Data Strategy and AI Readiness: Three Critical Questions for Business Decision Impact

In the first blog in this series, I focused on the supply side of data strategy and AI readiness, specifically the quality, integration, metadata, and architectural design required to move beyond isolated pilots and support AI at scale. But delivery capability is only half the equation. Even when the data foundation is strong enough to support AI technically, a more fundamental and important question remains: are we building data and AI capabilities around the decisions that matter most to the business, or are we building around technology momentum?

In IIA’s view, the purpose of data strategy is not to produce visible AI activity. It is to improve the availability, timeliness, and usefulness of data for the people making decisions across the business. That means the demand side must come first. Before organizations decide what to build, they need to be clear on which decisions matter most, where those decisions fall short today, and what kind of information environment is required to improve them.

This challenge shows up differently across the industries we work with. In tightly regulated industries like banking, insurance, and healthcare, decision quality is deeply tied to control, auditability, and defensibility. In companies with complex operating environments like manufacturing, transportation, energy, and distribution, the pressure is often on coordination, speed, and performance. But the underlying issue is the same. If AI is not helping the organization make better decisions, faster and with greater confidence, then it is not a strategy. It is simply another layer of activity.

With that in mind, here are three demand-side considerations enterprise data and analytics leaders should be asking as AI pressure continues to build. 

Demand-Side Consideration #1: “Have we tied our AI efforts to business context or to AI momentum?”

One of the easiest ways to lose the plot on AI is to let urgency become a substitute for strategy. Pressure from the market can explain why leaders are paying attention. It cannot define what the organization is trying to accomplish.

That sounds obvious, but many enterprises are still getting this backwards. They begin with the question, “Where can we apply AI?” rather than, “What business outcomes are we trying to improve?” The result is usually a portfolio of loosely connected initiatives that generate interest but lack strategic coherence. Teams can point to experimentation, but they struggle to explain why certain investments matter more than others, who is supposed to benefit, or how progress will be measured in business terms.

A stronger starting point is to frame the conversation around business context rather than technology momentum. What are the most important performance issues the organization is trying to address? Where are decisions slow, inconsistent, weakly informed, or difficult to scale? Where is poor data creating friction for growth, service, risk management, operational control, or cost discipline? Those are the questions that should shape the agenda.

AI is not a business objective on its own. It is a means of improving how the business operates. In some organizations, that may mean better pricing decisions, more consistent risk assessment, faster claims handling, or more informed outreach to customers and members. In others, it may mean better inventory positioning, stronger maintenance decisions, faster response to disruptions, or improved coordination across plants, fleets, or business units. The point is not that every enterprise needs the same use cases. The point is that every enterprise needs a clearer connection between its AI efforts and the decisions that drive performance.

When D&A leaders skip this context-setting work, the enterprise tends to drift toward what is easiest to launch or easiest to talk about. That may create the appearance of momentum. But it rarely creates the kind of alignment required to sustain long-term value.

Demand-Side Consideration #2: “Do we know which decisions matter most or are we still talking in vague use cases?”

Once the business context is clear, the next challenge is specificity. Many organizations can describe broad areas where they want AI to help: customer experience, operations, forecasting, automation, risk, productivity. But those categories are too abstract to guide a serious data and AI agenda.

The real unit of value is not the category. It is the decision.

Enterprise discussions about AI often remain too loose at exactly the point where precision matters most. Under current conditions, that is understandable. Leaders are being asked to move quickly on AI, often before the business has clearly defined what it wants AI to improve, where it should be applied, or how success should be measured. But “personalization,” “intelligent automation,” or “supply chain AI” does not tell us what decision is being improved, who makes it, what information they need, what constraints they operate under, or what better performance looks like. Without that level of precision, it becomes difficult to prioritize investments, define requirements, or judge whether a proposed initiative is strategically important.

A better approach is to push every use case discussion down to the level of the decision itself. What exactly is the business trying to decide? Who owns that decision? What are they doing today instead? Where does the current process break down? What is the cost of delay, inconsistency, poor visibility, or low trust in the data? What would improve if the decision became faster, more accurate, more repeatable, or better coordinated?

The goal is to materially change the conversation at the business-level before investing in data and AI platforms or labor-intensive builds. A pricing organization does not need “AI for pricing” in the abstract. It may need better support for exception handling, elasticity analysis, or margin tradeoffs in volatile conditions. A claims function does not simply need automation. It may need better triage, faster escalation, or stronger fraud-related decisions. A logistics team does not just need predictive analytics. It may need better decisions about routing, shipment prioritization, recovery actions, or network capacity under changing conditions. Not to mention, many of these business needs can be solved with traditional analytics and AI solutions that have been used for years.  

Demand-Side Consideration #3: “Are we prioritizing AI initiatives for business decision value?”

Most enterprises are not short on AI ideas. They are short on a clear way to decide what deserves investment first.

Prioritization is often an inflection point in AI strategy. In practice, many organizations favor the projects that are easiest to launch: the cleanest data, the strongest local sponsor, the lowest political friction, or the fastest path to a visible pilot. Those factors do matter. But when they become the primary criteria, the portfolio starts to tilt toward convenience rather than business importance.

A stronger prioritization plan begins with decision value. Which opportunities improve decisions that matter most to enterprise performance? Which ones address a real operational, financial, or risk-management problem? Which ones create visible business benefit if they succeed? And just as importantly, which ones help build reusable data, governance, and delivery capabilities that make the next wave of work easier?

A good use case should do more than prove that AI can work. It should improve an important decision and strengthen the enterprise’s ability to support similar decisions over time. That may mean creating reusable data flows, clearer ownership, better controls, stronger lineage, or more consistent engagement between business and technical teams. When prioritization works well, the business gets value from the immediate use case and leverage from the capabilities built around it.

In our experience, the most strategically important early use cases are rarely the flashiest. A high-profile assistant or generic productivity pilot may generate enthusiasm, but a less glamorous initiative tied to pricing discipline, risk assessment, claims leakage, receivables management, service recovery, maintenance prioritization, or network performance may create more lasting enterprise value. It may improve a decision the business already knows is important. It may produce clearer measures of success. And it may create more reusable infrastructure for what comes next.

Across industries, the specific decision domains may differ, but the principle is the same. In banking and healthcare, for example, the highest-value use cases often sit where decision quality, consistency, and control are tightly linked. In manufacturing, logistics, or transportation, they more often sit where decisions must be made repeatedly across complex, interconnected operations. In both cases, the real priority is not what is easiest to demo or easiest to position internally. It is where better decisions will materially improve business performance and where the resulting capability can scale.

So, data and AI leaders, take a look across your current AI agenda and ask a direct question: which of these efforts are truly designed to improve important business decisions, and which are simply evidence that we are doing something with AI? The answer is often revealing. Because if AI is not helping the enterprise make better decisions — faster, more consistently, and with greater trust — it is not yet a strategy. It is activity. And activity is easy to mistake for progress.