Clinicians may keep using ambient documentation because it fits their encounters and leaves them with less work, less mental strain or more attention for patients. Others may stop when the draft needs too much repair, the workflow interrupts care, or the tool does not suit the notes they need to finish. A falling usage count cannot tell you which explanation applies.
An ambient scribe captures a clinical conversation and generates a draft note for review. For a health system's implementation lead, the adoption question is whether that reviewed workflow remains useful and appropriate in ordinary practice. Increasing the number of recordings is a poor substitute for finding out why a clinician has disengaged.
This article focuses on documentation in ambulatory care, where patients attend scheduled visits. Inpatient accounts can suggest questions, but they do not establish what will work in that setting. The broader roles and industries guide distinguishes documentation from other administrative work and clinical decisions.
First establish what changed
Start with the clinician's opportunity to use the tool. Leave, a different clinic schedule, a change in visit type or missing access can reduce recorded use without a decision to abandon it. A monthly active-user count also treats one trial encounter and sustained use alike.
Keep these situations separate:
| Observed situation | What you still need to establish | Useful first question |
|---|---|---|
Never started | Whether access, training and an appropriate encounter were available | Was there a real opportunity to try the approved workflow? |
Uses it selectively | Which encounters benefit and which do not | What makes you choose it for this visit? |
Tried it and stopped | The last useful or unsuccessful experience and any later change | What happened the last time you used it? |
Record shows little recent activity | Whether the record covers the work actually done | Were eligible visits or documentation responsibilities different? |
In an October 2026 Johns Hopkins study, researchers distinguished any use from the proportion of eligible ambulatory visits using a scribe. Their six-month observational analysis of 2,901 clinicians found different patterns, including sustained growth, transient use and early use followed by stopping. Uptake also varied by practice setting and specialty.
Those patterns locate a question; they do not answer it. The researchers did not measure why particular clinicians used the tool, whether notes were good or whether use benefited patients. Specialty and setting were descriptive comparisons, and the trajectories were not validated personality categories. An implementation team should therefore resist turning a usage pattern into a label such as reluctant adopter.
Useful for one note can be wrong for another
A clinician who uses a scribe for some encounters may have found its useful boundary. That is different from failing to learn it.
In a November 2025 Reddit discussion, a Reddit user described an inpatient documentation routine that combined spoken chart review, the patient encounter and a spoken assessment and plan. They found it useful for an admission history and physical, the account of a patient's history and examination. They were less convinced by progress notes, the daily record of changes during a hospital stay, and thought ordinary dictation might suit those better. They also acknowledged that they had not explored progress notes extensively.
That is an anonymous account, not a comparison of products or evidence about ambulatory effectiveness. Its useful detail is the difference between two notes in the same person's work. Asking only whether they like the tool would miss it. Asking what they have to say aloud, what remains elsewhere in the record and what they still rewrite gets closer to the actual fit.
For an ambulatory rollout, examine the encounter types your organization has approved. A long narrative history, a brief follow-up and a visit involving several speakers may create different documentation demands. Observe the actual work before deciding whether the problem calls for configuration, training, a narrower use case or another method. Do not ask clinicians to add unnecessary narration simply to make a use metric rise.
Check the work that a time metric misses
The evidence gives reasons for both interest and caution. In a randomized UCLA trial published in November 2025, 238 outpatient physicians across 14 specialties received one of two scribes or usual care. One scribe produced a modest reduction in measured time writing notes compared with the control group; the other did not show a significant difference on that measure. Some survey outcomes suggested improvements in the experience of work, with confirmation in larger studies still needed.
The trial also exposed a measurement boundary. Its electronic health record metric did not capture editing inside the scribe platforms. The authors warned that this could overstate the recorded saving. The study was short, conducted at one institution and restricted to English-language encounters. It cannot tell every health system which current tool will save time.
For the implementation lead, the practical implication is to inspect the whole route to an acceptable note. A faster draft can coexist with a slower review. A clinician may also value being able to listen more comfortably even when the recorded time difference is small. Keep those outcomes distinguishable and ask about both.
Check where the remaining effort goes:
- Preparation and capture. Starting the correct encounter, following the approved consent process and managing interruptions.
- Factual review. Checking what happened, who said it, relevant omissions and statements that the encounter does not support.
- Clinical structure. Making the assessment and plan express the clinician's intended meaning, rather than merely improving the wording.
- Completion. Moving the accepted material into the record and finishing any remaining documentation through the approved process.
A polished draft does not remove the second step. In an October 2026 LinkedIn reflection, Stephon Proctor, an associate chief health informatics officer at Children's Hospital of Philadelphia, described copying forward less content while using ambient scribes. He also questioned whether newly generated wording could make old errors harder to recognize. This was a practitioner reflection, not an audited improvement in note quality. It illustrates why fluent new text and trustworthy current information need separate scrutiny.
