Health systems launch clinical transformation initiatives with genuine conviction. The strategic rationale is usually clear: improve care quality, reduce unwarranted clinical variation, close gaps in care coordination, or position the organization for risk-based contracting. The investment is authorized. The implementation partner is engaged. The steering committee is constituted. And then, somewhere between the kickoff presentation and the first operational review, the initiative runs into something that was never on the project plan: the organization itself.
Physicians who were not meaningfully engaged before the work began resist workflows they had no voice in designing. Operational leaders who received the initiative as a directive rather than a shared problem to solve implement it defensively, protecting their departments from disruption rather than driving adoption. Frontline staff who experience the change as something being done to them rather than with them find workarounds that preserve the familiar, undermining the new.
The pattern is consistent enough across health systems of every size and configuration that it deserves a direct name: clinical transformation initiatives fail most often not because the clinical model was wrong, but because the organizational conditions required to sustain change were never established before the work began.
This is a pattern confirmed by leaders across the industry. A 2026 Becker’s Hospital Review survey of 35 healthcare executives found a consistent theme: technology rarely fails on its own. Change management, clinician trust, workflow integration, and organizational readiness are almost always differentiating factors. As one respondent put it, “AI adoption improves when organizations treat it less as a technology rollout and more as a workforce and workflow transformation.”
This blog is written for executive leaders, including the CMIO, CMO, Chief Clinical Officer, COO, and Chief Ambulatory Officer, who are either preparing to launch a clinical transformation or already in one and recognizing that the organizational conditions are not what the project plan assumed. The goal is not to add complexity to an already demanding initiative. It is to identify the specific investments that determine whether a clinical transformation delivers its intended value or becomes another entry in a long list of initiatives that generate activity without lasting change.
The Framing That Shapes Everything: Transformation vs. Implementation
The distinction between clinical transformation and clinical implementation sounds semantic. It is not. It determines the governance model, the resource investment, the leadership behaviors required, and ultimately whether the work produces the outcomes it was designed to achieve.
Clinical implementation is the deployment of a defined solution into an existing organizational context. The solution is known. The technical and operational requirements can be specified. Progress can be measured against a plan. Implementation is appropriate for operational initiatives with well-defined parameters: deploying a specific clinical protocol, launching a new service line, or configuring a technology platform.
Clinical transformation is something fundamentally different. It involves changing how clinicians think and work: the mental models, professional habits, team dynamics, and organizational incentives that shape clinical behavior at scale. The solution is not fully known at the outset. The path from the current state to the desired state requires iteration, adjustment, and sustained leadership engagement. Progress cannot be fully measured against a project timeline because the changes being sought are cultural and behavioral, not just operational.
Most clinical transformation initiatives are framed and governed as implementations. The project plan has milestones and deliverables. The steering committee reviews status reports. The go-live date anchors the organization’s sense of when the work is done. And when the go-live date arrives, and the underlying clinical behaviors have not changed, the organization declares success and moves on, leaving the transformation it launched still waiting to happen.
The digital health graveyard is full of initiatives that succeeded at implementation and failed at transformation. A Memorial Hermann Health System analytics leader reflected that “we measured deployment and usage; we should have measured decisions.” Their analytics products had users, not owners. Tracking logins and usage rather than outcomes meant the work was considered done at delivery. The lesson: “Delivered is not done. Decided is.”
The go-live date is not the end of a clinical transformation. It is the moment when real work becomes possible, because the new operational context now exists to support different clinical behavior. Organizations that understand this design their post-go-live investment accordingly.
What the CMO and Chief Clinical Officer Must Own Before Anyone Else Is Engaged
Clinical transformation requires clinical ownership. This is understood in principle by most health system executive teams. What is less consistently understood is what clinical ownership demands of the most senior clinical leaders, and how early that demand begins.
The CMO and Chief Clinical Officer are the executives who determine whether a clinical transformation initiative is experienced by the medical staff as a leadership-led change they are being invited into or as an administratively driven mandate they are being subjected to. That distinction is established in the first conversations the medical staff has about the initiative, before any operational details are defined, and shapes every subsequent engagement dynamic.
