Why Clinician Adoption Fails Even When the Playbook Is Followed

Why Clinician Adoption Fails Even When the Playbook Is Followed

When a regional medical center bought an ambient AI scribe, ran the training, and secured leadership endorsement, the internal medicine department it was meant for mostly never used it. The standard playbook — find physician champions, train more — would not have fixed it. That case, and the lessons it produced, sit at the center of a conversation with Aditi U. Joshi, MD, MSc, FACEP, CEO of Ardexia, a digital health adoption advisory firm.

Joshi's account is a useful corrective to the way adoption is usually discussed online, where it is often framed as a matter of explaining benefits, securing buy-in, and running training sessions. Her experience points somewhere less tidy: the clinic day, the moment a tool costs someone time, and the weeks after go-live when support teams have already moved on.

When the Usual Advice Doesn't Work

Asked about a rollout that defied the standard guidance, Joshi described the ambient AI scribe deployment. The tool was purchased, training happened, and leadership endorsed it. The intended department still did not adopt it.

She said "The usual advice says to find your physician champions and train more but neither would have fixed this. The barriers were practical and a little humbling, and a few of them had nothing to do with the technology."

One intervention stands out for its simplicity. A staff member was placed in the clinics for a week to help with workflow and answer questions in the moment. According to Joshi, "Within three weeks, use had gone up by about 10 percent." She is careful about the result, noting it is early data and that she is following up to see whether it holds. Still, the takeaway is clear: "the cheapest fix is often the least glamorous one."

What the Online Advice Leaves Out

The gap between published guidance and clinical reality is where much of Joshi's work happens. She argues that adoption is routinely misdiagnosed.

She said "Online advice treats adoption as a communication problem: explain the benefits, get buy-in, run training. In practice it is a workflow problem." The failure point is specific: "Clinicians drop a tool the first time it costs them time during a busy clinic, and nobody asked them what that moment looks like." Just as important, she adds, is that "We also do not ask them what it will take to get them back."

A second omission concerns timing. "Adoption isn't a launch event," she said. "Use drops off in the weeks after go-live, when the support team has moved on, and almost nobody budgets for that stretch." That post-launch window, often treated as an afterthought, is where sustained use is won or lost.

Joshi also draws a distinction between compliance and adoption. She points to her time at Thomas Jefferson University Hospital, where faculty use of tools reached 100 percent without any requirement. "People used them because we built the conditions that made them worth using; it was stepwise and took a few years but that's what it takes," she said. Her view of mandates is blunt: "A mandate gets you logins, perhaps, but it doesn't get you people who rely on the tool."

The Decisions Made Too Quickly

Brought in before a rollout begins, Joshi sees the same premature choices repeatedly. The first is who serves as champions. Teams typically pick the most enthusiastic or most senior physician, "who is rarely the person whose daily routine will be hardest to change."

Second, timelines and training plans get locked in before anyone has mapped what the tool changes about a clinic day. The result, she said, is that "the schedule drives the rollout instead of the other way around." Third, success metrics are chosen too early, usually logins or licenses activated, which she says hides whether anyone is using the tool in a way that matters.

Her recommendation is a deliberate pause. "I'd rather a team spend two weeks watching how work gets done before they decide anything else," she said.

Practical Takeaways

For teams preparing a rollout, Joshi's account suggests several concrete steps:

• Watch the workflow before choosing champions. Identify the people whose daily routine will be hardest to change, not simply the most senior or enthusiastic clinicians.
• Map the moment a tool costs time. Ask clinicians what that moment looks like in a busy clinic, and what would bring them back after they drop the tool.
• Budget for the weeks after go-live. Plan support beyond launch, when use typically declines and support teams have moved on.
• Delay the success metric. Logins and activated licenses can mask whether the tool is being used in a meaningful way.
• Consider low-glamour support. In Joshi's example, a staff member present in the clinic for a week accompanied a measurable early rise in use.
• Separate compliance from adoption. A mandate may produce logins, but not reliance on the tool.

The broader point is that adoption work is less about persuasion than about conditions — the practical, often unglamorous arrangements that make a tool worth using in the middle of a clinic day. As Joshi put it, building those conditions is "stepwise and took a few years but that's what it takes."