A process should have to explain what it is doing here.
Not with a flowchart. Not with a statement about consistency. With a useful answer to a plain question: what gets better because we do it this way?
Consider a fictional example. A customer asks for a small change. The request passes through seven people, two spreadsheets, a meeting, and a person who is away until Thursday. Everyone involved agrees that several steps add nothing. Removing them, however, requires another meeting.
The customer wanted a correction. We have provided a pilgrimage.
Nobody in this example needs to be stupid or malicious. Each person can be following instructions, protecting a legitimate responsibility, and doing their best. The result can still be ridiculous.
That is what makes broken processes so durable. Local obedience can coexist quite comfortably with collective failure.
Busy is not the same as useful
A workflow can produce an impressive amount of activity without moving the work very far.
Requests are acknowledged. Tickets change status. Someone updates a tracker to explain why another tracker has not been updated. The organization can demonstrate that people followed the procedure while the person waiting for the result experiences only delay.
The procedure has become its own customer.
The cost lands somewhere. A client waits. A colleague repeats an explanation. Someone checks the same information a third time because the first two checks did not travel with the request. A small defect becomes a support issue, which becomes evidence that we need another checking step.
Before long, the process is expanding to manage the problems created by the process.

Why reasonable people defend it
Sometimes the missing explanation is a good one.
A review may prevent an expensive error. An approval may establish who can commit money. A seemingly duplicated check may catch a different kind of failure. Someone with more history may know exactly which disaster inspired that annoying form.
Ask before removing it. People doing the work deserve more respect than an outsider announcing that everything looks easy.
But history explains how a step arrived. It does not automatically explain why the step still belongs.
The constraint may have changed. The risk may now be handled elsewhere. The person who introduced the step may have left, while their precaution has acquired the status of natural law.
There are less comfortable reasons for defending it, too. A workflow may be the only part of a job someone controls. Changing it may expose the limits of a past decision. A familiar task may provide visible evidence of usefulness, even when the task itself is unnecessary.
Those are understandable human pressures. They are poor measures of whether the workflow serves its purpose.
The workaround makes the system look healthy
Broken systems often survive because capable people quietly finish the job.
They know whom to call. They catch the missing information before it reaches the customer. They remember the exception nobody wrote down. They translate a request into language the next department will accept.
Their intervention becomes part of the real process, while remaining absent from the official one.
Then someone points to successful delivery as proof that the procedure works.
It may be proof that a person compensated for it. Those are different findings, with different consequences for staffing, reliability, and the next time that person is unavailable.
This is the same dependency described in The Load-Bearing Employee: competence can conceal the weakness it is holding together.
Make the step explain itself
When a process causes recurring friction, start with its purpose rather than its reputation.
What outcome is this step meant to improve? What failure does it prevent? Who uses the information it produces? Is that information already available? Who carries the cost when the step delays everything else?
“We need accountability” is a starting point. It is not yet an explanation of why three people must copy the same number into three places.
Likewise, “it has always been done this way” describes persistence. It says nothing about performance.
A useful answer should connect the activity to an outcome. If nobody can make that connection, investigate. Do not assume the activity is worthless, but do not award it permanent residence because asking questions feels uncomfortable.
Try something small enough to learn from
The alternative to defending every step is not tearing everything down on a Friday afternoon.
Choose one recurring problem. Record the current result: elapsed time, errors, repeated work, and the effort required from everyone involved. Include the people receiving the output, not just the people processing it.
Then change one thing in a bounded trial. Perhaps one approval can happen at the point of decision. Perhaps information can travel with the request. Perhaps two checks can become one check with a clear owner.
Keep the control if it serves a real constraint. Define what would make the trial stop, and how to restore the previous arrangement. Give people time and authority to participate instead of adding improvement work to an already impossible workload.
Afterwards, compare the results. Did the change remove work or merely move it to someone quieter? Did errors fall, or did they become harder to see? Did the customer get a better result?
Keep what improves the outcome. Revise what does not. Record why, so the next person inherits an explanation instead of a ritual.
Your process should be allowed to retire
There is dignity in building a useful system. There is also dignity in recognizing that the system has outlived the problem it solved.
Changing a workflow does not erase the judgment that created it. Good judgment includes noticing when circumstances have changed.
The question is not whether the process deserves loyalty. The question is whether the people depending on it are being served.
If the answer is no, protecting the procedure is not protecting the work.
Let the process change. Give the people doing the work a better chance.