Chapter 11: Design for Adoption and Organizational Change

The paper that survived the training

It is the third week after clinic one goes live, and Sam Otieno is walking the floor to see how the training held up. His records say it held up well. All forty care staff completed the platform course, thirty-eight passed the competency check, and the check was not easy: timed chart entries, a medication reconciliation, a simulated interruption. Ninety-five percent of clinic one is trained. Sam is proud of the number until he reaches the treatment room and finds Thelma, a nurse he trained by name, writing a patient encounter on paper while the platform terminal waits at the end of the corridor, still signed in.

“Paper,” Sam says. “After the check you passed?”

Thelma does not stop writing. “I passed the check with the machine in front of me. Here, look.” She nods at the corridor. “To chart on the platform I walk to the end of the hall, I go through two screens, I come back. It takes me three minutes I do not have. Between patients I have thirty seconds, sometimes less. This form is at my desk.” She finishes the line. “It wins.”

Sam starts to say something about the workflow standard, and Thelma cuts him off, not angry, just tired: “And Sam, when I am behind, I do not think about the platform at all. I think about the queue.” She looks at him the way someone looks at a colleague who is about to write the wrong conclusion. “Do not tell me I was trained and chose not to. I was trained and the clinic made me choose.”

That afternoon Nora Kariuki’s dashboard shows both numbers side by side: trained, 95 percent. Adopted, measured as the share of care staff whose last ten patient encounters went through the platform, 35 percent. The project reports the first number, and the number looks like a success, until someone asks why it differs from the second. That question is this chapter. Adoption is the behavior that turns an installed system into a realized benefit, and it is a design problem before it is a people problem. Clinic one was delivered on time, certified, and trained, and the benefit the grant pays for, access outcomes in six neighborhoods, depends on behavior that is not happening at the point of work. The Alignment lens and the Delivery lens meet here: Project Mastery = Judgment × Alignment × Delivery × Learning, and the multiplier that decides whether the delivered system becomes a delivered benefit is the change in what people do after go-live.

Installation is not adoption

The first discipline is vocabulary, because the two words are constantly treated as one. Installation is the state where the system is in place and the people can use it: the platform is configured, the interfaces are open, the staff are trained, the certification is passed. Adoption is the sustained change in behavior at the point of work: the clinician charts through the platform, the nurse checks the medication list before prescribing, the records clerk stops keeping the second book. Installation is about what the project produced. Adoption is about what the organization now does. The difference between the two states is where benefits are made or lost, and the difference is not measured by anything the project controls directly.

The two states get confused because the metrics for installation are beautiful. Trained, 95 percent: countable, owned, and finishes on time. The behavior metric is the opposite: noisy, lagging, owned by operations, and it only exists after go-live. The project measures what it can control and then reports it as what it achieved. That is the machine of the 95/35 gap: not a training failure, and not a people failure. A measurement failure. The field signal is a dashboard where “trained,” “certified,” and “ready” sit in the same column as “using,” as if passing a course and changing a practice were the same event.

The corrective frame is the output-to-benefit chain from chapter 1, and it does most of the work. The project produces outputs — the configured platform, the migrated records, the trained staff — that enable a capability, and the capability produces an outcome only when it is used. The chain breaks exactly where the behavior is supposed to happen, and nothing downstream of the break can compensate: a platform nobody uses is not a deliverable that fell short; it is a cost with no benefit attached. The decision this chapter improves is the allocation of the project’s attention and evidence: where do you put effort and proof so that behavior, not installation, becomes the unit of delivery?

Go-live is the midpoint of that chain, not the finish line. The project that treats go-live as the end is the project that celebrates the 95 percent, and the grant pays on the 35. The discipline is to design the adoption workstream with the same seriousness as the build, from the first gate, with owners, evidence, and consequences, because the work after go-live is where the value was always going to be made.

Today, tomorrow, and the awkward in-between

The second discipline is describing the change as a movement between three states, because the middle state is where the damage happens and almost nobody budgets for it.

The current state is the day as it actually runs, not as the poster describes it. At clinic one, the current state is a paper encounter form at the nurse’s desk, a paper queue at reception, a pharmacy that phones the doctor about refills, and a records clerk who files what everyone else writes. It is not a bad state. It works, after a fashion, and the people in it have tuned it: the form is placed where the hand lands, the queue is managed with eye contact, the pharmacist knows the doctors’ handwriting. The current state has momentum, and momentum is the first thing the project underestimates.

The future state is the intended day: encounters charted through the platform at the point of care, the queue visible to the whole clinic, refills routed, the records clerk auditing rather than filing. The future state is where the benefits live, and it is the only state the project designs in detail. That is the mistake: the organization does not jump between states. It passes through the transition state.

The transition state is the awkward in-between: the platform is up, so the records clerk stops filing; the clinicians are not fully on the platform, so the pharmacy sees neither a clean queue nor a clean paper trail; some encounters exist twice, once in the terminal and once in the form; the staff carry two systems in their heads and neither works well. This is where reversion happens, because the transition state punishes the new behavior: it is slower, more error-prone, and less trusted than the current state was. Thelma is not reverting from the future state to the current state. She is refusing to live in the transition state, where charting takes three extra minutes and the queue is long. Her choice is locally rational, and the project needs to hear that before it writes her off.

