Skip to main content
Steady at the Top
← All articles

How to restart a change effort after the last one failed

Restart a failed transformation by explaining what changed in resources, decisions, and accountability before asking employees to participate again. Acknowledge the prior gap with confirmed examples. Offer a testable restart commitment rather than a new label for the same unsupported plan.

A psychological contract is a person’s understanding of obligations between themselves and another party, informed by perceived promises. Rousseau and McLean Parks discuss this perspective in their 1993 synthesis, The Contracts of Individuals and Organizations. Distinguish that understanding from legally enforceable contract terms. A relaunch needs to address what employees reasonably heard without pretending that every interpretation was identical.

A private moment that makes the issue visible

Composite scenario, not a client case: You rehearse a relaunch message at home and hear phrases used in the previous campaign. Employees remember training that never arrived and decisions that remained unresolved. The new name has changed faster than the operating conditions.

Mark the familiar phrases in your draft and identify the earlier commitment behind each one. Ask what an employee who tried the first rollout would expect to happen next. If the new plan supplies no different answer, postpone the motivational language. Work first on the missing resource, unresolved decision, or handoff that the old phrase concealed.

Use the research within its limits

Levitt and March’s 1988 review links organizational learning to experience carried in routines; Levitt and March (1988): Organizational Learning. In a relaunch, inspect the routine that allowed the previous gap to persist. A renamed program or a new sponsor does not demonstrate that this condition has changed. The restart commitment review is a proposed planning tool, not a tested intervention derived from their paper.

Use industrial/organizational psychology, the psychology of work and organizations, to examine the promised exchange around participation. Specify what employees are being asked to contribute and what leadership has confirmed it will supply. Keep skepticism about a repeated promise separate from a judgment about an employee’s willingness to work.

State the decision before discussing personalities

The decision is whether the organization has removed a meaningful cause of the earlier failure. If it has not, improve the plan before improving the announcement.

Choose whether to restart, redesign, or continue the pause. Set a minimum condition for that choice, such as confirmed implementation coverage. Ask its owner to show that the condition exists now. An intention to resolve it after launch is a new risk to approve, rather than evidence that the cause of the earlier failure has been removed.

Compare an employee’s earlier request for training with the training actually made available. Record the gap without inferring that every later problem came from it. Ask whether the same task can now be performed with the supplied materials and access. This keeps the relaunch discussion tied to a repair employees can inspect instead of a demand to forget the past.

Use the change-restart commitment review

  1. Describe the prior gap. Use verified examples of what was promised and what occurred. Avoid blaming employees for expectations leadership created. Select a verified example of the earlier promise and its delivery. Show the relevant wording to the people responsible for the work and invite factual correction. Do not reconstruct the history solely from the sponsor’s closing presentation. Describe what remains unknown, including whether an individual example reflects a broader process problem.

  2. Identify a changed condition. Name the resource, decision right, or dependency that now differs. Verify it with the responsible owner. Obtain the actual staffing, access, or approval commitment required by the repair. Name the owner who can keep it available during the test. If the condition is provisional, state the condition under which it could disappear. Avoid describing a requested budget or unsigned resource plan as a completed change.

  3. Define a bounded restart. Choose a real application that can test the change. State what would cause another pause. Limit the restart to a task that depends on the repaired condition. Agree what would stop the test and who makes that call. Preserve the paused state for work outside the pilot. Otherwise a small trial can become a full relaunch through meetings and instructions that quietly commit the rest of the organization.

  4. Make the commitment explicit. Explain what participants will receive and what leadership will review. Mark unresolved matters without inventing reassurance. Explain the employee contribution and leadership’s matching obligation in observable terms. Name the coverage provided, the decision authority, and the reporting commitment that actually apply. Keep the date for unresolved decisions separate from an assurance about their outcome. Ask managers to identify any sentence they would struggle to honor in practice.

  5. Report the test honestly. Show gains, drawbacks, and remaining work. Revise the plan before expanding an approach whose assumptions failed. Compare the pilot with the earlier failure condition, including an example where the repair did not hold. Publish the next decision, not only the pilot’s most favorable result. If leadership proceeds despite an unresolved dependency, name the tradeoff and its owner. Do not shift responsibility for accepting that risk onto employees asked to participate.

