Almost every editing tutorial I have watched starts with material already roughly in order and proceeds to refine it. The stage before that, where you decide what the thing is, gets skipped almost entirely.
It is the stage that determines whether the edit works, and doing it badly means spending weeks polishing something that was never going to hold together.
What an assembly actually is
Everything you might use, in order, at full length, with no attention paid to how any of it cuts.
It is long, ugly and unwatchable. A twenty minute finished piece might come from a ninety minute assembly. That is normal and it is the point.
What it does is make the shape of the material visible. You cannot see the structure of a piece while the material is in bins. You can see it when it is laid out in a line.
Why people skip it
Because it feels like wasted work. You are laying down material you know you will cut, and the instinct is to make decisions as you go and save the trouble.
The problem is that decisions made shot by shot are local decisions. You are deciding whether this piece is good, not whether the piece belongs where it is or whether the section it is in should exist.
I have edited whole sequences beautifully and then removed them entirely, and every one of those was a sequence I fine-cut before I knew what the structure was.
Watching your own assembly is where the work happens
The actual step, which sounds trivial and is not, is sitting down and watching the whole assembly without touching anything.
No pausing to fix something. No adjusting an edit point. Watch it, take notes, and only then start working.
The notes from that pass are almost always about structure. This section is in the wrong place. This point is made three times. This thing that I thought was the centre of the piece is boring. This throwaway moment is the best thing in the material.
Those are the observations that change the piece, and they are only available when you are watching rather than working.
The order in which I now do things
Assembly, at full length, in roughly the order the material suggests.
Watch it, take notes, do nothing else.
Restructure at the section level. Move whole blocks around, delete whole blocks, without touching individual cuts.
Watch again. Repeat until the shape is right.
Only then start cutting within sections, tightening, trimming, choosing between takes.
And fine detail — frame-level trims, transitions, the last few seconds off each section — right at the end.
The discipline is not going down a level until the level above is settled, and the temptation to skip ahead is constant.
Documentary and interview work especially
This is where it matters most, because the structure is not written in advance.
With interview material I now build what people sometimes call a paper edit or a radio edit. Take the transcript, mark the passages worth using, arrange them in a document until the argument works, and only then assemble the video.
Working in text is enormously faster than working in a timeline for this stage. Reordering paragraphs takes seconds. Reordering timeline sections takes minutes and involves fiddly technical work that distracts from the thinking.
And a structure that does not work in text will not work on screen. Discovering that in a document costs an afternoon. Discovering it in a timeline costs a week.
Length as a diagnostic
A useful check I picked up. Note the assembly length, and note the target length.
If the assembly is twice the target, that is comfortable — you have choices.
If it is barely above the target, you are in trouble, because it means you will be keeping nearly everything and you have no ability to cut what is weak.
If it is five times the target, you may not have a structure, you may have a lot of material, and the assembly stage is where you find that out rather than three weeks later.
Showing the assembly to someone
Controversial, and I have come round to it.
The instinct is to hide the ugly version and show something polished. But feedback on a polished cut is about details, because that is what is visible. Feedback on an assembly is about structure, because that is all there is.
Structural feedback is worth vastly more and can only be acted on early. A director who says the middle does not work at assembly stage saves a fortnight. The same note on a fine cut is devastating.
The trick is warning people about what they are watching, so they do not spend the whole time noting that the audio is not mixed.
The general point
Editing is not primarily about cuts. It is about what is in the piece, in what order, and how long each thing lasts.
Cuts are the last ten percent and they get ninety percent of the attention, probably because they are the visible craft and the structural work looks like sitting there doing nothing.