Imagine a dispatch clerk who keeps departures on time by chasing information that should already be on the dispatch paperwork. Today, it’s a delivery booking reference. She knows which customer contact will answer, makes the call, updates the paperwork and gets the driver away before the cut-off.
The departure report records a success. Her colleagues know it took twenty minutes of chasing. The manager may know too. But unless that effort enters the conversation about how the operation works, the missing reference remains tomorrow’s problem.
That is how capable people can hide a broken process without concealing anything deliberately. Their work protects the service result, while the recurring obstacle receives less attention than its cost warrants. The business becomes accustomed to getting through the day this way.
DataDocks CEO Nick Rakovsky raised this distinction when discussing staff who repeatedly correct stock records during an automated replenishment rollout:
“We need to start doing some trend analysis of what isn’t getting picked up automatically and why it’s not.”
Every correction gets the operation moving. Taken together, those corrections can show where the operation needs to change. Someone has to make time to examine them.
Researchers Anita Tucker and Amy Edmondson observed a similar pattern in hospitals. Nurses worked around missing supplies and information so they could continue caring for patients. These immediate remedies often left the underlying problem untouched. The researchers distinguished restoring the interrupted task from taking action to prevent the problem recurring. Their study concerned nursing, but the distinction is useful when examining operational recoveries elsewhere. Tucker and Edmondson’s hospital study.
For the dispatch clerk, pursuing the missing reference is entirely reasonable. The truck is ready. The driver needs to leave. Investigating how booking information reaches the warehouse can wait. If tomorrow brings the same pressure, however, that investigation keeps losing its place. Training another clerk to make the same calls would provide cover, while leaving the avoidable work intact.
There is a difficult judgement for managers here. Removing every unusual case would be expensive and probably impossible. But a team’s ability to cope is weak evidence that a recurring problem is cheap. The cost may show up as unanswered queries, postponed checks or preparation pushed into the next shift. Ask what people had to set aside to make the successful departure possible.
A useful example of organising a different response comes from This American Life’s reporting on NUMMI, the Toyota–General Motors venture in Fremont, California. Former workers described how, under the old GM operation, defective cars continued down the line and needed repairs afterwards. The reporting then showed Toyota’s approach at NUMMI: pulling the andon cord summoned a team leader. The team tried to resolve the problem promptly; if it remained unresolved, the line stopped. Workers also described being asked for improvement ideas and seeing those ideas acted on. This American Life, “NUMMI”.
The practical lesson is that raising a problem had a response attached to it. Assistance arrived, and there was a route towards changing the work. An exception log in a warehouse needs an equivalent commitment. Otherwise, recording the rescue simply adds another task for the person already dealing with it.
There is also a reason to be careful before deciding that every workaround should disappear. Sometimes the person doing the work has found a sensible response to conditions the official process overlooks.
In a March 2026 column, logistics researcher Chris Caplice illustrates this with a driver who ignores a software-generated route because a turn is too tight for the trailer. The deviation reveals a constraint missing from the plan. Understanding it could improve the routing system for other drivers. This is an illustrative example, but it makes a useful point: enforcing the original instruction would preserve the defect in the plan. Caplice’s March 2026 column.
A February 2026 paper by Nesi Outmazgin, Pnina Soffer and Irit Hadar offers a structured way to investigate these situations. Its framework uses recurring workarounds to examine why the prescribed process fails to support what people are trying to achieve, then identify possible improvements. Developed across three organisational settings and demonstrated in a fourth, it also treats the workaround itself as a possible improvement to evaluate, rather than assuming it must be eliminated. The 2026 study.
For a warehouse manager, a practical starting point is to choose one repeated rescue and follow it far enough to make a decision. Returning to our dispatch example, pick several recent departures that required someone to chase a booking reference. Establish when the reference became available, where it was recorded and why dispatch could not obtain it through the normal process.
Those details determine what would help. Perhaps the reference sits in a customer service inbox because nobody agreed where to record it. Perhaps it is entered correctly but omitted from the dispatch paperwork. Or perhaps the customer only issues it shortly before departure, making some late coordination unavoidable. Each explanation calls for a different response.
Making the reference mandatory when the order is first created could make things worse if it does not exist at that point. Staff might hold orders unnecessarily or enter placeholders to get them moving. A change needs to fit when the information becomes available and who can act on it. The clerk’s experience is valuable evidence for that design.
Also establish what the rescue costs: the people involved, the approximate time spent, the work interrupted and the consequence if the information cannot be found. A rare, manageable exception may justify a clear fallback procedure. Repeated calls across several departures may justify changing the information flow. The decision should reflect the burden and consequences of the problem.
Nick’s point about authority matters here: “You need to have the right people in the right place that can make those decisions—and that are allowed to make those decisions.”
In this example, that might mean bringing the dispatch clerk together with the customer service manager and whoever can change the relevant system field or document. Give them a specific issue to resolve, time to examine it and a date to review the proposed change. If the clerk is expected to contribute, someone needs to cover that part of the shift. Improvement cannot depend on finding spare minutes between the same recurring emergencies.
If the business decides a manual fallback is sufficient, explain that decision too. Agree who handles it, what information they need and when it should be reconsidered. A fallback that is manageable for a few deliveries may become burdensome as volumes grow. Staff should be able to see what happened to the problem they raised, even when the answer is to keep a measured amount of manual work.
Then check whether the change actually reduces the need for intervention. Are references reaching dispatch in time? How often does someone still have to chase them? Are departures remaining reliable? Has customer service simply inherited the same searching and retyping? Fewer recorded exceptions alone would tell you very little if staff have stopped logging them or the work has moved elsewhere.
At the next operations review, take one departure that went well because somebody rescued it. Ask that person to walk through what happened, then agree who will address the recurring obstacle and when the team will hear back. Keep the successful departure in the report. Put the work required to achieve it on the improvement agenda as well.