If an immediate operational failure requires a temporary workaround, authorize and document it through the responsible owner. Keep that decision separate from the relaunch promise. Tell participants when the workaround will be reviewed, and do not present emergency effort as evidence that the intended implementation arrangements can now support ordinary work.

Put the boundary into words

Use this draft restart explanation after the operating owners have verified the changed conditions:

Last time, we asked for this change without supplying these conditions. Here is what now differs and how we have verified it. We will test the restart in this work area and report the result before extending the commitment.

Ask a manager who experienced the first rollout to identify what the new promise means for the next shift. Check whether their answer matches the confirmed coverage and authority. Revise any phrase that implies more. The announcement should be clear enough to expose a gap before employees discover it while trying to perform the task.

Work through a bounded example

Hypothetical application: A system relaunch previously failed because managers lacked implementation time. The restart includes confirmed coverage and a smaller test. The announcement explains that difference instead of asking people to leave skepticism behind.

Extend the hypothetical by inspecting the coverage schedule during the pilot. If the assigned cover is repeatedly withdrawn for customer work, the repaired condition is not reliably reaching the test. Report that result even if the software itself works. Decide whether to change the coverage commitment or reduce the pilot, instead of declaring that resistance has returned.

Choose evidence that can change the decision

Create a restart record with the earlier failure condition, the changed arrangement, and evidence that the arrangement reached the pilot. Separate a purchased tool from access to it, and scheduled training from completed practice. Ask an operating owner to verify the critical dependency in use. Keep the first failure account beside the new evidence so a change in terminology does not conceal recurrence. When the repaired condition disappears, reopen the restart decision rather than asking communications to explain why the promise still counts as fulfilled.

Write a restart condition that can fail

For the hypothetical, avoid a condition such as managers feel supported. Specify the implementation period, the duties to be covered, and who can authorize a change to that coverage. Ask the pilot owner to record any withdrawal of cover and its effect on practice. This is an operating condition for the restart, not a wellbeing measure.

Explain what follows if the condition fails. You might pause the affected task, reduce its scope, or return the resource decision to the executive team. Choose the response before inviting employees to rely on the promise. If leadership reserves the right to proceed anyway, state that discretion and its reason openly. Calling the condition a gate while routinely bypassing it would leave employees with another gap between announced safeguards and the work they encounter.

Check the failure mode before scaling

Do not claim lessons learned without a changed operating arrangement. Avoid a fresh promise of certainty. Use employment, privacy, and technical advisers where the restarted plan requires them.

Have the responsible technical owner review the repair and employment counsel review changed job terms or obligations. Use the appropriate privacy adviser for new monitoring. Explain the limits of employee consultation before asking for input. Do not invite discussion of a contractual choice that leadership cannot offer, or imply that participation waives an employee’s existing concerns.

Review the work, then revise the arrangement

Review whether the changed condition reached the test and whether the promised report occurred. Keep the earlier failure visible enough to prevent the same dependency from disappearing again.

At the promised checkpoint, return to the precise resource or decision gap identified in the restart explanation. Show whether it was supplied when needed, what work followed, and what exception remains. Publish the checkpoint even if the pilot is inconclusive. Failing to issue an uncomfortable report would create another unfulfilled commitment inside the attempted repair.

Keep a route for an employee to say that the supposedly changed condition still does not exist. Ask for the task and timing needed to investigate, and protect unnecessary personal information. Do not require a positive view of the relaunch before hearing the report. A specific recurrence should reach the person who can pause or repair the affected work.

Edmondson’s 1999 study concerns psychological safety and team learning; Edmondson (1999): Psychological Safety and Learning Behavior in Work Teams. It does not demonstrate that a relaunch restores confidence. For this proposal, check whether employees can report that the pilot repeats an old failure. Use those accounts to inspect the work, without calling participation or silence proof of renewed belief.

Connect this decision to the next one

Place the restart’s verified difference inside the company change story. Choose the announcement channel after deciding what employees need to question. Apply the euphemism audit to phrases that blur the earlier gap. Use broken-promise repair when the relaunch is only one part of a wider obligation leadership must address.

Before requesting renewed participation, identify the earlier promise that failed and show the changed condition in the proposed test. Make a reporting commitment the owner can keep. Restart only the work those conditions support, and allow the next checkpoint to produce a decision to pause again.

Sources

Steady at the Top