A digital twin project can begin with a temptation: model everything. The result may be visually impressive, yet difficult to use. Our starting point is narrower: identify a decision and work backwards.
Name the decision
“Understand the building” is a broad ambition. “Compare two ways of operating this space” is a question a team can investigate. Write down who makes the decision, how often they make it and what evidence would help.
Draw a boundary
Every model leaves things out. Decide which assets, relationships and timescales matter for the first question. Keep a separate list of what the prototype does not represent. That list is part of the model, not an embarrassment to hide.
Define a useful first result
A useful first version might connect a small number of records, make missing data visible or compare a limited set of assumptions. Agree on what would count as evidence of usefulness before adding more detail.
Keep people in the loop
The people who operate a system know its exceptions. Bring them into review early, use their observations to challenge the model and record disagreements. The goal is a clearer conversation about reality.