What is a data and AI strategy and why do you need one

The data and AI strategy must take into account significant changes in your organization’s business environment that will affect its ability to create value from data and AI. Such trends may arise in multiple different domains:

• Changes in the technological environment, such as the rise of cloud computing platforms, or the application of generative AI to business context

• Changes in the economic environment, such as the emergence of new platform business models, or the availability of funding for AI initiatives.

• Changes in the social and cultural environment, such as attitudes toward sharing personal data, or the expectation for supply chain transparency

• Changes in the regulatory environment, such as the requirements for privacy protection or the explainability of AI solutions

These changes may open up new opportunities for value creation, but can also make existing solutions obsolete.

The data and AI strategy must consider your organization’s role in the wider business ecosystem. Of particular interest are the flows of money, services, and data between the players. More often than not, some of the key data assets needed by your organization are owned by some other player in the ecosystem, and you must try to find a way to access them.

We frequently see new data and AI ideas stemming from the questions “What data do we have?” and “Which exciting new algorithms could we apply?”.

This kind of ideation may sometimes uncover significant business opportunities, but more often the outcome is fancy one-off gimmicks that sit outside the core business processes and fail to generate long-term value. This is not a solid foundation for a data and AI strategy. Instead, we recommend starting from the important strategic challenges that your organization has identified and finding ways in which data and AI can help to solve them. This way, the data and AI strategy can build a bridge from the overall strategy into practical data and AI development.

The data and AI vision tell a compelling and inspiring story about what your organization is trying to achieve with data and why. This kind of storytelling is essential for facilitating organizational change by inducing alignment, engagement, and commitment to the shared vision. To enhance the effect, we recommend inviting the key stakeholders to participate in shaping and formulating the data and AI vision. Furthermore, insightful visualizations help to crystallize the data and AI vision and communicate it more effectively throughout the organization.

A common pitfall is ending up with vague and watered-down data and AI vision that tries to encompass every possible opportunity and please every business function. Instead, we advocate making tough decisions about prioritizing the most important focus areas, to ensure that they receive consistent attention and investment in the future.

Opportunity mapping can be the most exciting part of a data and AI strategy since it describes the concrete ways in which data and AI will create value for the business and customers (we often call these business use cases). We have identified four major areas of data-driven value creation, where data and AI opportunities can be found:

• Improving the efficiency of core business processes

• Developing new customer-facing data-driven services and experiences

• Selling and exchanging data and insights

• Using open data to catalyze a business ecosystem

When describing data and AI opportunities, the trick is to find an optimal level of concreteness. A good approach is to start by generating a variety of high-level idea headlines, then proceed to iteratively select the best ones and flesh them out in a bit more detail, using a predefined template. You can prioritize your data and AI opportunities using these four lenses (adapted from IDEO):

• Desirability: Does the proposed solution generate value for the customer or the end user? What evidence do we have for it?

• Feasibility: Can the proposed solution be implemented with reasonable cost and effort? Does it involve significant technological uncertainties?

• Viability: Will the proposed solution be profitable? What are the most important drivers of profitability, and how sensitive is the business case to them?
• Sustainability: Is the proposed solution compliant with data processing regulations? Is it in line with the policies and values of your organization (for example, avoiding discrimination)? What kind of impact does it have on society and the environment?

We recommend using service design methodology for discovering and refining the data and AI opportunities. In particular, iterative hypothesis-based co-design together with the customers or end users can help ensure that effort is not wasted on building data and AI solutions no one will use.

In order to turn your data and AI vision and objectives into reality, you will need various kinds of data and AI capabilities. Organizations often focus on technological solutions, but we advocate taking a holistic view that fully incorporates the human perspective. To this end, we have formulated a framework we call The Data-Driven Business Puzzle.

For example, from the technology perspective, you will need to consider which solutions you should migrate to the cloud, or perhaps even redesign cloud-natively, to reap the full benefits of cloud computing. At the same time, you will need to carefully choose the algorithms that best fit your business needs, and integrate them seamlessly into your business processes. But since your algorithms are only as good as the data you feed them, you will also need to identify and acquire the most valuable data assets for your business and nurture them with data management and governance processes.

From the human perspective, your organization will probably need to gain access to new data and AI skills, either through training, recruitment, or outsourcing. New ways-of-working are necessary to leverage familiar DevOps practices, such as version control, continuous integration, and automated testing, in the arena of data and AI (this is sometimes dubbed DataOps and MLOps).

The art and science of fostering a data-driven business culture may seem enigmatic to people coming from a technical background, but there are tried-and-tested methods for alleviating uncertainty, building confidence, inspiring engagement, and motivating change. Furthermore, the positioning of the data and analytics practitioners within the organizational structure needs to strike a balance between ensuring strong data domain knowledge and avoiding counterproductive silos. Finally, the leadership team should play a crucial role by painting a clear vision, ensuring sufficient resources, and setting an example of data-driven decision-making.

Once you have an idea of your data vision and opportunities, we recommend conducting a holistic data and AI capability maturity assessment across all pieces of the data-driven business puzzle to identify the most important bottlenecks in your data and AI capabilities.

Now that your organization has a good idea of where it’s going with data and AI, you need to translate this vision into clear objectives, and define unambiguous metrics that can be used to measure progress.

We recommend developing a balanced KPI framework that covers the key focus areas for data and AI opportunity and capability development. One tried-and-tested approach for this purpose is the Objectives and Key Results (OKR) method, made famous by Andrew Grove at Intel and John Doerr at Google.

The end goal of the organization’s data and AI initiatives is typically to make an impact on financial business metrics such as revenue and profit. However, the feedback cycle between practical data and AI development activities and financial outcomes is often too slow to enable quick feedback loops for iterative learning. Therefore, we recommend complementing those lagging indicators with leading indicators that provide near-instant feedback on development efforts, allowing your organization to continuously hone its plans as you learn what works and what doesn’t.

The final element of the data and AI strategy is the roadmap that outlines the sequence and timeframe for achieving the data and AI objectives. The data and AI roadmap should cover both the most important data opportunities (business use cases) that will deliver tangible business impact, as well as the data and AI capabilities (enablers) that your organization needs to improve. Again, remember to take a holistic view to capability development, looking beyond mere technology. Both opportunities and capabilities should be clearly linked to the overall data and AI vision and objectives that your organization is striving to achieve.