How to improve the data foundation for AI at a strategic level?

Imagine a robust data foundation as having a well-build road. It is crucial for leveraging AI and driving innovation—and here's why.

There is no AI without data. Different AI applications serve different purposes, but they all require a robust data foundation. A key business objective is to create value through the effective use of data and AI. By implementing business-driven use cases, organizations can effectively leverage their data and AI capabilities. This includes everything from the data itself, processes and architectures, to tools, models, leadership and other essential skills. These elements, combined with effective ecosystem management and strategic partnerships, form cornerstones for the success of data and AI initiatives.

Driving a sports car on a rocky road illustrates the challenges when AI and data strategies meet reality.”

A robust data foundation enables organizations to fully maximize their data assets and support strategic initiatives such as AI, machine learning and advanced analytics. Building an effective data foundation that is AI-ready involves several critical elements:

1. Contextual information: Enhance data usability and governance through effective metadata management, providing context that makes data more meaningful and actionable.
2. Data quality measures: Ensure that data is fit for purpose by implementing rigorous processes such as data cleansing, monitoring and validation. These steps help maintain high standards of accuracy, completeness and consistency.
3. Timely access to information: Facilitate data-driven decisions by ensuring data security, access, and sharing. This involves implementing comprehensive data architecture, seamless integrations and strong data governance practices. These measures ensure consistency across various platforms and make the right information available at the right time.

Effective data governance and quality measures act as regular car maintenance, ensuring the system remains reliable and performs at its best over the long journey ahead.

StrategyOps - From paper tigers to action heroes

StrategyOps focuses on ensuring that strategic plans are not only formulated but also effectively implemented, monitored, and adjusted as needed to achieve the desired outcomes. StrategyOps includes various data and AI initiatives, user interfaces (e.g. chatbots, robo-advisers, and credit decisions), end-to-end life cycle management, architectural decisions, and improvements on the roadmap.

In ERP implementations, there is often insufficient interest in data and its organizational value."

Organizations on all levels require training and advisory to become data literate. A lack of data background can easily lead to failures. It is like driving a car without paying attention to the quality of the fuel, resulting in poor performance and missed opportunities.

Skilled personnel should calculate and showcase Return on Investment (ROI) for initiatives like AI projects. While ROI, cost-benefit and net present value calculations provide valuable insight into the financial potential of a data project, it is crucial to ensure that these projects are aligned with the organization’s overall strategic objectives. Additionally, the potential impact on non-financial factors such as customer satisfaction, operational efficiency, and regulatory compliance (e.g. fines) should also be taken into account.

Data ownership may not be an issue at the top level, but it can become problematic at the next level. It requires middle management to make decisions for the common good. Concrete examples include investments in data governance, like interoperable data catalogues or data quality monitoring.

Top management should also consider basic scenarios for an enterprise data strategy:

1. Growth engine: Treat data and AI as a growth engine rather than just operational fuel.
2. Data collaboration: Leverage data with partners and ecosystems to drive growth, rather than just using it for internal operations.
   
   

AI and how to set the course for meaningful innovation

From a business perspective, AI models should be trustworthy, robust, legal, and ethical. When selecting AI technology, align it with your strategic goals whether it is improving efficiency, driving innovation, or transforming workflows. For example, GenAI is ideal for content creation and image generation based on patterns and probabilities, while deep learning models are more effective for segmentation tasks. In addition, remember that different user groups - such as data scientists, business analysts, marketers, customer service teams and developers - may require different AI technologies tailored to their specific needs. Choosing the right AI technology is like choosing the best tires for your car to get the best performance on different terrains.

From an innovation perspective, key success factors include mastering data, increasing AI literacy, creating a clear strategic vision, using agile tools and infrastructure, prioritizing customer-centric approaches, demonstrating strong leadership, and complying with ethical standards and regulations. Documentation becomes crucial as the speed of innovation skyrockets in cloud environments.

Documentation is like a GPS, keeping your car on the right track as the pace of innovation accelerates."

Data availability ensures you have sufficient and relevant data for training and validating AI models. You need different datasets depending on the focus of your work, such as customer interfaces, internal operations, regulatory compliance and surveillance, or financial markets.

Creating a sound data foundation

Think of a robust data foundation as the engine oil of a high-performance car; without it, even the best AI applications will sputter and stall. While AI and advanced tools are exciting, they only deliver value if built upon a solid data foundation. For example, feeding unstructured data into a large language model (LLM) can yield results, but they may be speculative.