The physician engagement investment that most initiatives underfund
Meaningful physician engagement in clinical transformation is not a communication plan. It is a design process. Physicians who are engaged as designers of the change, who participate in defining the clinical model, identifying the barriers to adoption, and shaping the workflows that will govern their practice, develop ownership of the outcome. Physicians who receive communications about a change designed by others develop, at best, compliance.
The investment required to move from communication to design engagement is real. It requires structured forums that go beyond town halls and department briefings: working groups with protected time, clear decision rights, and genuine influence over outcomes. It requires identifying the clinical champions who carry credibility with their peers and investing in their leadership development alongside the clinical content of the transformation. And it requires the CMO and Chief Clinical Officer to be personally and visibly present in those forums, not as presenters of a pre-determined plan, but as partners in a shared problem-solving process.
The evidence from the field reinforces this. At Jefferson Health, leadership formalized the connection between clinical strategy and operational execution through eight “SWAT teams” (Synchronizing Workflows and Technology), each staffed with clinical, informatics, and IT leads operating on a standing biweekly cadence. As Jefferson’s Dr. Judd Hollander summarized: “It really is all about the workflows and operations and not the technology.” Health systems that make this structural investment before implementation begins consistently report faster adoption, lower resistance, and more sustainable behavioral change.
What Genuine Physician Co-Design Looks Like, and How to Run It
The gap between co-design and what organizations usually call co-design: town halls, department briefings, and online surveys, is the single largest organizational change management failure mode in clinical transformation. Town hall theater produces the appearance of engagement and the reality of resentment. Physicians who sit through a presentation of a pre-designed care model and are then asked for feedback understand immediately that the decisions have already been made.
Genuine physician co-design has a specific structure. It is not a meeting. It is a repeating working session series, conducted before the clinical model is finalized, with the following characteristics:
- Small groups (6–10 physicians), organized by clinical domain or care setting, not by seniority or committee membership. Frontline clinicians who will actually use the new workflows must be in the room.
- A current-state problem framing that precedes any solution discussion. The first session in a co-design series should be entirely diagnostic: what does not work about how we deliver this type of care today, in your experience? This question, answered by clinicians before anyone presents a proposed solution, generates the clinical problem ownership that makes subsequent solution design feel like professional problem-solving rather than administrative compliance.
- Explicit decision rights stated at the outset. Clinicians need to know which elements of the clinical model are truly open to design input and which are constrained by regulatory, contractual, or system-wide requirements. Ambiguity about decision rights is the fastest path to cynicism when the final model differs from what was discussed.
- A visible record of every design decision, including who made it and why alternatives were considered or rejected. This “decision log” serves as the evidentiary foundation for physician trust in the process, and as the institutional memory that sustains the model when the original design participants are no longer in the room.
- A closing “so what” session that translates design decisions into specific workflow implications for each participating role. Physicians who leave a co-design session without a clear picture of how their daily practice will change, concretely and step by step, will fill that gap with assumptions and anxiety.
A well-run co-design series for a single clinical domain typically requires three to five sessions over six to eight weeks, with three to five hours of physician time per session. This is the investment that separates clinical transformation initiatives that achieve durable adoption from those that achieve temporary compliance.
The clinical variation conversation that must happen at the leadership level
Most clinical transformation initiatives involve, at some level, an effort to reduce unwarranted clinical variation: standardizing care pathways, aligning practice patterns across facilities, or implementing evidence-based protocols that replace individually variable approaches. This is clinically appropriate and operationally necessary. It is also organizationally complex because it directly implicates physicians’ professional identity and autonomy in ways that other operational changes do not.
The CMO who launches a clinical variation reduction initiative without first engaging the medical staff in the evidence base, and without creating forums for clinicians to distinguish warranted variation that reflects appropriate clinical judgment from unwarranted variation that reflects habit, preference, or system dysfunction, will encounter resistance that no amount of implementation methodology can overcome.
