Change Management and Embedding Change

A fictional water utility replaced paper work orders with a scheduling application on ruggedized tablets. The software launched on schedule and every crew member attended a half-day session. The board closed the project at go-live, but no operating owner accepted responsibility for checking adoption or benefits. Six months later, an audit of completed work orders found that crews had closed 40 percent directly in the field application. The other 60 percent were completed on paper and entered by a clerk on Fridays. All eventually had application records, so record presence alone would have concealed the different workflows. This article examines organizational change across its lifecycle, with particular attention to sustained use after launch: individual and organizational models, line managers and sponsors, adoption evidence, concerns, and continuing ownership. The utility and all its results are invented; the public guidance provides concepts to assess, not proof that its remedies work everywhere.

What change management is, and what it is not

GovS 002, the UK government project delivery standard, describes organizational change management as preparing and supporting people and organizations to adopt and embed changed approaches. Within its scope, planning these aspects from the start and throughout delivery is a requirement. It recommends continued readiness assessment, outcome milestones, and monitoring after implementation. The standard distinguishes mandatory “shall” from advisory “should”; its monitoring recommendation is not a universal legal duty. A utility borrowing the approach should define its own obligations, acceptance criteria, and continuing responsibilities. In this case, a launch milestone and an attendance record did not establish that crews could complete work in their actual environment.

The operational practice described under change enablement addresses authorization, risk, and evidence for changes affecting services. Organizational change management addresses changed work, readiness, and sustained behavior. Their purposes differ, but user impact, training, support, and transition readiness can concern both. A technical approval does not by itself establish adoption, and a people-focused plan does not replace release authorization. The utility needed the two activities to share evidence about field conditions.

Two models, two units of analysis

Prosci’s ADKAR model examines individual change through awareness, desire, knowledge, ability, and reinforcement. It directs support toward the earliest unmet element and calls for reassessment as needs change. Use it to ask whether a person understands the reason, supports participation, knows what to do, can perform the work, and has support to sustain it. Training attendance establishes none of those answers by itself. Keeping paper forms may reflect a tool constraint, uncertainty, a skill gap, disagreement, or several causes. Ask the person, check understanding, and observe representative work before assigning a barrier or prescribing another course.

Kotter’s eight steps address organizational mobilization. The firm distinguishes the original linear model from its current accelerators: urgency, coalition, vision, participation, removing barriers, short-term results, continued momentum, and institutional support. These activities can recur as conditions change. For the utility, an unsupported field workflow and no post-launch owner were concrete gaps to investigate. Early results and sustained practice deserve attention here, but do not replace participation, a credible purpose, or removal of the practical obstacles.

	ADKAR	Kotter’s eight steps
Unit of analysis	One person	The organization
Question it supports	Which individual needs and barriers require investigation?	Which organizational conditions and actions support the change?
Use for a technology leader	Choose and reassess support using evidence about people and their work	Connect mobilization, barrier removal, learning, and sustained organizational support
Where it misleads	Treating a structural problem, such as a system that makes the job slower, as an individual deficit of desire or ability	Manufacturing urgency; treating the sequence as a checklist; declaring the last step done at go-live
Evidence needed	A completed checklist is not proof of adoption. Verified task performance and sustained behavior can provide evidence, whether gathered through a model-based assessment or another method

A form marked complete is different from an observed task performed successfully. Examine the evidence behind each assessment and its coverage. A pattern across a crew warrants looking for shared constraints, but does not identify one cause or rule out individual learning needs. The article on working conditions and performance treats the person-versus-seat question as a hypothesis to investigate. A supervisor might need coaching while the product team also needs to fix an offline workflow.

The managers who carry it

The utility had a sponsor, communications staff, and a training vendor, but had given its twelve field supervisors no defined adoption role. Prosci’s resistance guidance describes complementary communication roles for senior leaders and line managers, with sponsorship and team support working together. The UK Civil Service line management standards likewise connect support through change with assessing results against agreed objectives. Applied here, supervisors needed a route to raise field problems, a clear account of the changed work, and time to help crews practice. A supervisor could not alone commit the vendor’s work, approve every operational exception, or decide the clerk’s future role.

Prepare managers in time to support their teams. Advance learning may help; joint design and supported learning with a crew can also work. Check demonstrated readiness rather than enforcing a universal training order. Explain workload and role impacts, allocate time and cover, and agree how performance will be judged during transition. Unchanged targets may create conflicting incentives, but do not establish that a manager is unwilling to help. The process owner coordinates workflow and acceptance, product and supplier teams address defects, and resource owners confirm commitments. The executive sponsor supports decisions and continuing justification within an explicit mandate. That oversight complements the supervisor’s daily support.

Adoption is measured, not assumed

The digital take-up measure in the GOV.UK Service Manual counts completed digital transactions, including assisted digital transactions, relative to all channels in the same period. It is not a count of unique people or proof that every internal step uses a new system. The Manual sets measurement and channel-planning expectations for its service context. An internal utility workflow needs its own definition. The article on adoption and benefits explains how to define that operational measure and connect it to the promised outcome.