Here are the priorities in the data and analytics domain:

1. Data foundation components: Ensure robust data security, privacy, and data quality management.
2. People-centric elements: Foster a data-driven culture, implement effective data governance, and enhance data and AI literacy. Promote self-service analytics and real data discovery.
3. Technology: Select and integrate the right technologies to support your data and analytics initiatives.

Strategic business transformation should be driven by clear business cases, such as improving customer experience, driving product innovation, improving financial performance, optimizing the supply chain or workforce planning. This often requires tailored data sets specific to each use case and key performance indicator (KPI). Establishing a mechanism for managing data across diverse use cases is recommended. Conducting sanity checks is valuable as many AI and data initiatives fail. Determine what data is needed for each purpose and identify where and how to source this data. Regardless of the type of AI, you need high-quality, accurate and relevant data. Such data is the foundation for building effective and reliable artificial intelligence systems. Good data ensures that the models are trained properly, leading to better performance, more accurate predictions, and actionable insights. Without good data, even the most sophisticated algorithms and models can produce misleading or erroneous results.

Applying agile methods in the data world is challenging due to the extended timeframes for data tasks, the need for thorough documentation, and the dependencies on existing data structures. Foundational data stability requires careful planning, and regulated industries need upfront impact analysis to prevent future issues. These factors often conflict with agile principles of short iteration cycles, minimal documentation, and flexible planning. By prioritizing these steps, organizations can build a robust data foundation that maximizes the value of AI and advanced tools. This reduces the risk of project failure.

Building a strong data culture

The phrase "data culture eats data strategy for breakfast" aptly captures a key reality: without a strong data culture, even the best strategies are doomed to fail. Data culture encompasses the set of values, beliefs, and behaviors within an organization that promote the effective and ethical use of data for decision-making, process improvement, and innovation.

The cost of expertise might slow down data culture revolution like a speed bump on the road.”

Improving data culture can be approached from facilitation and enablement angles. Facilitation includes data strategy, leadership, and governance, while enablement covers data literacy, communication, and access. A key success factor is assigning responsibility for data culture. Best-in-class companies have dedicated teams at all levels, whereas laggards have fragmented responsibility or none at all. Data silos, one of the biggest inhibitors of digital transformation, often arise from organizational structures, disparate technologies, lack of communication, or cultural barriers.

Management needs to make crucial decisions related to data democratization and access principles, balancing the "need to know" versus the "right to know." This approach requires balancing innovation opportunities with the imperative to protect sensitive information. Investments in data culture pay off through improved decision-making, continuous process improvement, cost reduction, revenue growth, greater acceptance of decisions, and competitive advantage. Change to culture is a marathon, not a sprint. Regularly sharing and celebrating success stories can shape data culture effectively.

Investing in common good for data and AI initiatives

AI-enabled business transformation is a journey. It requires a clear structure and KPIs to anchor the change. Typical success patterns for achieving business value from AI include demonstrating tangible benefits, selecting a strategic area to focus efforts, defining key changes, integrating with existing initiatives, and gaining IT commitment for technology support. Engage a committed business owner to drive and support the change, with designated change agents to manage day-to-day operations and sustain momentum.

Data alone doesn't drive business value; its true potential lies in what it can achieve. The actual value of data becomes evident when it is harnessed to generate insights or products. This involves extracting, refining, and processing data to unlock its potential. However, this process is complex and requires significant effort to transform raw data into valuable business outcomes. Therefore, when selecting a data project, it is crucial to ensure that it will create more business value than it incurs in costs.

Creating value from your data comes with associated costs which can be significant. These include:

• Data storage: Ensuring compliance with data regulations and laws.

• Data security: Implementing measures such as firewalls, encryption, and intrusion detection systems.
• Data quality, governance, and analytics: Allocating dedicated and potentially expensive resources to maintain accurate, consistent, and properly labeled data, which is crucial for effective decision-making.
• Data archiving and disposal: Managing these processes can be complex and time-consuming. Improper archiving and disposal can result in significant regulatory and legal penalties.

Two practical tips to start with:

1. Begin your data journey with "no regrets" moves, like enhancing data quality, while developing various strategies for digitalization, data, and AI. For example, upgrade data quality improvement capabilities (e.g. a tool that is able to put tags on the data traffic automatically).
2. Minimize data volume by identifying necessary data and removing duplicates. Reducing data volume simplifies quality improvement and offers significant cost savings, especially in cloud environments. Establishing a data lifecycle strategy to manage data from creation to archival and deletion is essential.

While these may seem like operational concerns, they require top management's attention and support. Investments in data enhancements benefit the entire organization. Top management should encourage mid-management to invest in areas that serve the common good. Investing in data foundation is essential to tap into the potential value of your data and AI while mitigating risks.