The conversation that must happen at the leadership level, before the initiative is operationalized, is an honest and data-driven examination of where the organization’s clinical practice patterns diverge from the evidence and from peer benchmarks, and why. That conversation is the foundation on which physician ownership of the transformation is built. Without it, the transformation is an imposition. With it, it is a shared professional commitment.
What the COO and Chief Ambulatory Officer Need to Solve that the Clinical Model Cannot
Clinical transformation initiatives are designed by clinical leaders and implemented through clinical workflows. But their operational context, including the scheduling systems, the staffing models, the facility configurations, and the access structures, is owned by operational leaders. When the operational context is not redesigned in parallel with the clinical model, the clinical transformation operates in an environment that was built for a different way of working.
The result is predictable: clinicians who are willing to work differently cannot, because the operational infrastructure that would support the new model is not in place. Physicians who were trained in the new approach revert to old patterns because the old patterns are what the operational environment rewards.
Ambulatory transformation and the access redesign it requires
Clinical transformation in ambulatory settings represents the majority of patient touchpoints for most integrated delivery networks, which are the primary arena for population health management, chronic disease management, and value-based care performance, and cannot be separated from access redesign.
The care model changes that ambulatory clinical transformation typically pursues, including team-based care, proactive outreach, care gap closure, and care coordination across specialties, require fundamentally different scheduling architectures, staffing ratios, and visit type structures than the traditional physician-centric visit model was built around. A clinic that is scheduling 15-minute follow-up visits for a physician-driven care model cannot execute a team-based care transformation without redesigning how its schedule works.
The urgency of this redesign is amplified by broader demographic and financial pressures. With 61.2 million Americans now aged 65 and older, a figure projected to reach 72 million by 2030, and 90% of the CDC’s estimated $4.9 trillion in annual healthcare expenditures tied to patients with chronic conditions, the ambulatory system is bearing increasing structural load. Organizations that redesign ambulatory operations in parallel with their clinical models are better positioned to absorb this demand without compromising care quality or financial performance.
The Chief Ambulatory Officer who treats clinical transformation as a clinical protocol deployment without addressing the operational infrastructure in which those protocols will be executed is setting the transformation up for the most common form of failure: the initiative that succeeds in training and fails in practice.
Inpatient operational redesign and the throughput implications
In inpatient settings, clinical transformation initiatives, whether focused on care pathway standardization, length of stay management, discharge planning, or interdisciplinary rounding models, have direct and measurable implications for throughput, capacity utilization, and operational flow. The COO who understands the operational design requirements of the clinical transformation, and who invests in redesigning the operational systems that the new clinical model depends on, becomes the transformation’s most important operational enabler.
This requires a level of collaboration between the clinical and operational leadership teams that most health systems have not institutionalized. Clinical leaders design the care model. Operational leaders design the environment in which it will be executed. When these two design processes happen in parallel, with genuine coordination, the result is a clinical transformation that works in practice. When they happen sequentially, the clinical model first and the operational adaptation second, the result is a clinical model deployed into an operational environment that was never designed to support it.
The CMIO’s Role: Building the Data Foundation That Clinical Transformation Requires
Clinical transformation initiatives generate an operational dependency on data that most organizations underestimate at the outset. The care models being implemented, whether standardized clinical pathways, population health management protocols, or value-based care programs, require clinicians and leaders to have access to timely, accurate, actionable information at the point of care and in the operational decision-making process.
When that information is not available, whether because the EHR is not configured to surface it, because the analytics infrastructure to generate it does not exist, or because the data that feeds it is not reliable, clinicians cannot practice the new model effectively even when they want to.
Clinical decision support that enables rather than interrupts
The CMIO’s most consequential contribution to clinical transformation is the design of a clinical decision support environment that enables the new care model rather than creating alert fatigue and workflow friction. At Johns Hopkins Medicine, the CMIO noted that autonomous AI tools in the EHR have broadly underdelivered relative to their promise: specifically, AI-assisted patient messaging and automatic inbox routing require so much provider oversight that net time savings are marginal. The contrast with ambient AI documentation is instructive: that category delivers real, measurable time savings precisely because it augments what physicians do naturally rather than asking them to supervise processes not yet reliable enough to trust.