The three states have a respectable lineage. Kurt Lewin’s mid-century account of social change described unfreezing the current equilibrium, moving to a new one, and refreezing it into new norms; the crucial observation is that an existing way of working holds itself together and resists replacement the way a weight resists being lifted. The practitioner tradition that followed, notably William Bridges’s writing on transitions, sharpened the same idea: the psychological work of change begins with an ending, and people cannot arrive at the new beginning without passing through a neutral zone where the old rules no longer hold and the new ones are not yet trusted. This book uses its own words for it: today, tomorrow, and the awkward in-between. You do not design a change by describing the future state; you design it by designing the transition, and the transition is where the project’s promises are actually tested.

Two consequences follow. The first is the performance dip: when a new practice is adopted, performance falls before it rises. The first hundred chart entries are slower than the paper routine ever was, and the dip lasts weeks, not days. Michael Fullan gave this the name that stuck, the implementation dip, and it is a feature of real change, not a failure of it; the project that does not budget for it will punish its own adoption, because the targets in place during the dip are current-state targets that make the new behavior a losing choice. The second is decay: knowledge learned in a course fades quickly when it is not practiced, so a one-shot competency check on a Friday is a fragile foundation for behavior on the following Tuesday under load. Training creates capability with a short half-life. Adoption requires practice, and practice requires the environment.

The identity dimension deserves its own paragraph, because it is the part of the transition state that polite impact assessments miss. Thelma said the honest version of it: the record is my memory. For a nurse, charting is not a data-entry chore; it is the professional act of vouching for what happened to the patient, and the paper form is how the profession has always vouched. A change that asks the nurse to record in a machine down the corridor is asking her to trust the machine with her professional identity. That is a real loss, and it is not overcome by a slide deck. The same loss appears in every role: the receptionist whose skill is the paper queue, the pharmacist whose value is knowing the doctors’ handwriting, the records clerk whose filing the platform will do. The change takes something from each of them, and the honest impact assessment starts there, not with the list of what they gain.

Six dimensions of impact

The third discipline is the impact assessment, and it is a decision tool, not a report. The question it answers is specific: for each role, what actually changes, and what must be true for the new way to be the easy way? The minimum viable form is one page per role, or one table for the whole team: today, after, what changes, what must be true. Nothing more. The failure mode of impact assessments is completeness: forty pages describing the future state that nobody reads. The working version is a map of decisions, each row ending in a requirement someone can act on, and the requirement is almost always about the environment, not about the people.

The dimensions of impact are six, and they travel together: role, what the job now is; process, how the work flows; technology, what tool the person uses; policy, what rule governs the work; skill, what the person must learn; and identity, what the person believes they are. The first five are easy to find; identity is the one nobody writes down, and the one that decides whether the other five stick. The assessment is done role by role with the people in the role, not for them, and the output is a set of “what must be true” statements that become requirements: the terminal where the paper was, the consult target amended during the dip, the pharmacy interface accepting the routed queue, the policy that still says “paper records retained” amended.

At Meridian the assessment surfaces three things the project had shipped as assumptions. The workflow standard, which Hana Lindqvist owns, was written for the training room: it describes how to use the platform correctly, and it never describes where the platform will live in a clinic built for paper. The terminal placement was decided by facilities on cabling and wall space, not by workflow. And the ten-minute consult target, which governs how Thelma spends her day, belongs to the current state; nobody checked whether the platform could meet it during the transition. All three are environment failures, not training failures, and all three are fixable by someone with authority. The assessment’s job was to say so, with names.

The readiness heat map

The fourth discipline is readiness, and it is three separate questions that are constantly collapsed into one. Capability: can the person perform the new behavior? Willingness: does the person choose to? Environment: does the context keep the new behavior alive after the push? The collapse is the disease: most “readiness assessments” measure enthusiasm, or attendance, or skill, and treat the answer as a single score, and then a row that is green on capability but red on environment is reported as “ready” and fails on day one.

The three questions have different owners and different cures. Capability is the training and performance-support question: skill, access, tooling, time. It is the only one the training program owns, and it is the one projects over-invest in, because it is the one they control. Willingness is the motivation question: motives, identity, trust, effort. It is the one with the longest literature, because it is the one with the least leverage: you cannot mandate willingness into existence. The environment is the reinforcement question: defaults, reminders, incentives, physical layout, workload, consequences. It is the most neglected and the most tractable, because it does not require converting anyone; it requires rearranging the situation so the right behavior is the easy behavior.

The field tool is the readiness heat map: roles down the side, the three questions across the top, each cell marked high, medium, or low, with a symbol, never color alone, because a map read in monochrome must say the same thing. The map is built in the room with the role leads, and it is kept to one page.

The map’s verdict at clinic one is the sentence that should stop every meeting about “attitude”: every row is capability high, willingness medium or high, environment medium or low. The people are not the problem. The situation is the problem. The map makes that visible as a pattern, and the pattern is why the map earns its keep: it converts a moral argument, they should try harder, into a design argument, the environment makes the new behavior the losing choice, and a design argument has a fix.

