The launch went well. People were engaged, the reasoning was accepted, the early weeks produced visible activity.
Somewhere around the fourth month it stops. Nobody announces opposition. Attendance at the meetings declines, the old process reappears alongside the new one, and the leader finds themselves asking about it more often.
Why it happens there
The timing is not arbitrary. Three curves intersect.
The novelty has worn off, so the change is no longer interesting and is now simply extra work.
The costs have fully arrived. Everybody is doing the new thing badly, which is slower than doing the old thing well, and the difficulty is being experienced daily.
And the benefits have not. Most organizational changes return something on a horizon of six months to two years. At month four the ledger shows all of the cost and none of the return, which is not a perception problem. It is an accurate reading of the current position.
People are therefore behaving rationally when they revert. They are optimising against the evidence available to them, and the evidence at that moment genuinely favours the old way.
The leader’s attention has also moved
The other half of the collapse is on the leader’s side and gets less examination.
By month four the person who initiated the change has been living with it for considerably longer, since they were thinking about it before the launch. They are bored of it, they consider it settled, and something else has become urgent.
So attention withdraws at precisely the moment the organization most needs to see that this still matters. The withdrawal is read accurately: if the leader has moved on, the change was a phase, and reverting carries no risk.
A change programme requires the leader’s visible attention for roughly twice as long as they expect to give it, and the second half is the part that determines whether it holds.
The signs
The collapse is quiet and it announces itself in specific ways.
The old system runs in parallel, described as a temporary safeguard. Exceptions are granted, each for a reasonable case. Compliance persists where it is checked and lapses where it is not. Questions about the change stop arriving, which is read as acceptance and is usually disengagement. And nobody has mentioned it in a meeting for three weeks.
The parallel system is the most reliable indicator. Where the old process still exists, it will win, because it works and the new one does not yet.
What to do about it
Say it will happen, in advance. At the launch: around month three or four this will feel like a mistake, because we will be paying all the costs and seeing none of the return. That is expected and it is not evidence the change is wrong. Naming it beforehand converts the experience from evidence into a milestone.
Remove the old system. Not gradually. As long as it is available it will be used, and the change is competing against something that works.
Find and publish an early indicator. The real benefit is eighteen months out, but something measurable moves sooner: a cycle time, an error count, a complaint volume. Whatever it is, report it, because the organization needs evidence and currently has none.
Increase your visible attention rather than reducing it. Ask about it in the meeting where you would normally have moved on. This is the single most effective intervention and it costs nothing but the discipline of continuing to care about something that has stopped being interesting.
Distinguishing collapse from a correct rejection
Not every reversion is fatigue. Sometimes the change is wrong and the organization has found out before the leader has.
The distinguishing evidence is whether people can articulate a specific defect. Fatigue produces vague dissatisfaction and quiet non-compliance. A genuine problem produces particular objections: it does not handle this case, it takes four times as long for that category, it breaks the thing downstream.
Those objections are frequently correct and they come from people closer to the work than you are. Treating them as resistance is the error the next entry addresses, and it converts recoverable information into an argument.
The useful question at month four is therefore not how to push harder. It is what specifically is not working, asked of somebody doing it daily, with the willingness to alter the design in response.
Edited by Patrick J. Wolf, PhD