As Qventus CMO Jason Cohen, MD, put it: “The gap today is not insight. As a clinician, please do not give me another alert, but start automating the next step in the workflow for me.” A sepsis alert that stops at the notification is a half-built solution. A sepsis alert that surfaces the missing lactate order, flags the chest X-ray finding, and asks the clinician if they want to initiate the order. That is clinical decision support that enables rather than interrupts.
Prerequisites before deploying AI-enabled clinical decision support
The design principle above, that CDS enables workflow completion rather than surfacing information, is only achievable when the underlying data infrastructure meets a set of non-negotiable prerequisites. Deploying AI-enabled clinical decision support into an environment that does not meet these standards does not produce the enabling CDS described above. It produces the alert fatigue and clinician disengagement that undermines the transformation.
The CMIO should treat the following as an operational checklist before any AI-enabled CDS goes live:
- EHR data completeness and timeliness standards. Define and measure the percentage of structured data fields required for each CDS use case that are populated in real time: vital signs, medication lists, problem lists, and lab results. Set a threshold (typically >95% completeness and <4-hour data lag for inpatient use cases) before activating AI-enabled alerts. Alert a clinician about a missing lactate only when the system can confirm the order has not already been placed.
- Algorithm transparency documentation. Before clinical deployment, each AI tool must have a one-page clinical summary stating: what data the model uses, what population it was validated on, what its false positive and false negative rates are at the thresholds being deployed, and what the intended clinical action is. This document is reviewed and signed off by the relevant clinical champion before go-live.
- Physician validation cohort. Run the AI model in “silent mode” for 30–60 days before activating clinician-facing alerts. During this period, a designated physician review group receives the model’s outputs in a dashboard (not as live alerts) and assesses clinical plausibility against their own judgment. Discordance rate greater than 15% triggers model recalibration before live deployment.
- Alert burden baseline and cap. Measure the current alert burden per physician per shift before adding new CDS. Establish an organizational cap on total interruptive alerts per physician per shift (a commonly cited benchmark is no more than 5–8 interruptive alerts per shift). New CDS tools may not go live if they push any physician category above the cap without a corresponding retirement of lower-priority alerts.
- Post-live override tracking and review. Track override rates for each AI-enabled alert from day one. An override rate above 70% is a signal that the alert is firing inappropriately or that the recommended action does not fit the workflow context. The Clinical Informatics Working Group reviews override rates at every biweekly meeting during the first 90 days of a new CDS deployment.
These prerequisites add time to CDS deployment. They also eliminate the three most common failure modes of AI-enabled clinical transformation: clinicians ignoring alerts because they fire too frequently, clinicians distrusting alerts because the underlying data is unreliable, and alerts triggering at workflow moments where the recommended action cannot actually be taken.
Analytics that drive accountability without creating surveillance
Clinical transformation requires performance measurement. Clinicians and operational leaders need to know whether the new care model is working, where it is not being adopted, and why. But the way performance data is designed, presented, and used in the governance of the transformation is itself a leadership decision with significant behavioral consequences.
Performance data that is experienced as surveillance, including individual physician scorecards deployed without context, benchmarking that is punitive rather than developmental, and metrics that measure activity without accounting for patient complexity or case mix, generates defensiveness and disengagement. Performance data that is experienced as a shared learning tool generates the reflective practice that drives sustainable behavioral change.
The CMIO who designs the analytics infrastructure for clinical transformation is making choices that shape the organizational culture of the initiative. As the Memorial Hermann experience illustrates, the shift from tracking usage to tracking decision velocity, meaning how fast an insight moves to an action, changes not just what is measured, but how the entire initiative is built and governed.