Resistance is information

The fifth discipline is the reading of resistance, and it is the one that separates the leaders from the enforcers. Resistance is the first evidence about the design, not the first evidence about the people. Before any intervention, ask the question that costs nothing and saves months: is the new behavior actually better in the real context? If the trained behavior is slower, harder, or riskier than the old behavior at the point of work, then the people resisting it are doing the project a favor by not adopting it, because widespread adoption would institutionalize a worse process. The repair is the behavior: the workflow, the tool, the layout, the target. Not the people.

The behavioral economics of the question are worth naming, because they explain why even good changes meet honest resistance. People are loss-averse: the pain of losing what they have outweighs the pleasure of gaining what they are offered, the asymmetry at the heart of prospect theory from Daniel Kahneman and Amos Tversky’s work in 1979. The related status quo bias, documented by Kahneman, Jack Knetsch, and Richard Thaler in 1991, is exactly the pattern at clinic one: the current state is known, tuned, and survivable, and the proposed state asks for losses first — three extra minutes, a riskier record, a threatened identity — with the gains arriving later, promised by people who will not be there to suffer the dip.

Hesitation under those conditions is not irrational. It is a rational response to a poorly designed transition, and the design job is to shrink the losses, make the gains visible and early, and absorb the dip with the environment rather than with exhortation.

The field signal that the design question has been skipped: every conversation about adoption becomes a conversation about attitude. “They are resistant.” “They are set in their ways.” “They were trained; they know better.” When the vocabulary of the discussion is character, the evidence of the design has been dismissed without examination. The counter-move is short and concrete: read the heat map row aloud, capability high, willingness high, environment low, and ask the room to point at the person and then at the corridor. The second gesture answers the first.

A scene from the steering discussion shows the trap from the inside. A clinic supervisor proposes a mandate: anyone who charts on paper after week four gets a written warning. Nora is quiet; Dana is uncomfortable; Sam points out that the mandate punishes the people the environment failed. The supervisor is not cruel. She is accountable for the queue, and the mandate is the only tool she has that moves behavior today. What resolves the scene is evidence, not persuasion: Thelma’s row on the map, and the measurement showing the platform chart took three minutes longer than the paper form under load. The mandate was not wrong because it was harsh. It was wrong because it punished the symptom of an environment failure, and a mandate applied to an environment failure produces compliance theater or resignation. The honest sequence is the reverse: fix the environment first, ask for the behavior second, and look at the person only if the behavior still does not arrive. Resistance is information, and the order of investigation is the discipline.

The sponsor who is seen

The sixth discipline is the sponsor, because the change the project needs is not something the project can do. The project manager can build the platform, schedule the training, and measure the gap. Only the sponsor can absorb the transition costs and remove the obstacles the change creates, because those are decisions about the organization, not about the project.

Elena Marchetti does the two things that make a sponsor real, and both are small and both are visible. The week after the 95/35 discussion, she walks clinic one with Thelma. She does not deliver a speech. She asks one question, the one the impact assessment should have asked in month nine: “What would you change first?” Thelma says, “Put the machine where the paper is.” Elena looks at the corridor and says, “Then that is what we do,” and the terminal moves that week, because moving a terminal is a facilities decision with a small budget and a priority, and a sponsor can make both.

And Elena changes the second thing, the one that costs more: she takes the ten-minute consult target to the board and replaces it, for the transition quarter, with a transition target that includes charting time, so that Thelma is no longer measured against the current-state speed while learning the future-state process. That decision is the sponsor’s, and only the sponsor’s: the throughput target belongs to the operating plan, the operating plan belongs to the board, and a project manager who changes it is overstepping. Elena does not overstep. She removes the obstacle at the level where the obstacle lives.

The research on organizational transformation says plainly why this is the sponsor’s seat and not a nicety. John Kotter’s work since Leading Change (1996) returns to the same finding: transformation fails most often when the effort is delegated to managers while the leadership stays in the steering room, and succeeds when the leadership is visibly, repeatedly attached to the new behavior. The visible attachment is the thing. A sponsor who announces and leaves is a poster; a sponsor who is seen on the floor, who asks what would change first, and who returns a month later to ask what changed, is a force. The distinction is not charisma. It is recurrence: the new behavior stays on the leader’s agenda until it becomes the old behavior.

The empty-seat lesson from chapter 8 returns here in structural form. Nora Kariuki owns adoption; the charter says so, with the grant targets beside her name. But Nora cannot move terminals across six facilities, and she cannot amend a throughput target. Elena can. The seat and the authority must meet in one decision, or the owner is accountable for an outcome she does not control, which is the fastest way to turn a named owner into a scapegoat.

The steering committee meeting after clinic one makes the rule concrete: the adoption number is on the agenda as a decision item, and Elena says the sentence that defines the practice. “When the adoption number is red, I want to know what we decided about the terminal, the target, and the training. I do not want to know who is being counseled.” The sponsor’s job is not to own the number. It is to own the decisions that move it.

The change system, not the training event

The seventh discipline is the instrument mix, because training is the smallest instrument in the change system, and projects invert the proportions: they spend the most where the effect is shortest, and the least where the effect lasts.

Training creates capability, and that is all it does. A course cannot create willingness, and it cannot fix an environment, and its capability decays on a schedule that Ebbinghaus described more than a century ago: steeply at first, then slowly. The training design that respects the decay is the one that practices the behavior in the real workflow and rechecks it in the workflow, not in the classroom. Testing the behavior improves long-term retention far more than re-reading or re-lecturing — the finding at the heart of the work on test-enhanced learning by Henry Roediger and Jeffrey Karpicke (2006). The translation for a clinic is that the competency check that matters is the one administered at the desk, under load, thirty days after go-live, by the super-user, not the one in the training room on Friday. Sam’s training was good and was measured wrong.

