What makes manufacturing software hard to design?
Manufacturing software connects office and shop floor roles, physical material and records, plans and what actually happened. Each connection is a place where data can go wrong.
The designer needs to understand production well enough to ask about rework, partial quantities and traceability before drawing screens.
Key facts
- One product serves roles from operator to auditor, with different horizons and devices.
- Physical conditions on the floor (gloves, noise, shared terminals) set design rules.
- Material identity (lots, serials, units of measure) runs through every screen.
- Exceptions such as short picks, scrap and rework are daily events.
- Many sectors are regulated, and one gap against a standard can end a buyer’s evaluation.
Sources of difficulty
| Source | What it means for the interface |
|---|---|
| Several roles in one product | Operators, planners, supervisors and auditors need different screens over one data model |
| Physical conditions | Large targets, visual confirmation, fast sign-in, a keyboard path that works when scanners fail |
| Material identity | Lots, serials and units shown with every quantity, conversions visible at entry |
| Exceptions as daily events | Short picks, scrap and rework designed as normal paths |
| Regulation | Complete history, corrections as separate records, reports built from captured events |
Office and floor
Office users sit at a desk, compare many records and can take time. Floor users stand, handle material, and record one step between physical tasks. A product designed only from the office point of view works in a demo and slows the floor down, and the floor is where most records are created.
A plan says what should happen. The floor records what did happen: different quantities, substitutions, delays, rework. Manufacturing software has to show both and the difference between them, and the records of what happened are only as good as the capture screens that produced them.
Common mistakes
- Designing from the office point of view only
- Treating exceptions as edge cases
- Leaving the data model until screens are done
- Showing quantities without units, or lots without their supplier lot
Where we’ve worked on this