An Operator’s Guide to Generating ROI From AI

For all its upside, the artificial intelligence boom has also created a core challenge for operators. Despite significant investment in AI adoption, many operators still aren’t seeing meaningful ROI materialize on the balance sheet.

In fact, while global spending on AI is expected to reach $632 billion by 2028, an MIT analysis found that only about 5% of enterprise AI pilots deliver measurable financial returns, with the vast majority generating little to no ROI. This gap has created mounting pressure on operators to translate dollars into impact, often leading to resources wasted on failed pilots or hasty investments in solutions that look promising on paper but fall short in practice.

The reality is that success in the AI era won’t be defined simply by the novelty or sophistication of a new technology, but by how discerning teams can be in understanding their fundamental challenges and choosing tech-enabled solutions that deliver real value. There’s no silver bullet for getting it right, but a few considerations can help get your team moving in the right direction.

Avoid the Urgency Tax

One key barrier to AI ROI is letting the fear of being left behind guide decision-making. When this mindset influences strategy, organizations can pay an urgency tax, burning valuable time, energy, and resources in an effort to keep up with the latest trends.

Internal and external forces can trigger that pressure. When leadership sees a competitor touting a new AI capability, a rapid descent into the comparison trap can follow, and what starts as a desire to stay relevant quickly turns into a reactive race to respond.

Investments made from this starting point fail for many reasons, but one of the most common is insufficient readiness. While a competitor may offer a similar product or service, an organization’s data foundation or operational maturity may not be strong enough to support the same technology, turning what appears to be a strategic move into a risky bet.

That’s why managers and directors closest to day-to-day operations are often best positioned to inform technology decisions. When a seemingly must-have technology comes to market, these teams should be tasked with first assessing whether there’s a clear problem it can solve and whether the organization is truly ready to support it. Because they understand where friction exists, where time is being lost, and where technology might make an impact, they can help ground AI decisions in operational reality rather than chasing novelty.

Conduct a Bicycle Audit

Another common tech procurement pitfall is over-buying. This differs from the urgency tax because it occurs after determining that a true need exists and you’re operationally ready to purchase an AI solution. At this point, the question becomes not “do we need something” but “what do we actually need”?

This problem is particularly prevalent in legacy-bound industries like logistics, which has gone from 0 to 60 with technological possibilities in recent years. Where once our challenge was tackling modern complexities with outdated systems and processes, today it’s choosing from the infinite technology wish lists available from third-party providers or through in-house development.

A “Bicycle Audit” can help immensely before reaching the point of purchase. It challenges decision-makers to answer a simple question: Do we need a Ferrari or a bicycle? Ambitious tech teams love to dream big, and third-party providers typically aim to offer their top-tier solution right out of the gate. Both are valid, but investing in Ferrari-level horsepower doesn’t make sense when a bicycle will get you where you need to go.

Audit With Metrics

One way to make that decision is to understand the problem you’re trying to solve across three metric levels: Primary, Secondary, and Tertiary. Assessing all three together helps clarify where friction exists, what optimal performance looks like at each layer, and how much investment is necessary to close the gap.

Tertiary metrics represent core operational behaviors. Significant inefficiencies often live at this layer, and bicycle-level solutions that enable improvements like cleaner data capture and more efficient execution can make a big impact with relatively small investment.

Secondary metrics reflect the real performance drivers — think customer conversion rates and other levers teams can influence through increased productivity. Solving inefficiencies here typically requires something more advanced than a bicycle but less complex than a Ferrari, such as sophisticated automation that can handle larger datasets.

Primary metrics are the big rocks like revenue. This is where Ferrari-level solutions tend to appear. It’s typically high-ticket technology that promises material impact on the bottom line. While worth exploring, it’s critical to remember that unless secondary and tertiary challenges are addressed first, these solutions can fall short of their true ROI potential.

Smaller, targeted investments at lower levels are often the best place to start because they tend to deliver rapid results. They also create opportunities to learn what works while providing incremental gains that compound over time, ultimately helping build toward the same or greater total impact as larger investments, with far less risk.

Together, the Bicycle Audit and this three-tiered metric framework help organizations mitigate risk by right-sizing solutions to real problems. The point isn’t to avoid advanced AI, but to start small by solving the most impactful problems with the least investment required and scaling from there.

Be Strategic About Startup Partners

The recent surge in AI-related venture capital has flooded the market with new startups. These disruptors will come to the table with pitches promising innovation and results compelling enough to sway even the most discerning procurement teams.

But buyer beware: both the products and the people behind many of these newcomers are often unproven. Becoming an early adopter carries inherent risk, including the possibility that you might be unknowingly building the product alongside them. While that can offer upside, it should be a conscious choice — because when you’re trying to move the needle on problems with real financial implications, spending valuable resources helping a vendor fine-tune its latest update can introduce unnecessary headaches.

Once a vendor is integrated, much of the outcome sits outside your control. Their roadmap, customer support scalability, pricing dynamics, and ability to sustain performance as they grow are all subject to change. Those shifts can shape the long-term value of the partnership in ways that aren’t fully visible at the outset.

Navigating that uncertainty requires patience and discernment on the front end. Taking time to validate a solution through a proof of concept, understanding contractual commitments before deeper integration, and speaking directly with existing users helps teams choose providers positioned to deliver value over the life of the partnership.

Making AI Pay Off

Taken together, these considerations reinforce the reality that practicing strong discernment is the first and most critical factor in generating ROI from AI. When teams focus on identifying real friction, results improve because inefficiencies are removed and time is reallocated to higher-value tasks. That’s what true ROI looks like, and it’s only earned through discipline, clarity, and pragmatic decision-making that benefit the bottom line over time.