Performance support is the instrument that fills the gap between the course and the fluent behavior: the one-page card at the desk, the default that preselects the right path, the super-user one desk away. The design goal is stated in one sentence and it is the whole craft: make the new behavior the easy behavior. Not the correct behavior, the easy behavior. Thelma’s paper form won because it was one hand-movement away; the platform lost because it was a corridor away. Every environment fix in this chapter is an instance of the same rule: the queue card at reception, the pharmacy matching, the terminal by the desk, the target that includes charting. None of them requires anyone to change their mind. They require someone to change the path.

Communication is the instrument with the most ceremony and the least practice. The launch email and the town hall are not the program; they are the announcement. The program is repeated, specific, two-way sensemaking: the staff hear what changes and why from their manager, their peer, and the leadership in the same week, and they are heard back, with questions answered and answers visible. The practitioner’s rule — a message is real only when it has been heard three times from three directions — outlasts the single email, which has the half-life of a meeting announcement.

Incentives are the instrument where projects defeat themselves, and the pattern is old enough to have a canonical name. In 1975 Steven Kerr published the classic essay on rewarding A while hoping for B: organizations that want one behavior while rewarding another get the behavior they reward. The clinic is a textbook case — the project wants adoption, and the operating plan still rewards the ten-minute target, which punishes adoption during the dip. The fix is not a permanent change to the clinic’s performance system; it is a dated transition target, announced as a transition, that includes the learning time and the charting time, and reverts when the behavior is fluent. The transition target makes the new behavior winnable, and it belongs in the adoption plan with an end date, so it does not become a permanent subsidy for slow work.

Local champions are the instrument that uses how adoption actually spreads. Everett Rogers’s diffusion research showed that adoption moves through a social system along its communication channels, not its org chart, and that the population sorts into rough groups: a small set of innovators, a larger set of early adopters, the big middle of the early and late majority, and a final group that holds out longest. The design consequence: find the early adopters — the people for whom the new way is genuinely better — and put them where the majority can see them, because the majority adopts when their peers demonstrate that it works in their context, not when the project proves the system works. Geoffrey Moore named the gap between enthusiasts and the pragmatic majority the chasm.

The tipping-point research adds the number that makes champions a program instead of a hope: in 2018 Damon Centola and colleagues reported experiments in which a committed minority could overturn a majority convention, with the required share around a quarter of the group for conventions that spread slowly, and as little as a tenth when the new convention spread quickly. The mechanism is visibility and consistency: the minority behaves the new way, consistently and observably, until the majority’s assessment of the group flips. The translation for Meridian is the super-user program, and it is the instrument the clinic gets right: seven super-users out of forty, trained first, chosen for being the people their colleagues already ask, deployed at the desks during the transition with time in their week and an escalation path to Sam. Their job is not to answer questions; it is to be seen charting. The environment caveat matters as much as the mechanism — a visible minority cannot tip a group into a practice that loses them three minutes a patient — which is why the environment work comes first in this chapter, and the champions second.

The figure below is the chapter’s primary picture: the adoption curve as a decision instrument.

Figure 11.1: The adoption curve with intervention points and evidence
gates. Adoption is the share of care staff whose last ten patient
encounters went through the platform, measured weekly at the point of
work; the dots are observed values at clinic one. The dashed band is the
target corridor agreed with Nora at gate two: above 50 percent by week
four, above 70 percent by week eight, above 85 percent by week sixteen.
The vertical lines are the evidence gates that decide the waves: week
eight decides the wave-two go-live, week sixteen the wave-three. The
markers W, S, and I are the interventions: the workflow redesign (week
four), the super-user escalation (week five), and the transition target
(week six). The band between 25 and 45 percent is the reversion zone,
where behavior slips under workload peaks — a statement about the
environment, not about the people.

  100%|                         . . . . . . . . . . . . . . . . . . . .
      |                       . . . . . . . . . . . . . . . . . . . . .
   80%|                     . . . .   target corridor  . . . . . . . . .
      |                    . . . .  . . . . . . . . . . . . . . . . . .
      |                   . .  . . . . . . . . . . . . . . . . . . . . .
   60%|                  . . . . . . . . . . . . . . . . . . . . . . . .
      |                 .  .  . . . . . . . . . . . . . . . . . . . . . .
   50%|              . .    . . . . . . . . . . . . . . . . . . . . . . .
      |             .  .    . . . . . . . . . . . . . . . . . . . . . . .
      |           .    .    . . . . . . . . . . . . . . . . . . . . . . .
   35%|---- observed dots sit here through week six ---------------------
      |          . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
   25%|  reversion zone: behavior slips under workload peaks  . . . . . . .
      |        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
      |     . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
    0%|_________________________________________________________________
      week  1     4     8      12     16     20     24
                   |            |             |
                 [8] gate:    [16] gate:    (24) end of the adoption
                 wave-two     wave-three    window; benefits measured
                 go-live      go-live       from here, chapter 44

      Markers: W = workflow redesign (terminal and paper swap places)
               S = super-user escalation (peer coverage at the desks)
               I = transition target (the ten-minute target, amended)

The numbers that explain the gap are worth doing explicitly. Clinic one runs about 120 patient encounters a day. At 35 percent adoption, roughly 40 of those are charted on paper, and each one becomes a second data entry later, at about four minutes each: roughly 160 minutes of rework a day, more than two and a half hours, on top of the queue it creates and the transcription errors it plants in a migration the project already certified. The gap between 95 and 35 is not a percentage point gap. It is about 160 minutes a day, forever, until the environment changes, plus the errors, plus the queue, plus the grant’s access numbers that never move. That is what a measurement failure costs.