For this utility, the six-month audit covered unique work orders marked field-complete during month six, reconciled with application and paper records through the following Friday. Field-direct closure meant the crew recorded the required completion information in the application without a paper-to-clerk handoff. Of every 100 audited completed orders, 40 followed that path and 60 used paper followed by re-entry; the invented initial audit had no unclassified cases. Later reviews would keep approved paper exceptions, other mixed paths, and unknown paths visible rather than silently remove them. Count each order once and report unfinished work separately. Report depot counts and workloads alongside percentages: nine high-use depots do not establish high use across the utility if another depot handles most orders.

Read access, workflow use, and outcomes separately. An account or issued tablet is not demonstrated readiness; direct use is not proof of faster or more accurate work. Define same-day closure as the proportion of eligible field-completed orders with a complete central record available by the end of that depot’s local calendar day. Include paper routes, disclose missing timestamps and records, and inspect unfinished work, errors, and rework so speed does not conceal omissions. Compare periods with similar work types and volumes. Actual reassignment of the clerk and the remaining re-entry workload also need verification. Benefits realization separates forecasts, observed improvements, attributed effects, and financial consequences.

Reading resistance

Prosci’s resistance guidance identifies possible concerns about understanding the change, job security, leadership, and uncertainty. In particular, insufficient understanding of the reason can be a barrier; awareness itself is not the problem. Such concerns can appear as questions, avoidance, disengagement, slower work, or opposition, but an observed behavior does not establish its cause. In the utility, the clerk’s planned redeployment raised questions about jobs while crews reported poor-signal delays. Both deserved investigation. The consultancy reports favorable survey associations for preventive work; those claims support asking questions early, not a guarantee that a particular communication plan will prevent resistance.

Separate what happened, whether an objection is valid, and why someone acted. A well-founded defect report can coexist with reluctance, fear, or a training need. Reproducing the six-minute delay supports a product diagnosis; merely predicting that a fix would end paper use does not prove it. If paper persists after a repair, investigate remaining constraints rather than treating the original complaint as false. A Friday backlog can reveal a hidden workflow and start an inquiry. Combine record tracing, observation, and conversations. Psychological safety can support candid reporting, but time, knowledge, accessible channels, and visible responses matter too. Assign each concern an owner, decision date, and response to the people who raised it.

Continuing review and ownership

Project closure and sustained organizational change are different judgments. GovS 002 calls for a controlled transition with ongoing responsibilities accepted; its change guidance recommends monitoring sustained behavior after implementation. A project may close while an operating owner continues that work. Here the defect was closing without that assignment and treating launch as proof of realized benefits. ADKAR’s reinforcement and Kotter’s institutional support direct attention to continued practice, feedback, and management arrangements. They do not make a review date or the disappearance of every old form sufficient evidence of success.

Agree review timing with the expected adoption and outcome cycle. The utility’s eventual 30-, 90-, and 180-day remediation reviews were its chosen schedule, not a general rule. Continue routine oversight during delays, add checks relative to when each cohort actually adopts, and escalate material defects or risks between meetings. Review current use and exceptions, outcome evidence, workload, and unresolved conditions; assign actions and the next decision. The article on executive communication recommends reporting the current position against a baseline or limit, changes, uncertainty, and required action or awareness. An unchanged serious risk belongs in that report too.

Concurrent work needs a capacity assessment. DORA’s 2024 research examines unstable organizational priorities and reports associations with productivity and burnout in its surveyed population. Its proposed explanations concern uncertainty, control, and workload. Several coordinated programs can have stable priorities, while one program can repeatedly change direction; a count of three does not diagnose saturation in field crews. Team Topologies’ discussion of continuing organizational adjustment is another prompt to examine manageable change, not a rule to serialize every initiative. Map demands on each group: training, duplicate work, support, new responsibilities, dependencies, and service obligations. Then decide whether to reduce scope, add support, proceed together, or move a date, and check the actual experience.

The utility’s remediation and follow-up

In the invented timeline, crews first reported the signal problem in launch week one. The clerk’s month-six audit exposed the 40 percent field-direct figure, but the issue remained unresolved. A new head of operations arrived around month eight and reviewed it in her first week. The company assigned her authority over the field workflow and its operating improvement budget, with separate supplier, security, and employment decisions following their authorized routes. She named an adoption owner and obtained support commitments before starting a remediation plan. Its day 0 was that plan’s approval, not the original launch. Reviews were set for days 30, 90, and 180, with event-based escalation and later cohort checks where needed.

During the first fortnight, observation reproduced a six-minute closing delay in poor-signal locations for the tested tasks; the team recorded its scope rather than treating it as every crew’s delay. The vendor accepted the offline-closure defect under the support agreement. Operations authorized a controlled paper fallback, with record tracking and reconciliation, while the repair was developed. A review date did not automatically terminate that route. Around day 60, the offline candidate passed tests for capture, synchronization, duplicate handling, access, and restoration before an authorized phased release. Each depot retired redundant routine paper use only after its readiness and support checks. Required fallback and approved exceptions remained available and measured.

