This article is part of the MiiWrap Institute’s series, Closing the Sustainability Gap: Building Human Services That Create Change Beyond Services. Read Part 1 here.

Organizations often talk about implementation as though it is something they eventually finish. A model is selected. Staff are trained. Supervisors learn what to expect. New tools and procedures are introduced. For a period of time, there is considerable attention to doing the work correctly. Eventually, the organization begins to think of the model as implemented.

But implementation does not remain stable simply because the initial work was done well. The organization keeps changing. Staff leave. New staff arrive. Caseloads increase. Supervisors take on additional responsibilities. Funding requirements shift. Documentation changes. Difficult cases expose parts of the model that are harder to use than they appeared in training.

Practitioners adapt to those realities, usually in small and reasonable ways. Conversations get shorter. Preparation becomes less thorough. A difficult practice is replaced with something more familiar. New practitioners learn partly from experienced staff who have already adapted the model themselves.

None of this requires anyone to decide to abandon the model. It is simply what happens when a complex practice encounters the conditions of everyday work.

That is why implementation is not an event. It is the infrastructure an organization builds to continually pull practice toward what it is intended to be.

Organizations are always reinforcing something

Every organization teaches practitioners how to work long after formal training ends. It does this through supervision, coaching, productivity expectations, documentation, performance reviews, case consultation, leadership decisions, and the informal norms practitioners observe around them.

Those signals may reinforce the practice model. They may also gradually teach something different.

A practitioner can be trained to slow down enough to understand what matters to a family, while simultaneously learning that successful employees keep meetings short and cases moving. They can be coached to transfer decision-making to clients while supervision focuses primarily on whether the practitioner got the plan completed. They can learn that building natural support is essential to sustainable change while workload expectations make it easier to make another professional referral.

None of those operational expectations is necessarily unreasonable. Organizations have real responsibilities for productivity, safety, documentation, timeliness, and accountability. The implementation question is whether those expectations have been designed to coexist with the practice the organization says it wants.

When they are not aligned, practitioners receive two different messages about what good work looks like. Over time, the message reinforced every day is usually the one that survives.

This is one reason implementation science treats organizational context as central rather than viewing adoption and fidelity solely as practitioner problems. Implementation strategies are intended to support not only adoption but continued implementation and sustainment, and research increasingly examines combinations of educational, feedback, facilitation, and organizational strategies rather than assuming one intervention will carry the work by itself. A 2024 systematic review of 129 experimental studies across health and human service settings found that implementation efforts used an average of nearly seven strategies, with educational meetings, audit and feedback, and facilitation among the most commonly tested approaches.

Fidelity tells us whether the model is surviving

Fidelity has an essential role in this infrastructure because organizations cannot strengthen practice they cannot see. But fidelity is sometimes expected to do more than measurement can accomplish.

An observation or assessment can tell us that practice is drifting. It cannot, by itself, reverse the drift. It can show that practitioners are consistently struggling with an important part of the model. It cannot tell us whether the solution is additional practice, stronger coaching, clearer training, better supervision, or addressing an organizational barrier that makes the expected practice unusually difficult. Measurement tells us where practice is. Implementation determines what happens next.

There is another important limitation. In complex relational models, fidelity cannot be reduced to whether practitioners completed the required activities. Procedures matter, but the purpose behind them matters too.

A practitioner can ask every required question without creating meaningful engagement. They can complete a planning process while professionals remain firmly in control of the decisions. They can follow the sequence of a model while missing the mechanisms that are supposed to make that sequence useful.

This is why strong fidelity systems have to preserve both consistency and professional judgment. Practitioners need to understand what is essential to the model, why it matters, and what can appropriately be adapted to the person, culture, setting, or circumstance in front of them.

The purpose of fidelity is not perfect replication. It is to protect the connection between the practice an organization claims to provide and the experience clients actually receive.

Fidelity information has to lead somewhere

Part 4 of this series focused on the developmental process practitioners need in order to change their practice. That process becomes part of the implementation infrastructure when observation, feedback, coaching, and reassessment are connected.

