumers, their behavior patterns, how they perceive prices, how they effectively make recommendations, and, in general, how to complement the commercial strategy with managerial economics applied in Python.
I have leveraged the knowledge of coders, managers, and specialists from whom I got the inspiration to author this book. All the code and data that’s been modified and used in this book belong to their respective owners and producers, to whom I am grateful. Most of all, I want to thank the data leaders, who, amid their incredibly busy day-to-day lives, have taken the time to answer the questions presented in this chapter.
I am thankful for the time you took to read this book, and hopefully, it has inspired you to use data and achieve your objectives while having fun and enjoying the process.