In three decades on manufacturing floors, I’ve watched more MES implementations go dark than I care to count.
The technology was never the problem.
The system is live. The screens are on. The data is flowing. The vendor shakes hands and heads to the airport. Six months later, operators are back to clipboards.
After watching this pattern repeat across industries, system types and budget levels, I’ve concluded it’s never one thing. It’s three, and they compound.
1. The operators don’t trust the data
It’s not irrational. The first time a screen told them Line 3 was running at 94% efficiency while they were watching it choke, they made a mental note. The second time, they stopped looking. By month three, the dashboard is background noise: wallpaper with a pulse.
Trust isn’t granted. It’s earned, one accurate reading at a time. And it requires someone accountable for data integrity after go-live. That person rarely exists.
2. The standards were set wrong, and nobody fixed them
Cycle times entered during implementation reflect the best case, not the typical one. Downtime codes are so broad they’re meaningless. Shift definitions don’t match how the floor actually runs.
Garbage in, gospel out. The system isn’t wrong. It’s faithfully reporting a fiction baked in on day one. And because no one owns the standard definitions after implementation, the fiction calcifies.
3. Training stopped at go-live
Operators learned how to use the system. No one taught them how to act on what it tells them. There’s a difference, and it’s enormous.
Knowing that OEE dropped four points means nothing if there’s no meeting structure to surface it, no escalation path to respond to it, and no leader accountable for closing the loop. The information existed. The behavioral system to receive it didn’t.
The technology worked
I’ve spent years watching manufacturers invest millions in systems that were fundamentally sound, and then watching those systems slowly go dark. Not from technical failure. From the absence of the human and behavioral infrastructure required to make information mean something.
- Data without interpretation is noise.
- Interpretation without structure is frustration.
- Structure without accountability is theater.
The MES didn’t fail. The implementation architecture around it did. That’s where transformations are actually won or lost: not in the server room, but in the last 100 feet, on the floor.
Part V of the book, Technology in Its Proper Place, covers how to deploy technology so it holds. The Execution Gap by Gerald Tice explains why manufacturing transformations fail in the last 100 feet, and lays out the method for closing the gap. Get the book, or start with the free Companion Workbook.