A fidelity observation might reveal that a practitioner consistently moves toward solutions before fully understanding what matters to the client. Recording that finding is useful. Giving the practitioner a score is useful. Neither is sufficient.

Development happens when the practitioner can examine what occurred, understand why that moment was difficult, practice another way of responding, apply it in real work, and receive feedback again.

Research on audit and feedback supports the usefulness of this general approach while also showing its limits. Feedback can improve professional practice, but effects vary, and feedback is more useful when people have support to interpret it and determine what to do differently. More recent work examining feedback facilitation similarly emphasizes helping recipients identify barriers, select improvement actions, and translate performance information into change.

The implication for organizations is straightforward: fidelity measurement should feed practitioner development, not sit beside it.

Coaching and supervision have to reinforce the same practice

Coaching is one place that development happens, but coaching cannot sustain a model if the rest of the organization continually pulls practitioners in another direction.

Coaching and supervision have different responsibilities. Coaching is primarily developmental. It helps practitioners examine practice, build skill, strengthen judgment, and learn from difficult situations. Supervision carries broader organizational responsibilities, including safety, documentation, productivity, compliance, workload, and employee performance. Organizations need both.

The problem arises when the two functions communicate different definitions of success.

A coach may help a practitioner learn how to create more client ownership. If a supervisor consistently rewards getting decisions made quickly, the practitioner still has to choose which expectation carries more weight. A coach may encourage thoughtful preparation and reflection, but that practice becomes difficult to sustain if workload expectations leave no realistic room for it.

This is not a coaching problem. It is an implementation problem. Practitioners should not have to choose between practicing the model well and being viewed as successful employees. Leadership has to make those things as compatible as possible.

Implementation has to anticipate drift

Organizations sometimes treat drift as evidence that practitioners have become less committed. More often, drift is much less dramatic.

People adapt. A practitioner finds a shortcut because the caseload has increased. A supervisor stops focusing on one part of the model because several new administrative requirements need attention. New staff learn from coworkers who have gradually modified their own practice. What began as small adjustments becomes ordinary practice.

Over time, the organization can continue using the language of the original model while delivering something meaningfully different. That is why periodic retraining is rarely an adequate sustainment strategy.

The goal is not to prevent every adaptation. Adaptation is unavoidable and often necessary. The goal is to maintain enough shared understanding and visibility that organizations can distinguish thoughtful adaptation from gradual erosion of the model. Implementation infrastructure provides that correction.

Observation reveals what people are actually doing. Coaching helps practitioners examine and strengthen it. Supervision keeps daily organizational expectations aligned. Certification establishes what competent practice means. Ongoing learning and recalibration keep experienced practitioners from being treated as permanently finished products.

The system continually asks whether the practice being delivered today is still the practice the organization intended to implement.

Certification should establish capability, not attendance

Certification matters for the same reason. If certification means only that someone completed training, it tells us very little about whether the person can actually perform complex work to an expected standard. Meaningful certification establishes that a practitioner has demonstrated capability in practice.

But certification should not create the illusion that competence is permanent. Practitioners continue learning, organizational conditions change, and practice can drift. For complex models, some form of ongoing observation, professional development, and recalibration is therefore part of maintaining quality rather than an admission that the original certification failed.

This is particularly important in organizations with turnover. New staff should not inherit whatever version of the model happens to be practiced by the person sitting next to them. The organization needs enough internal capacity to teach, observe, coach, and interpret the model consistently across generations of staff.

If implementation depends almost entirely on one exceptional coach, supervisor, or champion, the model remains vulnerable.

Organizations need places to learn collectively

Not every implementation problem belongs to an individual practitioner. If one practitioner struggles with a particular skill, that may be a coaching issue. If nearly every practitioner struggles with it, leaders should ask what the organization is producing.

