We put handheld scanners on a receiving dock so receivers could be completed in real time, at the delivery, instead of hours or days later. Within a week of the WMS go-live, paperwork was stacking up next to the computer in the receiving office.
Nobody announced it. Nobody filed a complaint. Receivers quietly went back to paper, collected it in a pile, and keyed it in later when the dock was clear. Three receiving personnel, 35,000 SKUs, deliveries all day.
I was on that rollout. The WMS team owned it and I supported it on change management, which is the first thing that went wrong. Change management was scheduled as a back-end activity, something you do to people once the requirements are locked. It belongs at the front, running alongside requirements gathering. We found that out on the dock, in week one.
01Nobody Was Resisting Anything
Two things broke at once. The handhelds were top of the line, chosen by IT, capable of nearly anything, which was the problem: too much device for a person holding a clipboard in one hand. And the supplier labels wouldn't scan, because nobody had made the suppliers comply with our bar coding requirements. So one of the receiving crew stood at an open dock door fighting a device and a bad label while a driver watched.
None of this was new. In 2008 a team led by Ross Koppel published a study in the Journal of the American Medical Informatics Association on what nurses actually did with barcode medication scanning. They shadowed the work, interviewed staff, pulled the override logs. Fifteen workarounds, thirty-one causes. The causes Koppel names first: barcodes that wouldn't scan — crinkled, smudged, torn, missing, covered by another label — and devices that fought the person using them.
That's a hospital, not a warehouse, and I won't pretend the error rates transfer. The mechanism does. A person scanning against a clock, holding a label that won't read, finds another way to finish the work. Our stack of paper is Koppel's first category: the scan step, omitted where it belonged.
The workaround was accurate information and a worse way to run the process, at the same time. The stack kept material moving and kept drivers from sitting. It also meant inventory was wrong for hours at a stretch, and the count problems that created ran well past go-live. A later Dutch study of nearly 5,800 medication administrations found workarounds associated with roughly three times the odds of an administration error. Observational, and the error measured was a patient getting the wrong medication, not a pallet in the wrong slot. Read it as a direction.
A workaround tells you the truth about your design and charges you for the privilege every time it runs.
02The Pilot Passed Because It Left Out the Driver
We piloted. That's the uncomfortable part. There was a pilot, it passed, and it told us nothing.
It ran in a conference room, in the WMS sandbox, staffed mostly by IT. Clean test labels. No dock, no truck, no driver. Under those conditions the handhelds performed exactly as specified, which is how that device selection survived to go-live.
The pilot tested the software. It never tested the process.
The two conditions that killed this rollout were a label that wouldn't scan and a driver standing there waiting. A conference room holds neither.
Safety researchers call this work-as-imagined against work-as-done: the work as designed differs from the work as performed, and insisting doesn't close the distance. Somebody has to go look. Lean says it in one word. Gemba. I know the principle. We piloted in a conference room anyway.
That same Dutch group found the strongest predictor of a workaround was load, not attitude. Nurses carrying six or more patients were far more likely to work around the scanner. Workarounds are a pressure reading.
03Assumed, Not Managed
Supplier bar coding compliance was assumed. It was never managed. No owner, no measurement, and not one supplier had been asked to change anything before we handed scanners to receiving.
That's a requirements failure sitting outside your own building, which is why it gets missed. Our new process depended on a change other companies had to make, and we'd never asked them.
Big-box retail worked this out a long time ago. Walmart publishes barcode and labeling requirements and charges suppliers who miss them. The compliance program came first, the automated receiving second.
The waiting driver was a meter running. Two hours of free time at the dock is the industry convention, with detention charged after that at $25 to $100 an hour. The American Transportation Research Institute reported in 2024 that drivers were detained on 39% of all stops the prior year. I can't tell you what our carriers billed us. I can tell you receiving understood that clock better than the project did.
04What Made It Stop
Three changes. We swapped the devices for something receiving could run in the time available. We turned supplier bar coding from an assumption into a managed compliance push, with the non-compliant suppliers named. And we defined the paper fallback: a real process, with a standard, measured.
The third one is worth stealing. We stopped treating the workaround as a discipline problem and promoted it into the design: an exception path with an owner and a number on it. How often the paper lane ran told us whether the rest was working. It stopped being a secret.
Measurement is the piece most rollouts skip. Prosci's benchmarking of 2,600-plus change practitioners found 76% of those who measured compliance and performance met or exceeded project objectives, against 24% who didn't. Their own research, self-selected sample, and they sell the methodology, so weight it accordingly. The fallback got fixed once it had a number on it.
None of this argues against training. Train people. But training can't fix a device an operator can't run in the time available, and it will never make a smudged label scan. When a stalled rollout gets diagnosed as an adoption problem, that's the last thing anyone checked and the first thing that's wrong.
The fix I'd actually want came earlier than all three. The three of them knew the labels didn't scan. They knew what a waiting driver does to a dock, and what happens to the four trucks behind him. That was requirements-grade information, held by the only people who had it, and nobody asked for it until the paperwork was already stacking up.
What is your operation doing instead of the system you paid for, and who has been told?