The Governance Model That Sustains Transformation Beyond Go-Live
Clinical transformation does not have a natural endpoint. The clinical model will need to evolve as evidence develops, as patient population characteristics change, and as operational experience reveals what works in practice and what requires adjustment. The governance model for a clinical transformation must be designed to sustain this continuous improvement orientation, not to close out a project.
Industry surveys find that while 85% of health systems are now using AI internally, only 17% have mature governance structures in place. The organizations moving fastest and most safely are those that have defined the rules of engagement before scaling, not as an afterthought.
The four governance bodies that sustain clinical transformation
Governance principles without a governance structure are not implementable. The following four-body architecture represents the operational design that sustains clinical transformation from launch through embedding. Each body has a defined membership, meeting cadence, and decision scope. Executive leaders should treat this as a starting framework, adjusting membership titles to reflect their organizational structure.
A few design principles that make this architecture work in practice:
- The Executive Steering Committee should not review operational details. Its role is strategic direction, resource allocation, and escalation resolution. Operational problem-solving that lands in the ESC is a signal that the Clinical Operations Council is not functioning as designed.
- The Clinical Informatics & Analytics Working Group is the body that makes the CMIO’s data requirements operational. It owns the alert burden baseline, the override rate tracking, the algorithm transparency documentation, and the AI validation cohort process. It should have a standing agenda item for CDS performance at every meeting.
- Frontline Implementation Teams are the most frequently skipped governance investment. They are also the most consequential for adoption. The clinical champion on each team is the person who knows when a workflow redesign is not working before it shows up in the performance data, and who has the peer credibility to address it before it becomes a pattern.
- Governance cadence should match the initiative phase. During active design and build, all four bodies meet at their full frequency. During the post-go-live stabilization period (typically months 1–6 post-launch), increase Frontline Team cadence to twice weekly. During the embedding phase (months 12–24), ESC can shift to quarterly while Clinical Operations Council maintains a biweekly cadence.
CMIO and CDAO Roles in the Clinical Informatics Working Group
Many health systems now have both a CMIO and a Chief Data and Analytics Officer (CDAO), and the boundary between their governance roles in clinical transformation is a common source of friction that the architecture above can inadvertently obscure. The practical distinction is this: the CMIO owns clinical workflow integrity and CDS design, specifically the questions of what information clinicians need at what workflow moment and whether AI tools are safe to deploy in clinical contexts. The CDAO owns the data platform, the analytics infrastructure, and the enterprise data governance that makes the CMIO’s CDS work possible. In the Clinical Informatics & Analytics Working Group, the recommended model is CMIO-chaired with CDAO as a standing co-chair or permanent voting member. CDS deployment decisions (go/no-go, alert thresholds, override rate response) are CMIO calls. Analytics platform architecture, data quality standards, and enterprise metric definitions are CDAO decisions. Both require the other’s concurrence on anything that crosses the boundary. Health systems that have not clarified this boundary typically find that the Working Group becomes a venue for unresolved executive jurisdiction disputes rather than a functional governance body, which is the single most common reason the four-body architecture breaks down in practice.
Decision rights, tie-breaking, and phase-transition triggers
Three governance design questions that the table above does not answer on its own:
- Tie-breaking in the Clinical Operations Council. The COC is co-chaired by the CMO and COO. When clinical and operational perspectives produce a deadlock on a care model or workflow design decision, the tiebreaker is the executive whose domain bears the primary implementation risk: clinical model decisions (pathway content, evidence thresholds, order set structure) break in favor of the CMO; operational design decisions (staffing ratios, scheduling architecture, facility configuration) break in favor of the COO. Decisions that cross both domains escalate to the Executive Steering Committee within one meeting cycle.
- Phase-transition triggers. The governance architecture has three phases: active design and build, post-go-live stabilization, and embedding, with different meeting cadences for each. Phase transitions are not calendar-driven; they are condition-driven. The transition from design and build to stabilization is triggered when 80% of planned workflows have achieved go-live and frontline adoption is being tracked at the department level. The transition from stabilization to embedding is triggered when sustained adoption rates exceed 70% across the primary physician cohort for two consecutive measurement periods (typically quarters) without requiring active intervention from Frontline Implementation Teams.