Perhaps the training is unclear. Perhaps supervisors need more support in reinforcing the practice. Perhaps a workflow makes the expected behavior difficult. Perhaps staff have developed an adaptation that needs to be examined more carefully. Fidelity and practice data become much more useful when organizations look for these patterns.

Shared learning structures also matter here. Communities of practice, calibration conversations, team learning, and cross-site learning give practitioners and implementation leaders places to compare experience and develop shared interpretations of difficult work.

The evidence for communities of practice as a standalone implementation strategy is less developed than the evidence for some other approaches, so they should not be treated as a solution by themselves. Their value is as part of a larger learning system: they prevent practice knowledge from becoming isolated inside individual practitioners and create opportunities to notice how the model is evolving across people and settings.

The MiiWrap perspective

MiiWrap approaches implementation from the assumption that complex relational practice will not sustain itself. Training establishes the foundation, but practitioners are expected to develop the ability to use MiiWrap through practice, coaching, fidelity observation, feedback, and demonstrated competency. Certification reflects the ability to perform the work rather than simply exposure to the curriculum, and continued observation and recertification recognize that fidelity and competence require ongoing attention.

The same structure extends beyond practitioners. Coaches develop the capacity to observe practice, provide useful feedback, and help practitioners strengthen their judgment. Process Mentors develop Coaches and help organizations interpret fidelity and implementation patterns at a broader level.

The intention is not to create layers of oversight for their own sake. It is to distribute enough knowledge and capability throughout an organization that fidelity does not depend on individual memory or a small number of unusually committed people.

MiiWrap also treats supervision and organizational alignment as implementation issues. A practitioner cannot be coached toward one model of practice while their workplace consistently rewards another. Leadership expectations, supervision, coaching, fidelity standards, and workforce development need to point toward the same definition of good work.

Shared learning then provides another layer of reinforcement. Practitioners, Coaches, Process Mentors, and organizations can learn from experience beyond their own immediate environment, compare interpretations, and recognize patterns that would be harder to see in isolation.

Taken together, these elements create a continuous learning system around the practice.

Implementation is what allows good practice to survive reality

The central leadership question is therefore not: have we implemented the model? It is: what is our organization reinforcing now in regards to how we use the model?

What are practitioners learning from supervision? What does coaching develop? What does fidelity measurement reveal? What happens after a gap is identified? What do new staff learn when they enter? What happens when an experienced practitioner begins to drift? Is the organization learning from patterns across its workforce, or treating every fidelity problem as an individual performance issue?

Those questions acknowledge something important about complex human services: implementation is never standing still. The organization is continually shaping the practice.

Training provides practitioners with a model. Implementation determines what happens to that model over time.

That is why fidelity is not maintained by asking people to remember what they learned. It is maintained by building systems that continually make good practice visible, reinforce it, develop it, and bring people back toward it as reality inevitably pulls them away.

That infrastructure is what allows a practice model to become more than a training initiative. It is what allows it to become the way an organization actually works.

In Part 6: What MiiWrap Changes About the Work

A practice model without implementation eventually becomes theory. An implementation system without a meaningful practice model becomes process.

The final part of this series will bring those two sides together: what MiiWrap asks practitioners to do differently to create sustainable client change, and how its continuous learning system is designed to help organizations sustain that practice over time.

Research & Further Reading

Proctor, E., Silmere, H., Raghavan, R., et al. (2011). Outcomes for implementation research: Conceptual distinctions, measurement challenges, and research agenda. Administration and Policy in Mental Health and Mental Health Services Research, 38, 65–76.

Moussa, L., et al. (2024). A systematic review of experimentally tested implementation strategies across health and human service settings: Evidence from 2010–2022. Implementation Science.

Brown, B., et al. (2024). Exploring the content and delivery of feedback facilitation co-interventions: A systematic review. Implementation Science.

Baskerville, N. B., Liddy, C., & Hogg, W. (2012). Systematic review and meta-analysis of practice facilitation within primary care settings. Annals of Family Medicine, 10(1), 63–74.