Measure behavior, not attendance

The eighth discipline is the measurement of adoption, and it is the discipline that keeps the rest honest, because an adoption plan without adoption evidence is an agenda.

The measures are behavior at the point of work: the share of encounters charted through the platform, workflow completion, time-to-task, abandonment, reversion events, support calls. Each one must have a witnessable definition and an owner, which is the acceptance discipline from chapter 10 applied to behavior instead of to requirements. “Adoption” is not a measure; it is a word. The measure is “the share of care staff whose last ten patient encounters went through the platform, counted weekly, owned by Nora.” That sentence is the entire difference between the 95/35 dashboard and a real one.

The measurement architecture separates leading from lagging. Adoption is the leading indicator; the grant’s access outcomes are the lagging one, and the access numbers follow by months, which is exactly why the project needs the leading measure: it is the early warning that the benefit is on its way, or not. The public-health tradition offers a useful lens, the RE-AIM framework from Russell Glasgow and colleagues (1999): reach, effectiveness, adoption, implementation, and maintenance. The framework’s contribution is the last three terms — adoption asks whether the people who should use it actually do, implementation whether it was delivered as designed, maintenance whether it holds. The project that evaluates only the pilot’s effectiveness and reports success has evaluated the wrong thing.

The benefit linkage is the point: adoption measures exist to predict and explain the benefit measures, and the owner of the benefit, Nora, is the owner of the adoption evidence, not the project team. The project hands her a measurement system, not a report.

The evidence gates are where the measurement does its work, and they are the governance of the rollout. Gate at week eight: is adoption inside the target corridor? Gate at week sixteen: does it stay there under the workload peak? The gates do not decide whether the project is on schedule; they decide whether the next wave of clinics goes live, and they convert the adoption curve from a report into a decision — the only kind of measure that survives contact with a steering committee. The gate question makes the difference: “What behavior do we now see that we did not see at the last gate?” A status update answers “how many are trained.” A gate answers what people are doing, with evidence.

The warning is the metric that is green while the project fails, the lesson of chapter 2 returning in its most seductive form: the engagement metric. Attendance at briefings, messages read, awareness scores, help-desk call volume: every one of these can be green while adoption sits at 35 percent, and every one of them feels like measurement. The tell is the dashboard where the engagement number is rising and the behavior number is flat. The repair is to ask what behavior each metric predicts, and to refuse the metrics that predict nothing.

The initiative tsunami

The ninth discipline is the crowd of changes that surrounds the project, because adoption does not happen in an empty organization, and the most common reason a good change fails is the other changes.

Change fatigue is a real state, not a metaphor. Bernerth and colleagues, developing a validated measure of it in 2011, defined it as the exhaustion that accumulates from the demands of change, distinct from general burnout: every change a person absorbs makes the next one harder, even when each one is individually reasonable. The clinic one staff are not adopting a platform in a vacuum. The same year, the network is consolidating two facilities and introducing a new staffing model. Thelma has been trained for the platform, briefed on the consolidation, and interviewed about the staffing model, in the same quarter, and the three messages arrive from three directions and land as one noise. The apathy that follows — “here we go again” — looks like resistance, and it is not. It is depletion, and depletion is treated by sequencing, not by motivation.

The tool is the change calendar, one page: every initiative touching each role, by month. The question it answers is not whether each change is justified; it is what any given person is changing at the same time. The calendar at Meridian shows the problem in one glance: three initiatives hit clinic one between months eighteen and twenty-four, which is the adoption window for the platform, and two of them were never scheduled against it. The fix is a decision, not a document: the consolidation defers, or the platform’s adoption window shifts, or the staffing-model roll-out starts at the clinics in wave three rather than wave one. The sponsor’s job includes protecting the adoption window, because the window is the narrow period when the new behavior becomes the old behavior, and a competing change inside the window resets it.

The discipline of no is the uncomfortable half of the practice. The project that cannot decline, defer, or re-sequence a competing initiative is financing its own adoption failure, and the refusal is not obstinacy; it is the same resource decision the book has made at every level, from portfolio selection in chapter 5 to scope control in chapter 10. The change calendar is where the project states, with dates, what it will not absorb, and the sponsor defends the statement. Without the defense, the adoption curve is not a measure of the change; it is a measure of everything else that was happening to the staff at the same time.

Waves, not a waterfall

The tenth discipline is the shape of the rollout, because the single biggest adoption decision is not how to train; it is how many clinics go live at once.