- Sunset criteria. The governance architecture is funded for 24 months post go-live, but it does not automatically dissolve at month 24. The condition for dissolving the Frontline Implementation Teams is that the clinical champion network has been incorporated into standing departmental leadership structures, and the onboarding process reflects the transformed care model. The condition for dissolving the Clinical Operations Council as a distinct body is that transformation performance metrics have been incorporated into existing quality and operations committee agendas with the same reporting cadence. Governance dissolution without condition fulfillment is the organizational equivalent of declaring victory before the battle is won.
How Long Does Transformation Take, and What Does the Evidence Say?
The governance model must be designed for the actual duration of clinical transformation, not for the duration of the implementation project. Most importantly, they are led by executives who stay engaged after go-live.
How long is that? The honest answer is that the research literature on clinical transformation duration is thinner than practitioners would like, and the timelines vary substantially by initiative scope, organizational size, and the depth of pre-launch investment in organizational readiness. What literature does support:
- Studies of EHR-enabled clinical pathway implementation published in journals including the Journal of the American Medical Informatics Association and Implementation Science consistently find that significant practice pattern change, measured as sustained adoption rates above 70% across a physician cohort, requires 12–24 months post-deployment, with the wide range explained primarily by the depth of pre-launch physician engagement.
- Kotter’s research on large-scale organizational change, while not healthcare-specific, identifies “consolidating gains and producing more change” as a phase that typically requires 24–36 months in complex professional organizations, a benchmark that clinical transformation practitioners frequently cite.
- Health system case studies reviewed in the NEJM Catalyst series on care model transformation consistently describe an 18–24 month window between go-live and the point at which the new model no longer requires active organizational support to sustain, a finding that reflects practitioner consensus rather than controlled study.
The 18–24 month framing should therefore be understood as practitioner-validated guidance rather than published empirical findings. Organizations that plan for less are consistently surprised. Organizations that plan more consistently outperform. The governance architecture described above should be funded and staffed for a minimum of 24 months post go-live.
The most common governance failure in clinical transformation is the disengagement of senior leadership once the implementation phase is complete, conveying the tacit organizational message that the initiative is finished. Clinical transformation requires sustained leadership presence and organizational investment for as long as it takes the new model to become the default way of working.
Clinical transformation is the most demanding and most consequential strategic initiative that most health system executive teams will lead. The organizations that succeed are not the ones with the best clinical model or the most sophisticated implementation methodology. They are the ones whose executive leadership understood, before the work began, that they were signing up to lead a sustained organizational change, not deploy a clinical program.
What This Means for the People Doing the Work
The preceding sections are written for executive leaders. But clinical transformation is executed by practitioners: the clinical informatics analysts building the CDS architecture, the data product managers designing the performance measurement infrastructure, and the care coordinators managing the patient-facing workflows on which the transformation depends. What changes for them under a well-designed transformation model?
When the governance model does not exist yet: what practitioners can do
The practitioner roles described above assume a functioning governance architecture. Many practitioners reading this will be working in organizations where architecture is absent or only partially built, where there is no Clinical Informatics Working Group with a standing override rate agenda item, no Frontline Implementation Team to escalate barriers to, and no co-design process that resembles the one described above. If that is your situation, the most useful thing you can do is make the gap visible in terms that executive leadership can act on. The language that tends to move governance conversations forward is concrete and operational, not abstract: “We have three active CDS rules with override rates above 70%, and no standing forum to review them: can we add a 30-minute standing item to the informatics leadership meeting?” or “I have escalated the same scheduling access barrier for four consecutive weeks with no response; we need a named owner and a response time commitment or the care coordination protocol will not hold.” Governance gaps surfaced as specific operational failures with named consequences are more actionable than governance gaps surfaced as abstract structural critiques. You are not asking for a reorganization. You are asking for the minimum structure needed to do the work the transformation requires.
