The numbers are out. Forbes and PwC’s latest data puts 56% of CEOs in the zero-ROI category on AI spend. That’s not a rounding error — that’s the majority of enterprise AI investment producing PowerPoint decks, proofs of concept, and stalled pilots instead of measurable business value.
The question worth asking isn’t why the number is so high. It’s what the other 44% are doing differently.
The answer isn’t a better model or a more sophisticated prompt. It’s structural.
The Common Pattern Among Failed AI Projects
When we look at organizations stuck in the 56%, a few patterns show up consistently:
No defined ownership. The AI initiative belongs to everyone and therefore to no one. IT owns the infrastructure. The business unit owns the use case. Data owns the pipelines. Nobody owns the outcome. When results are diffuse and accountability is distributed, projects drift.
Pilots designed to succeed in isolation. A proof of concept that runs on clean sample data, evaluated by the team that built it, compared against a manually established baseline will almost always look good. It’s when that pilot meets real data, real users, and real organizational processes that the seams show. Most pilots aren’t designed to survive contact with production.
Measuring activity instead of outcomes. “We’ve deployed AI to 300 employees” is an activity metric. “We’ve reduced contract review time by 40%” is an outcome metric. Organizations chasing adoption numbers instead of business impact will optimize for the wrong thing — and wonder why the ROI isn’t materializing.
Integration debt from day one. Every AI layer added on top of existing systems creates connective tissue that has to be maintained, secured, monitored, and eventually replaced. Organizations that underinvest in the integration layer are paying invisible carrying costs that compound over time. The ROI calculation never includes this, which is part of why the ROI often doesn’t close.
What the 44% Do Differently
The organizations consistently seeing AI returns aren’t necessarily working with better technology. They’re making better structural decisions before the first line of code gets written.
They start with a specific, measurable problem. Not “we want to leverage AI across the organization.” A specific process with a defined current-state metric and a target. Legal review. Customer support triage. Inventory forecasting. The narrower the initial scope, the faster the path to real numbers.
They treat integration as a first-class workstream. The AI model is 20% of the project. The integration into existing systems, data pipelines, user workflows, and governance structures is 80%. Organizations that staff and budget accordingly ship working systems instead of isolated demos.
They design for scale before they prove the concept. The pilot question shouldn’t just be “does this work?” It should be “does this work in a way that can scale across the org?” That requires thinking about data governance, access controls, monitoring, and change management before the pilot is over — not after.
They have someone whose job it is to own the outcome. Not a committee. A person. With a P&L implication tied to the AI initiative’s performance. Accountability structures that diffuse responsibility across teams produce diffuse results.
The Structural Question to Ask Before the Next Initiative
If your organization is in the 56%, the investment decision isn’t about which AI vendor to choose or which use case to pilot next. It’s about whether the organizational infrastructure exists to turn a working pilot into a deployed system that produces real returns.
That infrastructure — governance, integration capacity, outcome ownership, change management — is what separates the 44% from the majority. It can be built. It usually just isn’t, because it’s less exciting than the technology itself.
The organizations getting AI ROI aren’t smarter. They’re more structured.
If you’re preparing to make the case for H2 AI investment and want an honest read on where your execution gaps are, that’s exactly what VitaLink does. Get in touch.