The wave design at Meridian is deliberate, and it is the project’s strongest adoption decision. Clinic one goes live as wave one, an experiment the program is willing to learn from: the terminal lesson, the transition-target lesson, the super-user lesson, the reversion data, all of it feeds the design of wave two before clinics two and three go live, and wave three’s evidence feeds the last wave. The sequence converts the adoption curve from a risk into an engine: each wave is a cheaper, better-informed version of the previous one. The alternative, the big-bang cutover of all six clinics on a calendar date, is the seductive failure: it looks decisive, uses the certification once, and converts every lesson of clinic one into a six-clinic-wide mistake.

The delivery-style contrast lands here, because the rollout is a hybrid and the seam is explicit. The formal machinery is the gates: the readiness checklist from chapter 8, the acceptance criteria from chapter 10, the certification evidence, the non-negotiables of safety, privacy, and continuity, all of it runs on the predictive cadence with baselined evidence. The adaptive machinery is the weekly adoption review: the observed behavior, the workflow fixes, the training redesign, the terminal placement, all of it runs on the feedback cadence, reprioritized every week from the floor. The seam discipline is that the same thing is never governed twice: the workflow standard is formal, and the workflow fixes are adaptive, and the line between them is drawn in the adoption plan with owners. The hybrid fails exactly where the seam is invisible: a weekly adoption fix frozen into a baseline, or a formal gate waived because the calendar wants it waived. The gate decides the wave; the wave is never decided by the calendar.

The predictive and adaptive versions of the practice are the same discipline at different cadences. In a fully predictive environment, adoption is a controlled workstream; in a fully adaptive one, it is a continuous feedback loop in which the product changes to fit the behavior rather than the behavior being forced to fit the product. Meridian is between them, as most real projects are, and the deliberate hybrid is the point: the gates keep the non-negotiables honest, and the weekly loop keeps the negotiables learning.

How adoption programs die

The failure patterns deserve to be named together, because each one is produced by competent people in good faith, and each one kills the change in a different way.

The training-room theater: attendance and pass rates reported as adoption, one column where “trained” and “using” are the same number, a 95 percent metric with a flat behavior curve — the exact shape of clinic one’s original dashboard.

The announcement trap: the launch email and the town hall treated as the change program, an adoption curve that moves during announcement week and never again.

The convert-and-forget: champions selected, trained, and released with a badge and no time in their week, no shift cover, no escalation path, a super-user “available” in a room while the desks struggle.

The incentive contradiction: the current-state target surviving into the dip, punishing the new behavior exactly when it needs reward — “I would, but I cannot miss my number.”

The engagement theater: attendance, awareness, and messages read all green while behavior is flat, a green engagement score on the same page as a flat adoption curve.

The project manager as adoption owner: the plan written by the project team and approved by nobody who runs a clinic — the empty-seat lesson from chapter 8 returning in its most expensive form.

The all-at-once cutover: six clinics on one date, a go-live that “goes well” because nobody measures the week after, a launch report that ends at go-live.

The repairs are not new instruments; they are the disciplines of this chapter applied in order: fix the environment first, ask for the behavior second, and look at the person only if it still does not arrive.

The machine reads the dashboard; the leader reads the floor

The assistant has a real place in the adoption work, and its boundary is the same one this book has drawn at every step. It can do the tractable, verifiable parts: detect the anomaly in the usage data, the drop in chart opens at eight in the morning that marks where the paper queue wins; draft the first-pass impact table from the role descriptions; cluster the open-text feedback into themes; summarize the week’s adoption review for the steering committee. The source data is the approved, anonymized usage and feedback set, nothing personal, nothing clinical, nothing regulated; the summaries are drafts until a named owner confirms each conclusion against the floor; and the decisions, the terminal move, the target change, the wave gate, remain human, recorded, and auditable. A fluent paragraph about morale is a draft, not a diagnosis, and it must not become evidence on its own. The verification is the walk: the leader reads the dashboard, walks the floor, asks Thelma what would change first, and comes back with the one sentence the dashboard cannot produce. The two readings are both required, and neither substitutes for the other.

Practice

One. A quick classification. Each statement below is either output evidence or adoption evidence, and each names one of the three readiness questions. Classify both. (a) “Eighty-five percent of clinic staff completed the platform course.” (b) “The share of encounters charted through the platform reached 70 percent in week eight.” (c) “Thelma’s row on the heat map is environment low.” (d) “The clinic exceeded its transition target for the quarter.” (e) “The super-users answered 240 support calls in May.” (f) “The paper encounter form is no longer stocked at the nurses’ station.”

(a) is output evidence: training is a deliverable the project produced, and it says nothing about behavior. (b) is adoption evidence: behavior at the point of work. (c) names the environment question, the third of the three readiness questions. (d) is adoption evidence, with the transition target as its witnessable definition, and it is the incentive instrument doing its job. (e) is activity evidence, not adoption evidence: support calls answered is an engagement-style metric that can be green while behavior is flat, and the drill’s lesson is that help-desk volume measures the support system, not the change. (f) is adoption evidence of the strongest kind: the environment was changed so the paper path is no longer the easy path, and it is the workflow-redesign intervention made physical.

Two. A field drill: the change you lived through. Take a change your organization recently went through, or one you are in now. Produce the impact assessment: one row per role, columns for today, after, what changes, and what must be true for the new way to be the easy way. Then build the readiness heat map for the two roles closest to the point of work, capability, willingness, and environment, with a symbol in each cell. Then answer the diagnostic question honestly: was the trained behavior actually better in the real context, or was the environment punishing it? Write the single sentence that says which.

