Strategic Bets and Portfolio Thinking for AI Use Cases

If you’re responsible for analytics in a large enterprise, chances are you’ve already been pulled into the AI use case conversation. Leaders are under pressure to demonstrate momentum, launch pilots, and keep pace with competitors. But without a structure for choosing the right bets, organizations risk falling into a cycle of disconnected experiments that over-promise and under-deliver.

The truth is, AI success isn’t a matter of having the best model or the most novel use case. It’s about making better bets—decisions that reflect not just technical feasibility, but strategic alignment, sponsor commitment, and readiness to act on the results. And that means treating AI as part of your portfolio, as opposed to a collection of disconnected pilots.

At IIA, we’ve long advised analytics leaders to adopt portfolio mindsets. It’s one of the key disciplines in our Returned Business Value (RBV) framework, introduced in our first blog in this series. But as AI reshapes the analytics landscape, this mindset is no longer a best practice. It’s a necessity.

From Proof of Concept to Portfolio Strategy

In traditional IT governance, the idea of betting rarely enters the conversation. Projects are scoped, approved, and funded based on predictable inputs and known outputs. But AI doesn’t work that way. Nearly every strategic analytics effort is a bet—one made with incomplete information and a degree of risk.

What makes AI different is that the risk is mostly organizational. The model might work fine in testing, but if no one owns the outcome or is ready to act on the insight, the return never materializes.

That’s why the RBV model reframes analytics investment as strategic bet-making, and encourages leaders to define and manage their portfolios accordingly.

In a well-run analytics portfolio, you should be able to:

• Stack-rank projects based on strategic alignment and business priority
• Quantify both benefit and cost, including technical and opportunity costs
• Adjust for risk, using a consistent scoring model that reflects partner readiness
• Assign ownership for both the technical delivery and the business outcome

The goal is to make bets visible, so they can be prioritized, resourced, and evaluated for what they are: calculated decisions in uncertain territory.

Operationalizing Your AI Portfolio: What It Takes

In our advisory work, we see organizations get stuck not because they lack AI talent or use case ideas, but because they don’t have the organizational muscle to prioritize, evaluate, and manage those efforts as a portfolio. In other words, they’re making bets, but not managing risk. They’re launching proofs of concept, but not ensuring outcomes. And often, they’re missing the single most important success factor: business ownership.

If you’re leading analytics and AI work in a large enterprise, the first and most pragmatic step you can take is to enforce a simple rule: No business sponsor, no project. One of the analytics leaders we support put it best: “I won’t work on something unless it’s a true business problem, and at least a VP has acknowledged it’s important to their org.”

That discipline weeds out the shiny objects and protects scarce analytics capacity. But more importantly, it draws a line of accountability. If a business unit believes an AI use case will deliver X dollars of benefit or save Y hours of staff time, they should define how that value will be measured and reported. The analytics team can enable the outcome, but they can’t own it.

Value Tracking Is a Joint Responsibility

One of the hardest operational challenges with analytics ROI is tracking the value after delivery. And most organizations aren’t great at this. Some business teams, like utilization management or pharmacy, are naturally strong in this area. They’re already wired to track FTE reductions, turnaround time, or avoided costs. They write their own impact summaries for leadership and celebrate the analytics team publicly.

Others aren’t there yet. Teams like case management or IT may have compelling ideas but lack the operational maturity or measurement tooling to demonstrate impact. This unevenness is common, and it’s one reason portfolio thinking matters. Your best-performing partners should get more time and attention. Not because they’re easier to work with, but because they deliver results that justify continued investment.

In practical terms, we advise analytics leaders to assign a “maturity score” to business partners. Who’s consistently defining success upfront? Who’s measuring impact after the fact? And who’s evangelizing the work across the enterprise? Use these insights to guide prioritization, allocate resources, and reinforce expectations. If a partner can’t tell you how they’ll track value, the project may not be ready to fund.

Make Your Portfolio Visible to Everyone

Whether you use a dashboard, internal newsletter, or a literal “value thermometer” poster in the office, the goal is to make progress visible. Too often, analytics teams complete high-impact work but fail to connect the dots for leadership. And in the absence of a clear story, perception wins out over performance.

Make it easy for your stakeholders to see what’s in flight, what’s working, and what’s creating value. Don’t wait for quarterly business reviews. Share short summaries. Link outcomes to enterprise priorities. Where possible, report on hard-dollar value—savings, revenue lift, reduced call time. But don’t overlook operational metrics that build the case: time saved, decision latency improved, or customer satisfaction nudged upward.

And crucially, track the soft metrics too. Are your business partners coming back for more? Are they bringing new ideas to the table? Are they recommending the analytics team to their peers? These relationship signals matter, especially when hard-dollar ROI is still in the future.

Use Portfolio Categories to Stay Balanced

Not every project will deliver immediate, measurable value. Nor should it. A healthy analytics portfolio contains a mix of short-term wins, medium-term strategic bets, and a handful of long-range innovations. A successful mix might look this:

• 60% clear, high-impact bets tied to known problems and measurable outcomes
• 30% strategic plays aligned to corporate initiatives or long-term capability building
• 10% experimental ideas that could uncover new value areas down the line

This kind of balance protects near-term credibility while making room for long-term growth. It also creates cover for innovation. Analysts get time to work on future-state ideas that haven’t yet found a home. And when business priorities shift—as they always do—you’re not starting from zero.

This structure is especially important in the context of AI. Some use cases, like contact center summarization or routing, have quick feasibility and obvious benefits. Others, like CSAT optimization using LLMs or deeper patient segmentation, require longer timelines and more nuanced metrics. A portfolio model lets you manage both without letting one cannibalize the other.

Align to Strategic KPIs (Or Don’t Do the Work)

If you can’t tie a proposed AI use case to a strategic or operational KPI, stop and ask why. AI should be an enablement tool, not a science fair project. One of the best practices we see is having business units define the impact model up front: What metric will change? By how much? Over what time frame? And what happens if the outcome doesn’t materialize?

To be clear, these estimates aren’t always precise. But they should be grounded in fact, not guesswork. Use back-of-the-envelope modeling, Monte Carlo simulations, or benchmarking as needed. And revisit the estimates after deployment. Did the chatbot reduce handle time by 30 seconds per call? Did the predictive model reduce avoidable readmissions by 2%? These can powerful proof points for continued investment.

Strategic Bet-Making Is a Cultural Shift

Ultimately, portfolio thinking is a mindset. It requires business partners to take ownership, not just make requests. It asks analytics teams to push back, not just take orders. And it demands transparency in how work is selected, funded, and evaluated.

Put another way, you’re prioritizing both use cases and partners. This allows you to double down on teams that value the analytics relationship, track outcomes, and tell the success story up the chain. It can also help your team say no—with confidence—to efforts that aren’t ready or aligned.

If you’re trying to demonstrate ROI from AI, start by making fewer, better bets. Build a portfolio that reflects your strategic priorities and your team’s capacity. Track outcomes. Share the wins. And most of all, make sure the business has skin in the game. This approach builds the credibility and traction you need to secure future investment.