What stalls manufacturing IoT projects — and the fix for each.

On a factory floor, the obstacles are rarely technical capability. They are access to the machine, agreement about what a number means, and whether the shift can act on it before the end of the day.

What goes wrongHow you noticeThe fix
Retrofit assumed to need PLC accessThe project stops at IT and controls approvalsStart with the least invasive route that answers the question — many losses are visible without touching the control system at all
Counting the wrong eventThe system’s output disagrees with the operator’s tallyCount what the customer pays for, as late in the line as possible
Definitions differ between shiftsThe same line reports differently depending on who is onWrite down the changeover, stoppage and rework rules once, and apply them everywhere before comparing anything
Data arrives after the shift endsNothing changes on the floor, only in reportsPut the number where the shift can see it during the shift. Information that arrives after the decision is history
Old machines treated as out of scopeThe worst assets stay invisibleAge is rarely the obstacle. Current, vibration and simple stop/run sensing work on machines with no digital interface at all
Pilot on an easy lineThe result does not transfer to the hard onePilot where the problem is. A clean line proves the technology and settles nothing
Maintenance and production measured separatelyBoth teams can show they are doing well while output does not moveShare one set of numbers. Availability is a maintenance result and a production decision at the same time

The two we see most often are the invisible old machine and the easy pilot. Plants tend to instrument what is convenient to instrument, which is usually the newest asset with a network port — and then report that the data did not reveal much. The losses generally sit on the machine nobody wanted to touch, and reaching it is a sensing problem rather than an integration one.