The question is not whether a digital twin is useful, but which perimeter to pick first so the project survives beyond the demonstration.
The trap of a perimeter that is too wide
The request often arrives phrased like this: "we want a digital twin of the plant". That is a three-year project that will die in its second year, having spent its budget on data collection before producing a single decision. A useful twin starts small and is judged quickly.
Pick the perimeter by decision frequency
The right first perimeter is not the most strategic one, it is the one where decisions are most frequent. A bottleneck station whose parameters are adjusted weekly will produce more value than an entire workshop nobody ever changes. Our rule: if no decision has been made on that perimeter in six months, a twin will not help there.
Start with the data you already have
Most workshops hold more data than they think: machine counters, CMMS history, quality records, production files. It is scattered and poorly qualified, but it exists. Building the first twin on that base costs a fraction of an instrumentation campaign, and reveals exactly which sensors are genuinely missing.
The pivot identifier is the structuring decision
The technical choice that matters most is neither the platform nor the 3D engine: it is the identifier linking the modelled object, its CMMS record, its measurements and its history. Without it you get three systems talking about the same equipment without ever being able to confirm it. With it, every later addition connects without rework.
Judge on a decision, not on a demo
A twin proves itself the day an operational decision is taken from it — a setting, a planned stop, a re-layout. As long as the project only produces screenshots for the board, it remains an expense. So we set that exit criterion during framing: which decision, taken by whom, by when.
What has to be accepted
A digital twin is never finished nor perfectly up to date. Installations change, sensors fail, modifications are made without being documented. The right reflex is not to aim for completeness but to define, perimeter by perimeter, the freshness the intended decision actually requires.
Published 18 July 2026 Discuss this topic with us
