Name the question

Someone has said "we need a digital twin" or "can we model this?" Before any data or software, write one sentence that starts "We need to know whether…" and ends with a decision. If you cannot, you do not yet have a question; you have an interest, and a model will not sharpen it.

The sentence

Three parts, in writing, agreed by the person who will use the answer:

1. The question. "Which streets flood in a storm we see once in twenty years." Not "flooding".
2. The decision it informs. "…so that we can choose which three culverts to replace this year."
3. Who acts, and when. "The engineering manager, before the budget is set in March."

If the question has no decision behind it, it is research or curiosity. Both are fine, and both are cheaper than a model that pretends to be for a decision.

Split it

Most questions are two or three. "Will more trees cool the neighbourhood and what will it cost?" is a microclimate question, a budget question and a maintenance question, each answered differently. Write each one down. The question pages on this site each do this split for a common question, and reading the one nearest yours is a good way to find your own.

Decide what good enough looks like

Before you build, say what answer would change the decision. "If the model shows fewer than two streets flooding, we do nothing." A model that cannot tell those two cases apart is not worth building, however beautiful. This is also the sentence that stops a project growing forever.

Find out who already answered it

Someone has usually modelled a place like yours. Look in digital twins on this site, and ask your regional agency or the nearest university department. Their model may be reusable, and their mistakes are free.

Write it where people will see it

The sentence goes at the top of the project document and is read aloud at the first meeting. When the project drifts, and it will, this is what you read again.

Next

Talk to the people affected: they will tell you whether the question is the right one.