I think the biggest question, for me, when creating a scene is one of evaluation. How do I know if I'm filling my outline with useful scenes or junk? In the past, the old "write it and see" test worked well enough. But that a waste of time! There has to be a better way, right?
I've found a few things that help me do that quicker. This list is still under construction for now, but this is what I have so far.
Conflict and Stakes
I list both conflict and stakes because it's possible to have conflict without stakes, and that isn't too interesting. Some will say that each scene must have conflict, but that's a complex notion, more complex than it might seem at first. A scene may carry conflict from a plot point forward, or it may be building towards—promising—conflict in the future. Still, if a scene has neither conflict nor stakes, something is off.
Relevance and Focus
Lajos Egri talked about a dramatic premise. Others have offered variations on a logline or a thesis. There's a core to the story. If not, there's a danger of meandering and slogging. A scene should fit into the scope of that core idea that defines the story, whatever name you call it.
Questions and Curiosity
I hate a story that's so lousy that I just want it gone, except it's left me just curious enough about one or two things that I want to know how it ends too. It's a chore to read, so I skim pages looking for a few answers. True story—I've Googled books before, looking for a summary so I could stop reading.
It's a good lesson, though. Narrative questions keep the reader coming back. Combine that with strong prose and engaging scene construction and you've got a winner. A scene should be viewed as a landscape of questions and answers. How soon are questions posed? Which questions are answered? What is yet to be answered—perhaps an answer is teased?
And pondering the questions might show you some flaws in the scene. "Why is Alex acting like this" might be the question you want to leave hanging for a few scenes, or it might be killing your story.
An online friend told me recently that a scene should look like a dinosaur, skinny at both ends and fat with questions in the middle. (Never accused him of being brilliant with the metaphors.) But a scene that asks no questions and answers no questions falls back on the relevance and focus question.
Originality and Surprise The whole scene might surprise a reader, or just a single moment. This might seem contrary, but a good story can still be told in boring and predictable ways. Look for ways to surprise the reader. Keep them guessing, whether it's the setting, or the choice of POV or part of the narrative mode, or a line of dialogue, or some other detail.
Now, there's a host of other things I could explore, like imagery and theme, or characters, or dialogue, or description, or POV choices and so forth, but for me Stakes, Focus, and Originality are the keys to evaluating a scene down to a basic Red Light/Green Light purpose. Those other things are important decisions, but don't lend themselves to "is it working?" as well.
Writer of speculative fiction. Current work in progress is an untitled dark fantasy novella.