Walk through the last encounter before prescribing a fix
Suppose Claire Bennett leads an ambulatory scribe rollout and notices that Sam Taylor, a physician, is using it less. Instead of sending a refresher invitation immediately, she asks Sam to walk through a suitable recent encounter under the health system's approved privacy arrangements. They can examine the work without copying patient information into an adoption spreadsheet.
The same dashboard signal could lead to very different findings.
Sam’s lower use leaves two different questions open
Suppose Claire sees fewer ambient notes in Sam’s ambulatory clinic. The count alone cannot distinguish these explanations.
Fewer suitable encounters
Sam’s clinic mix changed. Claire checks the eligible visits and documentation responsibilities before proposing any habit repair.
The same visits need repeated rewriting
Sam still has suitable encounters, but repeatedly rebuilds the assessment. Claire investigates the draft and correction work before proposing a template change or retrial.
If Sam's clinic mix changed, there may be no failed habit to repair. If the same suitable encounters remain but Sam repeatedly rebuilds the assessment, more reminders will not remove that work. Claire needs to understand what the draft leaves wrong, what customization is possible and what must be escalated to the clinical or product owner.
Use a short conversation anchored in actual work:
- Locate the change. Ask when use fell and what changed in visits, responsibilities, access or the tool around that time.
- Follow one attempt. Ask what Sam expected, what the draft did, which checks and corrections followed, and whether manual completion was necessary.
- Name the obstacle precisely. Distinguish a login interruption, an unsuitable template, missing information, unacceptable review demands and a decision not to record an encounter.
- Agree one next action. Assign the specific issue to someone who can resolve it. Offer another supported trial only if the clinical, privacy and operational conditions permit it.
Record the category, owner and agreed follow-up, with only the information needed for the implementation decision. Explain that the purpose is to improve the workflow. Avoid public clinician rankings or an expectation that someone must defend every encounter they choose not to record.
If there is a possible documentation safety problem, use the health system's established clinical escalation process. An adoption interview must not substitute for that response, and a usage target must not override a patient's choice or a restriction on recording.
Make the follow-up answer the original problem
The next check should match the reason for stopping:
- Access failed. Confirm that the clinician can complete the authorized workflow.
- The template needed changing. Ask whether the relevant rewriting decreased on comparable encounters.
- Recording was inappropriate. Retain a supported alternative rather than treating nonuse as a defect.
Where a limited retrial is appropriate, agree its encounter scope, available support, review expectations and the conditions for stopping before it begins. Compare the clinician's experience with the earlier problem. Do not generalize a successful demonstration to every specialty or interpret an increase in recordings as proof of safer documentation.
Two existing methods can help with the follow-up. Use total human effort through accepted work when the claim is that the workflow takes less time. Use explicit output acceptance criteria when the concern is what the draft omits or changes. Clinical owners still need to define the appropriate standards for their records; a general business rubric does not validate a clinical note.
The useful outcome may be sustained use, selective use or a supported decision to stop. What matters is that the organization understands the reason and acts on the part it can responsibly change.
Questions about sustained ambient documentation use
Does stopping an ambient scribe mean a clinician resists AI?
Stopping alone does not establish resistance. A clinician may have fewer appropriate encounters, encounter technical problems or spend too much effort correcting the draft. Selective use may also reflect a sensible distinction between note types. Ask about the last actual use and the work that followed before choosing training or another intervention. The initial distinctions help separate these situations.
Should a health system set a minimum ambient-scribe usage target?
A raw minimum can reward recordings without establishing useful or appropriate documentation. First define eligible opportunities, approved encounter scope and the outcomes the workflow must preserve. Examine use alongside review effort, note quality and clinician and patient experience. If a use measure is retained, explain its denominator and exceptions, and do not let it override consent or clinical judgment about an encounter.
What if clinicians feel better but measured note time barely changes?
Treat reported experience and measured time as different outcomes. A clinician may feel less mentally burdened or more able to attend to the patient without a large reduction in recorded note-writing time. Conversely, that metric may miss editing outside the health record. Investigate both observations and their measurement limits before claiming a saving or dismissing the experience. The whole-workflow review explains the distinction.
What should happen when a patient declines recording?
Follow the health system's applicable consent, privacy and documentation arrangements and use an appropriate alternative. Patient refusal is not a clinician adoption failure. Keep the resulting nonuse distinguishable from technical failure or a clinician abandoning the tool. Resolve uncertainty about recording with the responsible local clinical and privacy owners before starting, rather than applying a universal consent rule from another jurisdiction.
What is a useful first step when clinicians abandon the tool?
Offer a willing clinician a supported walkthrough of one suitable recent attempt. Identify the expected result, actual corrections and any manual fallback, using approved privacy arrangements. Record a specific obstacle and give its follow-up to someone with authority to address it. A bounded retrial can test a proposed fix when use remains appropriate; reminders or more training should follow a diagnosed need. The encounter walkthrough supplies the questions.
