Connecting data, analytics, and AI

How to integrate your tech strategies and maximize business value

by Dr. Matthias Schlemmer, Andreas Odenkirchen, Agnes Heringer, Dr. Maik Hesse, and Justus Rödel

Data-driven decision making isn’t optional anymore; it‘s essential in today’s fast-evolving world. You already know this, and perhaps you’re already focused on transforming your company into a data-driven enterprise. But what does that really mean, and why do many companies fail to follow through on their data, analytics or AI strategy and achieve their goals?

Working closely with our clients, we‘ve identified one of the most common root cause for companies‘ failure to deliver, and it’s this: missing strategic alignment along the entire data value chain. In other words, to succeed, companies must connect the dots from strategy all the way down to execution, and across the whole data value chain, which refers to the range of activities from data collection and management via appropriate governance mechanisms, to the actual extraction of value from data via analytics and AI use cases and monetization. Let’s take a look at three examples to demonstrate why connecting the dots along the data value chain is imperative to becoming a truly data-driven enterprise yielding full business benefits.

Typical pitfalls in data-driven transformations

1. Industrial products company didn’t think through enablement challenges at outset

Recently, an industrial products company decided to embark on a data-driven transformation project. It started with the digital leadership team initiating the development of a cross-divisional analytics strategy. This strategy was built on prioritized analytics use cases aimed at efficiency improvements in internal processes and accelerated revenue growth from introducing data-driven products and services. It laid out the required skills and technology stack to enable the divisions to rapidly develop and scale analytics solutions. While the availability and quality of the company's data assets have rightfully been considered as crucial to execute the strategy, targeted data management and governance capabilities were excluded from the strategy itself. Instead, they were treated as “standard IT tasks” and ended in a general backlog over the course of several months. Most importantly, the consequences of this decision only became clear during the implementation phase, creating significant challenges, such as:

• Time and efficiency losses: Missing data ownership concepts caused enormous lead times in obtaining required data assets, thus delaying the development of analytics solutions.
• Poor data quality: With missing information and complicated analytical modeling, the completion of analytics solutions was often put at risk.
• Inability to scale: The lack of legal frameworks for the use of analytical models trained on customers’ data prevented scaling of analytics solutions across the company's client base.

What’s the take away from this example? Developing an analytics strategy without ensuring its enablement with proper data management capabilities is like trying to build a house on a weak, unstable foundation.

2. Chemicals company didn’t consider a coherent business case for its transformation

In another case, the IT leadership team at a chemical company wanted to strengthen organization-wide data management capabilities. Their strategy primarily focused on establishing data ownership, improving the quality of data assets, and setting up a modern enterprise data platform. It looked at required investments from the perspective of the IT department, such as hiring new data architects and engineers, upskilling existing talent, redesigning existing processes, and setting up a new technology stack. But the initiative came to a sudden halt when the executive board declined any further investment proposals as the IT leadership could not justify the business’ “return on investment” requirements without a clear data use case and monetization perspective.

What’s the take away from this example? While the importance of a data strategy was generally acknowledged—whether to deliver on a broader business strategy, for risk and compliance adherence, or as a part of technology foundation building—a coherent business rationale and case for the investments weren’t made. Hence, the company’s other divisions were not convinced that the data strategy would serve their specific needs and withheld support.

3. An auto company launched two strategies that were at odds with each other

Finally, the executive board at an automotive company initiated the development of two strategies related to the company’s overarching digital agenda: a data strategy driven by the IT department and an AI strategy driven by the business divisions. The data strategy focused on improving access, quality, and integration of data assets. It introduced a data catalog spanning across the corporate data warehouse and data lake, and a data marketplace for assets in selected data domains. The AI strategy focused on introducing new data-driven products and services. It highlighted required investments in talent and technology to enable product development of market-ready AI solutions. Although both strategies aimed to transform the company into a data-driven enterprise, they were doomed to fail during implementation because they weren’t integrated. In particular, the data lake and marketplace contained data assets for domains that were almost irrelevant for the prioritized AI solutions. Hence, product development teams faced significant challenges in collecting, processing, and integrating the required data for their solutions. The teams were forced to build their own data pipelines from scratch, while working on a technology stack independent from the corporate data warehouse and data lake.

What’s the take away from this example? At first sight, this company managed to overcome the hurdles of the previous two examples by both looking at data-driven transformation from a data management and an analytics perspective. However, the two strategies set in place were not synchronized, that is, the teams developing them did not strategically align and were partly even competing with each other. Continuous alignment of a company’s data efforts is vital to business success because data management, governance, analytics, and AI are deeply intertwined and need to be looked at coherently.

How to get it right

In order to trigger a successful data-driven transformation of your company, we recommend aligning initiatives strategically along the entire data value chain. Thus, your company should develop and implement one holistic organization-wide strategy covering the entire, end-to-end data value chain. Or, alternatively, you should make the alignment of data-driven initiatives in different organizational units a top priority.

When thinking about your company’s approach to become a data-driven enterprise, have you experienced similar issues, challenges, or roadblocks? Are your data, analytics, and AI strategies sufficiently aligned?

We’d love to hear from you! Please share your own perspective in the comments section, or feel free to contact us directly at andreas.odenkirchen@pwc.com.