The drill succeeds when at least one row is environment low, when that row explains a behavior that had been explained by attitude, and when the “what must be true” column contains at least one thing that was never true. The most common failure of the drill is the assessment written for the people instead of with them, which reproduces the exact blind spot the tool exists to remove: the terminal in the corridor is visible from a walk, and invisible from a document. If your heat map has no environment-low rows, you have not looked at the point of work; you have looked at the org chart.

Three. A decision room: the week-eight gate. It is week eight after clinic one go-live, and the evidence is in front of the steering committee. Adoption is at 55 percent, up from 35 percent at week three, after the terminal move and the transition target, and still short of the 70 percent corridor agreed for week eight. The super-user program is working; the records clerk row is fully green; the clinician row is the laggard. Wave two, clinics two and three, is scheduled to start configuration in three weeks, and the network’s facilities consolidation is now scheduled to touch clinic two in the same quarter. The board wants the wave to proceed; the facilities director wants the consolidation to proceed; Nora wants both protected. Decide what wave two needs before it can start, and say what you would demand at the gate and what you would defer.

The defensible answer starts from the evidence: 55 percent is real movement, up from 35 percent at week three, but it is still below the 70 percent corridor agreed for week eight, so the wave-two gate is conditional, not open. The conditions should be specific: the clinician row moves above the reversion band, with the workflow fix that makes the consult target survivable in the real flow; the super-user ratio and the transition target are copied into the wave-two plan, because they are the interventions that moved the curve, not decorations; and the facilities consolidation is sequenced out of clinic two’s adoption window, or the window shifts. The unsafe choice is proceeding on the calendar because the board wants the wave — the gate becomes ceremony; the equally unsafe choice is freezing the whole program on a single missed corridor — the gate becomes a cliff, and the rollout stalls on the laggard row. The judgment call is the clinician row, where identity and process meet, and the fix is environmental — the workflow and the target — not motivational. Reasonable but risky: letting the consolidation proceed while the change calendar is renegotiated quarterly, defensible if the renegotiation has a real owner and a real consequence when it is ignored.

Four. The mastery drill: trained but not adopting. You are assigned to a project where the training numbers are excellent and the behavior numbers are flat, exactly the shape of clinic one. Design the intervention. State the order of investigation, the first three actions you take, the evidence you demand, and the conditions under which you would conclude the people are the problem after all.

The intervention begins with the diagnosis, and the order is the discipline: walk the point of work before writing the plan, because the environment failure is visible from the floor and invisible from the dashboard. First, observe the actual workflow under load and time the new behavior against the old one — the evidence that decides the whole case. Second, build the readiness heat map with the role leads, sorting the failure into capability, willingness, or environment. Third, read the map with the sponsor in the room, which converts the diagnosis into the decisions only the sponsor can make: the environment fix, the transition target, the wave gate. The evidence you demand is behavior at the point of work, with a witnessable definition and an owner; the gate evidence that decides whether the intervention worked is the curve moving into the corridor and staying there under a workload peak. The conditions under which the people become the problem after all: the environment is fixed, the new behavior is genuinely better in the real context, the support is in place, the incentives are aligned, and the behavior still does not move. Only then is the row a people row — and even then the response is not counseling; it is a decision about fit, made by the owner, with the evidence in the file. The unsafe answer is the one the story tempts: more training. Thelma was trained, willing, and punished by the environment, and another course would have moved the number by nothing. The equally unsafe answer is the mandate, which punishes the symptom of an environment failure and produces compliance theater. The drill’s lesson is the chapter’s spine: adoption is a design problem before it is a people problem, and the design problem is solved by changing the situation, not by exhorting the people.

Five. The transfer question. In the last change you were part of, what was the metric that was green while the behavior was flat, who owned the transition target, and what was the corridor, the terminal, the policy that made the new behavior the losing choice? Who was the person who looked like they were resisting and were actually carrying the design’s mistake?

The durable principle: adoption is the behavior that turns an installed system into a realized benefit, and the project that delivers the system and measures the training has finished the easy half. Mastery means designing the transition as deliberately as the build: describe the three states, read resistance as information, make the new behavior the easy behavior, put the sponsor in the decisions that move the curve, measure behavior at the point of work, and roll out in waves that learn. The most common next failure is subtler than the ones named here: the adoption curve reaches the corridor, the project declares adoption achieved, the reinforcement stops, and the behavior decays back to the paper, unnoticed until the access numbers miss the grant. The curve that reached the corridor is a claim that must be maintained. The next chapter takes up the instrument that carries the whole change program: communication, negotiation, and alignment, because the adoption plan runs on conversations, and the craft of those conversations is the subject of “Communicate, Negotiate, and Build Alignment.”

