Access Is Not Adoption

I’ve been doing a deep dive into Deloitte’s 2026 State of AI in the Enterprise report to better understand why AI investments are not paying off the way most enterprises expected. If you’ve not yet had a moment to browse the report, I highly recommend you set aside some time to read through its insights.

Here’s a number worth starting with: workforce access to AI tools expanded by 50% in a single year. According to Deloitte, the share of workers equipped with sanctioned AI tools grew from under 40% to around 60% in twelve months. That is a remarkable deployment pace by any measure.

Here’s the number that should be sitting next to it: among those workers who now have access, fewer than 60% use it in their daily workflow. And that utilization rate is largely unchanged from last year.

So organizations dramatically accelerated access. Adoption barely moved.

Deloitte calls this the gap between access and activation, and they frame it as the primary barrier to AI value in 2026. I’d go further. I’d say it’s the most predictable outcome of how most organizations have been rolling out technology for the past three decades, and AI is simply the latest and most expensive example of it.

We’ve Conflated Two Very Different Things

Access means the tool is available. Activation (or, as I usually refer to it, engagement) means the tool is embedded into how work actually gets done.

These are not the same thing. They are not even close to the same thing. And the reason organizations keep treating them as equivalent is that access is measurable, reportable, and achievable on a timeline. You can announce that 60% of your workforce now has access to Copilot. You can put that in a board update. You can show a chart with an upward line.

Engagement is harder to measure, harder to report, and doesn’t happen on a deployment timeline. It happens on a behavior change timeline, which is slower, messier, and not particularly amenable to a project plan with a go-live date.

I’ve watched this dynamic play out with every major enterprise technology rollout I’ve been part of or observed over 35 years. We rolled out SharePoint and measured success by site creation and storage consumption. We rolled out Office 365 and measured success by license activation. We rolled out Teams and measured success by meeting minutes and message volume. In every case, the metric we chose was a proxy for adoption rather than a measure of it because actual adoption is harder to quantify and slower to demonstrate.

AI is following the same pattern, at greater speed and greater cost.

What Engagement Actually Looks Like

If access means the tool exists in your environment, engagement (which Deloitte calls activation) means it has changed how someone does their job. Not added a step. Not introduced an alternative. Changed the default. Made the old way feel slower and more effortful by comparison.

That shift happens at the individual level before it happens at the organizational level, and it almost never happens as a result of a training event or a rollout campaign. It happens when someone finds a use case that saves them real time on something they actually care about, experiences that value personally, and starts building the tool into their regular workflow.

The Deloitte report points to a structural reason why this isn’t happening faster: organizations are focused on deploying AI tools, not on embedding them into workflows. There’s a difference between giving someone access to a hammer and redesigning the assembly process around the fact that a hammer exists. Most organizations are doing the former and calling it transformation.

Hands-on, role-specific training and visible executive advocacy are cited in the report as the factors that materially shift employee behavior. That’s consistent with what I’ve observed in my workshop work. Generic AI literacy training — here’s what a large language model is, here’s how to write a prompt — builds awareness. It does not build habits. Habits come from repetition in context, which means the training has to be close to the actual work, not adjacent to it.

The 60% Who Aren’t Using It

Let’s talk about the people on the other side of that utilization number. Among workers who have access to sanctioned AI tools, more than 40% aren’t using them in their daily workflow. Who are these people, and why aren’t they engaging?

The Deloitte data gives us some texture here. Among non-technical workers, 13% are highly enthusiastic and actively seeking to use AI. 55% are open but not driving adoption. 21% would prefer not to use AI but will if required. 4% actively distrust it and avoid it entirely.

That’s not a monolithic resistant workforce. That’s a segmented population with very different relationships to the technology, and treating them as a single audience — which most AI rollout programs do — is a significant mistake.

The 55% who are open but passive are the most important group and the most neglected. They are not resistant. They are not enthusiastic. They are waiting for a reason that is specific to their work. They need to see the tool solve a problem they actually have, demonstrated in the context of their actual job, before they will change their behavior. A company-wide launch email and a Copilot training module on the intranet will not move them. A colleague showing them how it cut their weekly reporting time in half might.

