The closest thing to a universal belief in business today is that artificial intelligence (AI) has the power to deliver decisive competitive advantage. In fact, 91% of the 700 global C-suite executives surveyed by Forbes Insights agree that AI adoption will help them outpace their industry rivals in the years ahead.

But successful implementation of AI and machine learning requires fuel in the form of data, and a steady supply of it. And according to the survey, this is where the wheels fall off for many companies. Even though data is central to delivering the insights and predictive power of AI, just 12% of executives we surveyed say their organizations have built, and are executing, a company-wide data strategy. And almost 80% of executives say 40% or less of their data is available for sharing across their companies.

Without a comprehensive, enterprise-wide data strategy, organizations aren’t in a position to reap the competitive benefits of AI adoption. So how can you design a forward-looking data strategy? Consider the following:

• Reimagine your company’s vision of data. Traditionally, data has been an operational concern, focused on collection, storage, security and access. An AI-ready vision of data recasts it and ties it to outcomes versus operations, and as a business asset with significant value versus an IT-centric project that appears as a line item on an expense sheet. Instead of merely looking at data as a means of showing what happened, it becomes a guide to what can happen and how to make it happen. However, according to the Forbes Insights survey, only 45% of executives say the C-suite within their organization strongly evangelizes the need for a comprehensive, enterprise-wide data strategy.
• Get all key players across the business involved. An enterprise data strategy that will support, sustain and drive AI initiatives requires active involvement across the organization. A data strategy should be designed and carried out by cross-functional teams of business leaders, data scientists and IT. As a senior leader, this includes redefining your own role in the enterprise to that of coordinator and facilitator of data strategy, freeing up essential resources and serving as the bridge between IT and business stakeholders. This is an area where diligence can pay off, as only half of firms (54%) say they’ve built such a cross-functional team.
• Define the problem or opportunity. Involving the business is also important to defining the issue at hand. Business leaders should ask the questions: What are we trying to solve for with AI, what does success look like, and how is it measured? IT leaders can answer what data types you need to ensure success. Among other lofty goals, the major firms we surveyed were looking to use AI to optimize processes, evaluate investment opportunities and improve customer experience. Each of these goals requires its own set of data. For example, applying AI to customer experience may include looking at multiple data sources, such as customer satisfaction survey results, social media, advertising metrics, and browsing or purchase data to create personalized experiences.
• Determine your data’s AI-readiness. Data is the raw material of AI, but it comes with its own set of challenges. In fact, while 82% of the executives we surveyed say the data from at least some of their departments or functions is AI-ready, only 14% say their data is ready enterprise-wide. And with good reason. Data is constantly being created and updated from various sources, it’s often stored in different locations, and it comes in different forms. For example, in the customer experience example above, departments may have different ways of classifying a customer, creating multiple versions of a single customer’s data. The result is data that is potentially of poor quality, incomplete, overlapping, irrelevant to the business challenge at hand or even biased—all of which will impede successful AI adoption and implementation.Simplifying data management, access and protection across the enterprise is essential. This includes consolidating company data, defining clear labels and parameters, making it available to the business and enforcing its governance—essentially creating a single source of truth across your enterprise. It’s no wonder that 70% of executives consider data governance to be a work in progress, and less than half (48%) say their data management and integration systems are appropriately scaled.
• Develop a strategy to acquire new data. Once you’ve wrangled management and governance of your internal data, it’s time to start identifying any gaps in your data set and take the steps to fill them. Some data may be available publicly or by partnering. For example, a retail firm looking to forecast demand may want to partner with a provider of weather data in order to predict how certain weather patterns might impact foot traffic. This data then gets fed into the same data management processes you’ve already outlined. While this is a viable way to build a data set, the holy grail is to build a proprietary data set that your competition is unable to replicate. This goes beyond individual data collection projects and involves creating a virtuous data cycle by embedding data collection into the fabric of your products. This way, customers merely need to use your product to create new data that you can then use to improve your product. Those product improvements get your current customers to use the product more and also attract new ones, which ultimately creates more data that leads to more improvements, ad infinitum. One hypothetical example would be an exercise bike manufacturer comparing sensor data telling them how vigorously and frequently users were exercising against the type of cycling class content they consumed, in an effort to improve that content and increase usage. This virtuous cycle can be kicked into high gear through the development and support of a digital platform that encourages innovation and data sharing from internal and external partners, developers and customers.

If there’s one thing you take away from this exploration of AI data strategies, it should be this: It’s not too late to get started, yet time is of the essence, as the pressure to adopt AI is coming not only from competition but also from boards of directors. Seventy percent of C-level executives we surveyed say their boards are strongly urging greater use of AI—only slightly less than data scientists themselves (82%). So start thinking about this now. That way, the next time your board asks you, “What’s our AI strategy?” you’ll know exactly what to say.