Notes

• Meridian Community Health Network is a composite case created for this book; no real network, vendor, system, or people are depicted. Clinic one’s staffing of forty care staff, the 95 percent training pass rate, the 35 percent adoption figure, the ten-minute consult target, the three-minute charting penalty, the 120-encounter day, the four-minute second data entry, the 160-minute daily rework figure, the seven super-users, the week-eight and week-sixteen gates, the 50/70/85 target corridor, the reversion zone from 25 to 45 percent, and the facilities consolidation and staffing-model initiatives are author-created illustrative figures consistent with the earlier Meridian chapters. All named characters are composite.
• The three-state description of change (current state, transition, future state) is this book’s working vocabulary; the underlying idea follows Kurt Lewin’s field theory, including his account of unfreezing, moving, and refreezing as described in “Frontiers in Group Dynamics,” Human Relations 1, no. 1 (1947): 5-41, and the practitioner tradition of transition management associated with William Bridges, Managing Transitions: Making the Most of Change (Reading, MA: Addison-Wesley, 1991). This book uses its own vocabulary and does not reproduce either author’s wording.
• The term “implementation dip,” describing the decline in performance when a new practice is adopted, is from the work of Michael Fullan, most accessibly in The New Meaning of Educational Change, 4th ed. (New York: Teachers College Press, 2007); the concept is presented here in the author’s own words.
• The decay of newly learned material is from Hermann Ebbinghaus, Über das Gedächtnis (Leipzig: Duncker & Humblot, 1885), the classic experimental study of memory and forgetting; the qualitative pattern described here, steep early decay followed by slower decline, is its durable finding.
• The finding that retrieval practice, testing, improves long-term retention more than restudying is from Henry L. Roediger III and Jeffrey D. Karpicke, “Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention,” Psychological Science 17, no. 3 (2006): 249-255.
• Loss aversion is from Daniel Kahneman and Amos Tversky, “Prospect Theory: An Analysis of Decision under Risk,” Econometrica 47, no. 2 (1979): 263-291, and the status quo bias is from Daniel Kahneman, Jack L. Knetsch, and Richard H. Thaler, “Anomalies: The Endowment Effect, Loss Aversion, and Status Quo Bias,” Journal of Economic Perspectives 5, no. 1 (1991): 193-206. The chapter’s argument that hesitation under a badly designed transition is rational follows from these results; the application is the author’s.
• The account of rewarding A while hoping for B is from Steven Kerr, “On the Folly of Rewarding A, While Hoping for B,” Academy of Management Journal 18, no. 4 (1975): 769-783.
• The adopter categories, innovators, early adopters, early majority, late majority, and laggards, with approximate shares of about 2.5, 13.5, 34, 34, and 16 percent, are from Everett M. Rogers, Diffusion of Innovations, 5th ed. (New York: Free Press, 2003); the percentages are descriptive findings from his synthesis of diffusion research, not fixed laws. The idea that diffusion stalls between enthusiasts and the pragmatic majority is from Geoffrey A. Moore, Crossing the Chasm (New York: Harper Business, 1991).
• The tipping-point finding, that a committed minority of around a quarter of a group could overturn a majority convention in experiments, with a smaller minority, around a tenth, sufficient when the new convention spread quickly, is from Damon Centola, Joshua Becker, Devon Brackbill, and Andrea Baronchelli, “Experimental Evidence for Tipping Points in Social Convention,” Science 360, no. 6393 (2018): 1116-1119, published 8 June 2018. The experiments concerned laboratory groups choosing conventions; the chapter’s use of the finding for local-champion design is the author’s application, and the environment caveat is the author’s, not the paper’s.
• The finding that organizational change can accumulate into a measurable state of change fatigue is from Jeremy B. Bernerth, H. Jack Walker, and Stanley G. Harris, “Change Fatigue: Development and Initial Validation of a New Measure,” Work & Stress 25, no. 4 (2011): 321-337. The chapter’s treatment of apathy as a possible signal of depletion rather than resistance follows the authors’ conception; the change calendar is this book’s working instrument.
• The RE-AIM evaluation lens, distinguishing reach, effectiveness, adoption, implementation, and maintenance, is from Russell E. Glasgow, Thomas M. Vogt, and Shawn M. Boles, “Evaluating the Public Health Impact of Health Promotion Interventions: The RE-AIM Framework,” American Journal of Public Health 89, no. 9 (1999): 1322-1327; the chapter applies it to project adoption evidence in the author’s own terms.
• The five stages commonly applied to change (denial, anger, bargaining, depression, acceptance) derive from Elisabeth Kübler-Ross, On Death and Dying (New York: Macmillan, 1969), where they were proposed about grief, not organizational change; the adaptation to change management is a practitioner practice, the staged model is contested in the clinical literature, and this chapter deliberately treats the emotional content of change without adopting the five-stage vocabulary as a diagnostic tool.
• John P. Kotter, Leading Change (Boston: Harvard Business School Press, 1996), and his later writings argue that transformation fails most often when leadership attention detaches before new behavior is anchored; the chapter’s account of the visible sponsor follows that line of argument in the author’s own words and does not reproduce Kotter’s step frameworks.
• The readiness vocabulary in this chapter, capability, willingness, and environment, where environment is the reinforcement question, is this book’s own working model. It overlaps in intent with commercial change-management methods, notably the ADKAR model (awareness, desire, knowledge, ability, reinforcement) associated with Prosci; this book does not reproduce that proprietary material and readers should consult the original publisher for its current form.
• ISO 21502:2020, “Project management: Guidance on project management,” treats interested parties where this book says stakeholders and addresses organizational context at a general level; it does not prescribe the change-management instruments in this chapter. The chapter is a synthesis of the sources above plus the author’s experience; none of the sources is claimed to guarantee adoption, and the chapter’s prescriptions are the author’s judgment.