Two lighting states, and the move between them. That move is the only thing the audience actually experiences — the states themselves are just the endpoints.
Take the same pair of states and run the fade at eight seconds: it reads as an event. Something has happened, and the audience looks up to find out what. At twenty seconds it reads as a scene change, which is to say as machinery, and the attention goes to the room rather than the actor.
Fourteen sits in a strange gap where most people do not register the change starting. They register having arrived somewhere else. That is usually what you want and almost never what you get on the first try.
Not by calculation. You run it, you watch the actors rather than the stage, and you notice the moment the room’s attention shifts. If it shifts during the fade, the fade is too fast.
That is genuinely enough to start a conversation. Dates and budget help, but they are not the first question — the first question is what the room is being asked to do.