The 21% who prefer not to use it but will if required are a different conversation, and one that most organizations are avoiding because it’s uncomfortable. These people are not going to self-activate. They need a clearer answer to a question they’re often not allowed to ask out loud: what does this mean for my job, my career, and my value to this organization? Until that question gets answered honestly, their compliance will be performative and their adoption will be minimal.

The 4% who actively distrust it are not your immediate problem. They’re a small enough population that organic peer influence is probably your best tool. Trying to convert active skeptics with training programs is rarely a good use of resources.

The Missing Middle: From Access to Habit

Most organizations have invested in the beginning and end of the adoption journey and left the middle unaddressed.

The beginning: procurement, deployment, licensing, access provisioning. Check.

The end: enterprise-level metrics, such as productivity gains, efficiency improvements, cost reduction. These are the outcomes leadership wants to report.

The missing middle is everything that happens between a worker getting access and the organization seeing measurable impact. That middle is where behavior change lives, and it requires a different kind of investment than the beginning or the end.

It requires role-specific use case development — not generic prompting guides, but concrete examples of how AI changes the specific tasks that specific job functions do on a regular basis. A finance analyst and a project manager and a customer service rep are not going to be activated by the same demonstration. They need to see their work, not a representative approximation of it.

It requires peer-driven champion networks, not just top-down advocacy. Deloitte’s own recommendation is that high-performing implementations start with empowered employees who experiment, share early wins, and become internal champions. I’ve seen this work in practice. The most effective AI adoption I’ve observed has happened laterally — one person showing another person something that saved them an hour, and that person showing three more. Top-down mandates generate compliance. Peer demonstrations generate curiosity, and curiosity is what eventually becomes habit.

It requires time and explicit permission to experiment. This is the one most organizations won’t budget for. Workers don’t form new habits in the margins of an already full workload. If using AI tools is something people are supposed to figure out on top of everything else they’re doing, most of them won’t. The organizations seeing the strongest adoption rates are the ones that have carved out structured time for employees to explore, experiment, and share — not as a training event, but as a recurring practice.

Engagement Is a Measurement Problem Too

Part of why the access/engagement gap persists is that most organizations aren’t measuring the right thing. They’re measuring availability — licenses assigned, tools deployed, training modules completed. These are inputs, not outcomes.

Engagement metrics look different. They look like: what percentage of workers use AI-assisted tools to complete at least one core task per day? How has time-to-completion changed for specific high-volume workflows? Which teams show the strongest behavior change, and what do their rollout approaches have in common?

These questions are harder to answer than “how many licenses did we assign,” but they’re the questions that tell you whether you’re actually getting anywhere. Organizations that can’t answer them aren’t managing adoption — they’re managing access, and hoping the gap closes on its own.

It won’t. The Deloitte data is fairly clear on this point: the utilization rate among workers with access has not meaningfully changed year over year, despite significant investment in expanding that access. More access does not automatically produce more engagement. It produces more people who have the tool and aren’t using it.

The Harder Ask

Closing the gap between access and engagement requires organizations to make a harder ask of themselves than deploying software.

It requires accepting that adoption is a change management problem, not a technology problem. It requires investing in the human layer of AI rollout with the same seriousness applied to the technical layer. It requires measuring outcomes rather than inputs, having honest conversations with resistant employees rather than mandating compliance, and building the kind of peer champion infrastructure that actually moves behavior at scale.

None of that is impossible. But none of it happens by accident, and none of it is captured in a deployment dashboard.

The organizations that close this gap — that move from access to genuine activation — will look very different from the ones that don’t. Not because they had better technology, but because they treated the human side of adoption as seriously as they treated the technical side.

That’s not a new lesson. It’s just one that the industry keeps having to relearn with every new platform.

Access is the beginning of the journey. Engagement (or activation) is the destination. Most organizations are celebrating arriving at the trailhead and wondering why they haven’t reached the summit.