Why your AI initiatives aren’t delivering ROI—and how to fix it

Nearly every MSP knows that AI is shaking up the industry, and getting onboard is essential to not get left behind. However, generating measurable ROI from AI investments has proven to be more challenging than many businesses ever expected. Here’s a closer look at why organizations are struggling with AI initiative ROI—and what to do about it.

ROI on AI initiatives falling behind

Recent research from Gartner found that while AI remains a top priority and investment area for organizations, many are still failing to see meaningful returns. According to the numbers, 63% of boards of directors rank technology investments as a key strategic priority, and 72% of CEOs view AI as their primary growth driver. Despite this, only a reported 11% of CFOs have been able to concretely measure ROI from AI investments.

To begin resolving this discrepancy, organizations first need to understand that time saved does not automatically equal money saved. “What [Gartner] found is that increased productivity has not led to financial value,” says Gartner Distinguished VP Analyst Frances Karamouzis.

This comes down to the distinction between what Gartner calls blue money and green money. “Blue money is productivity [improvements] and efficiencies, but green money—which is direct cost savings or revenue growth—can only come from critical business changes, from changing the process workflow,” says Karamouzis.

In other words, AI-driven productivity gains may streamline operations, but they won’t impact companies’ bottom lines without significant, corresponding changes to workflows and business processes.

The real cause of the AI value gap

According to Karamouzis, the AI value gap is not a technological failure. Instead, it stems from shortcomings in the following areas:

Value creation delineation and transparency. “Can your teams and your functional leaders not only identify, but translate and communicate what the sources of value creation are? In other words, how do you delineate value, and how do you make that transparent?”

Financial rigor and governance. “The financial rigor of actually doing a mathematical calculation, not just a simple sort of, back of the envelope calculation, is critical. [Also key is] the financial governance to enforce whatever framework or rules that you have for how you are going to vet, prioritize, and decide what to going to fund.”

Business process change. “Ownership of the business transition, changing the process workflows, changing the metrics of the humans that do those processes, engendering trust, looking at the operating model—these are the things that the business team is supposed to be doing, while the IT and the data team are busy developing and testing the AI solutions.”

Filter initiatives through a funding funnel

The key to turning your MSP’s blue money into green money is to “implement a rigorous, repeatable, enterprise-wide framework for funding AI,” says Karamouzis. “Enterprise-wide means that it’s going to be consistent amongst all [departments]. It doesn’t necessarily mean that only one person approves everything, but [each approver will] use the same rigorous, repeatable framework.”

The framework starts with the funding funnel. “The fundamental premise is that there’s no shortage of ideas,” she says. “There are hundreds of ideas, but the business can only fund a very small, finite number of ideas. So, every idea needs to be vetted. They need to be prioritized—juxtaposed against each other—especially if they are disparate ideas or if they’re coming from different departments.”

The litmus test to determine which AI initiatives get funded should be based on your business strategy. This is not one-fits-all; it should be unique to your organization, and every proposed AI use case should connect directly to one or more strategic business objectives.

“The business teams are also the ones that can tell you the minimum scope and scale needed to get the value [desired]. [For example], what are the specific KPIs you need to focus on? So, while the IT team is working on developing and testing the solution, telling you the time horizon, the technical risks, and the cost of doing this in the feasibility analysis, the business people are going to say, ‘What is it going to take?’ and create that transparency and that delineation of value,” Karamouzis explains.

Three types of AI initiatives

After verifying that AI initiatives align with business strategy, organizations should categorize them into one of three groups:

Defend—use cases that give competitive parity, marginal incremental gains, and micro innovations.

Extend—initiatives that will give you the highest ROI by growing your market size, reach, or your revenue, or by decreasing costs.

Upend—game-changing initiatives; these typically have a large time horizon, traversing more than one fiscal year

Karamouzis recommends limiting investment in Defend initiatives because they generally do not generate substantial ROI. “You might want to have a cap on it, somewhere between five and 15%,” she says. Then, “you might want to have the lion’s share of your use cases—80%—fall into the Extend category.”

Upend initiatives may be better suited for later stages of AI maturity. “Not everybody has the wherewithal or the patience for Upend initiatives, so you might introduce these later, not in the first round of funding approvals. Depending on where you are, you might want to choose one or two of these, [making up] a small percentage of your budget, somewhere between five and 15%,” says Karamouzis.

AI demands a new approach to innovation

“In the old days, we used to design, build, and run new initiatives very sequentially. In advance, we might do a business case. [But] this old school methodology is no longer working; it doesn’t work in AI. [Previously], the business case was done once, presented to leadership, and once they approve it, you’re off to the races,” Karamouzis says.

AI initiatives require a far more adaptive process. “What’s really happening in AI is a very different approach. It’s dynamic, it’s iterative, and you might have to go to management and have discussions as many as five times.”

Instead of relying on a one-time business case, companies should start with proof of value to validate both feasibility and business impact of each suggested initiative. From there, projects should move through proof of concept, pilot, execution, and scaling phases. At each stage, teams should reassess whether the initiative continues to deliver measurable value before continuing forward.

 With a rigorous framework and a clear connection between AI initiatives and business strategy, MSPs can significantly improve their ability to generate measurable ROI from AI investments. For more of the latest in AI, check out these 21 AI tools MSPs should be using.