Supervisors received agreed responsibilities, practice time, and a route to the adoption owner; depot results informed local discussions and sponsor oversight. The people team and operations confirmed the clerk’s proposed reassignment and discussed it with affected staff, without assuming an announcement ended every concern. Before remediation, the analyst reconstructed a month-seven baseline from retained paper and application timestamps. Missing timestamps were reported, and comparisons used the same observed work categories with the limits of that retrospective baseline explicit. By day 30, an initial improvement in same-day central closure was observed after record reconciliation and support changes, before the offline repair. That observation could not establish the later repair’s effect.

At day 180, nine of the twelve depots reported field-direct closure above 90 percent for the latest monthly cohort. The report retained each depot’s order count, the other depots’ results, unknown paths, and the volume-weighted utility rate rather than calling the whole utility 90 percent adopted. It checked whether earlier same-day improvements persisted, using the stated scope and definitions, alongside record accuracy, rework, and uncompleted work. The analyst reported the observed changes alongside workflow repair, support, and reconciliation, with attribution still uncertain; the observations did not isolate a software effect. The clerk still spent part of the week on remaining re-entry, so full redeployment was not claimed. Operations retained an owner and further reviews for unresolved outcomes and later adopters.

The next proposed change was a safety-checklist application. The safety owner confirmed that the existing approved checklist and controls remained in force and assessed the consequences of delaying only the new application. Based on shared training and support demand, the authorized portfolio owner moved its planned rollout and set a day-90 decision to reassess readiness. A new safety requirement or unacceptable risk would trigger an earlier decision and additional support. The deferral was a choice under those conditions, not a rule that a second change would reset the first.

Suppose the fictional utility later introduces AI assistance that suggests a work-order category and scheduling priority. The process owner first defines the permitted task and which decisions remain with a qualified supervisor. Existing safety checks and escalation requirements remain in force. Managers and crews help redesign the actual workflow: staff examine the suggestion, check it against the work-order evidence, record an acceptance or override with the relevant reason, and route uncertain or safety-sensitive cases to the authorized decision owner. The product and model owners supply evaluation evidence, operating limits, and a tested fallback to the existing process. Supervisors need protected time to demonstrate that workflow with their crews, including unsuitable suggestions and unavailable service; course attendance or access to the tool does not establish readiness.

For that AI change, define adoption around eligible work orders where the agreed review and decision steps were completed, with exceptions and missing evidence visible. Report accepted and overridden suggestions separately. A high acceptance rate can reflect useful advice or uncritical agreement, so it is not the adoption target or proof of benefit. Compare accuracy, handling and review effort, rework, and operational outcomes for comparable tasks, including cases processed through the fallback. Each manager remains accountable for the support and observed working practices within their remit; the process owner owns the outcome and the route for unresolved constraints. Model and product owners investigate reported errors and changes in behavior, while the sponsor obtains decisions and resources beyond local authority. Continuing cohort reviews check whether crews still follow the agreed practice and whether the expected benefit persists. A model update or changed task scope requires renewed evaluation and an assessment of any additional manager and crew support before wider use.

What a reviewer should ask

• Was the change to people’s work planned from the start, with milestones for outcomes, or added as training before go-live?
• What does each model help investigate, and which observations support the assessment rather than merely marking an activity complete?
• Which shared constraints, individual needs, and valid concerns explain the observed work, and what evidence could change that explanation?
• Are managers prepared to support their teams, with time, suitable authority, and routes to product, process, resource, and sponsorship decisions?
• Are access, workflow use, and outcomes measured separately with clear units, periods, exceptions, missing data, and comparison limits?
• Can voiced concerns and hidden workarounds be investigated, assigned, decided, and answered without assuming one cause?
• Who accepted continuing ownership, when do regular and cohort reviews occur, and what triggers an earlier review or a change to the plan?
• What combined demands fall on the same people, and what evidence supports parallel work, extra support, or a schedule change?

The utility’s launch and attendance records described delivery activities. Field observations and defined workflow measures revealed what still prevented the intended work. Supervisors, operational owners, sponsors, and delivery teams had different decisions to make and evidence to retain. Sustaining the change meant supporting that work, checking outcomes, and revisiting unresolved conditions. A justified fallback could remain; a closed project could have a continuing operating owner. Neither an empty truck folder nor a high usage percentage settled those judgments.

References

• UK Government Functional Standard GovS 002: Project delivery, version 2.1, 1 September 2025, clauses 6.4.8 and 8.7–8.9, principle 8, and the overview on mandatory and advisory wording.
• Prosci, The Prosci ADKAR Model.
• Prosci, The Complete Guide to Managing Resistance to Change. Research claims on these pages are the consultancy’s own.
• Kotter, The 8-Step Process for Leading Change.
• UK Civil Service, The Line Management Standards: Core stage, April 2025, V1.2, Change-Focus.
• GOV.UK Service Manual, Measuring digital take-up.
• DORA, Accelerate State of DevOps Report 2024, on unstable priorities.
• DORA, 2024 Research Questions, organizational priorities.
• Matthew Skelton and Manuel Pais, Team Topologies: Key Concepts.