Anyone who wants to harmonise several plants digitally knows the pattern. Headquarters decides on a standard, the sites receive the requirement, and a year later local solutions are running again: spreadsheets, scripts, small databases. Officially everything has been implemented; in reality every plant works differently.
Why the islands exist
At every site there are people who have built solutions over years to keep production running. A dosing scale that is only connected this way there. A sequence that suits exactly this reactor. A hand movement that is in no instruction and still saves the batch.
That is not resistance to change. It is the factory’s will to survive. Anyone who treats these solutions as a disruption takes something away from the plant without giving it anything better, and gets the receipt in the form of pushback.
A model that takes both seriously
I work with a rule I call 80 to 20. The numbers are not a metric but a ratio by which decisions can be aligned.
The shared core, about four fifths. One standard applies to all sites here:
- master data with the same meaning everywhere,
- interfaces that are built once and then reused,
- traceability: logs, change tracking, permissions,
- a few central indicators that management really needs.
The local freedom, about one fifth. The plant decides here:
- plants and devices that only exist there,
- site-specific procedures and sequences,
- evaluations that matter for the work on site.
The freedom is not a lawless zone. It is defined so that deviations are visible and justified rather than growing in secret.
Rules for the freedom
In regulated environments, the local space must not break traceability. That is why the fifth also needs clear guard rails:
- Every local solution has an owner and a short description of what it is for.
- Solutions that change data or influence quality decisions go through the usual change process.
- Recurring local solutions are assessed regularly: does this belong in the core?
That way, good ideas move from the plants into the standard, and the core gets better instead of rigid.
Harmonisation is an offer
The insight that makes the difference: you do not harmonise by instruction but by offer. A shared standard prevails when it noticeably eases the operators’ daily work. Then they give up their own spreadsheets voluntarily, because the standard is better.
That requires clarifying the benefit on site before the roll-out starts: which work becomes easier, which evaluation disappears, which double entry goes away?
What follows
- Look at the islands first. An inventory of local solutions shows what plants really need.
- Keep the core small. The less it prescribes, the better it can be enforced.
- Name the freedom. Where may the plant decide for itself, and where not?
- Benefit before obligation. A standard that takes work away needs less enforcement.