The AI ROI Gap Is Still Wide Open. Here’s What the Latest Research Says About Closing It.
- Mar 10
- 4 min read
Last August, I wrote about MIT’s finding that 95% of enterprise AI pilots fail to deliver measurable ROI and what that meant for professional services marketers. You can read that post here. The response was strong, which told me the question is real and the pressure is not going away.
Deloitte’s freshly published 2026 State of AI in the Enterprise report, based on a survey of 3,235 senior leaders across 24 countries, confirms that the gap between AI adoption and AI results has not closed. In some ways, it has grown more complicated. The good news is that the path forward is becoming clearer.
Here is what the new data shows and what it means for our work.
The Picture in 2026: More Access, Not Enough Results
Access to AI tools inside organizations has expanded fast. Deloitte found that the share of workers equipped with sanctioned AI tools grew by 50% in a single year, reaching roughly 60% of the workforce. That is real progress.
The challenge is what happens next. Fewer than 60% of those workers with access actually use AI in their daily workflow. Revenue growth from AI remains an aspiration for 74% of organizations, with only 20% reporting they have actually achieved it. And only 34% of companies say they are using AI to deeply transform their business, creating new products, reinventing core processes, or rethinking their business model.
Deloitte describes this as the “activation gap”: the distance between having AI tools and getting genuine business value from them. The MIT findings we discussed last summer identified the same gap from a different angle. The consistent message across both reports is that adoption is not the hard part anymore. Activation is.
Why the Gap Persists
The Deloitte data points to several specific reasons organizations are not converting AI access into AI impact.
Training without restructuring. The number one way companies adjusted their talent strategy due to AI was education, teaching people how to use tools without actually rethinking how work gets done around those tools. Only 16% of organizations redesigned roles to align with AI capabilities. You cannot get transformation by layering new tools onto old workflows.
Poor user experience on enterprise tools. Deloitte found that enterprise AI applications are running roughly one to two years behind consumer tools in user experience. People who use ChatGPT personally every day encounter clunky, disconnected internal tools and disengage. The shadow AI economy MIT identified last summer is still thriving precisely because personal tools are simply better to use.
Governance and readiness gaps. Technical infrastructure readiness sits at 43%, data management at 40%, and talent readiness at just 20%. These numbers actually declined from the prior year, suggesting that as AI ambitions grow, organizations are falling further behind in their ability to support them.
Chasing efficiency instead of reinvention. The majority of organizations using AI are getting productivity and efficiency gains. Two-thirds reported those benefits. But only the companies using AI to reimagine their business, not just optimize it, are seeing the kind of strategic differentiation that justifies the investment at scale.
What the Leaders Are Doing Differently
Deloitte’s 2026 report identifies a clear leading cohort: the 34% of organizations that are using AI to deeply transform rather than simply automate. Several common practices set them apart.
Senior leadership drives governance, not just strategy. Enterprises where senior leaders actively shape AI governance, not just endorse it from a distance, achieve significantly greater business value. This is not about signing off on a policy document. It is about being engaged in how AI decisions get made day to day.
They redesign work, not just train workers. The companies seeing real transformation are rethinking how work flows, not just teaching people to use new tools. That means restructuring roles, redefining outputs, and measuring results differently.
They integrate AI into familiar tools. Successful implementations tend to bring AI into platforms people already use: Microsoft Teams, Slack, existing CRM and document management systems. Asking people to switch to a new AI-specific interface creates friction and reduces adoption.
They look beyond efficiency to reinvention. Productivity gains are a starting point, not the destination. The organizations pulling ahead are asking a harder question: how does AI help us do things we could not do before, create value we could not create before, or serve clients in ways we could not serve them before?
They partner rather than build alone. This finding from the MIT report holds up in the Deloitte data. Externally built and customized tools outperform internal builds. Vendors who understand your workflows can move faster and adapt more effectively than internal science projects.
Partnered Intelligence POV
The Deloitte 2026 report does not contradict the MIT findings from last summer. It updates them. The problem has evolved from “why are pilots failing?” to “why is broad access not producing broad transformation?” That is actually a more useful question for professional services marketers, because it moves us from diagnosis to action.
The activation gap Deloitte describes is one we can influence directly. We are often the people responsible for communicating AI’s value internally, helping teams understand how to use it, and building the workflows where it gets embedded. That is meaningful work, and this data gives us the business case for doing it more deliberately.
The firms that will close the gap are not necessarily the ones with the most sophisticated tools. They are the ones willing to ask harder questions about how work actually gets done and where AI can genuinely improve it. That work starts with people, not platforms.
Where are you seeing the activation gap show up in your own organization? And what is actually working to close it? I would love to hear what you are learning.