The twenty per cent that has to go

The default failure of every first edit

It is too long. Not slightly — typically by a fifth to a third. This is so consistent that you can treat it as a starting assumption rather than a diagnosis.

The reason is structural, not a lapse of judgement. Every section in your cut is there because you had a reason to put it there. The viewer does not have your reasons; they have only what is on screen. Material that is justified by your knowledge of the project rather than by its effect on a stranger is exactly the material that feels slow.

The exercise

Take your fine cut and remove twenty per cent of its running time. Not by trimming frames — by removing whole units. A section, a sequence, an answer, a shot.

Do it as a duplicate timeline so nothing is lost. Then watch both.

The result, nine times out of ten, is that the shorter one is better and you cannot reconstruct what you removed. That is the finding, and it recalibrates your judgement permanently.

Which twenty per cent

These are the reliable candidates, roughly in order:

Anything that repeats a point already made. Especially second examples. The first example teaches; the second confirms something the viewer already accepted.

Everything before the actual beginning. Most pieces truly start one to two minutes in. Watch your opening and find the first moment where something happens. Everything before it is usually preamble.

The section you had to work hardest on. If a sequence took three days to make work, that is normally a sign the material does not support it. The effort is a symptom.

The shot you are proud of that does not serve the spine. The drone move. The slow-motion detail. The one with the beautiful light. These are darlings and they are the hardest to remove precisely because they are good — the quality is what makes you keep defending them.

Explanations of things nobody asked about. Background, context and history that you found interesting during research.

The ending after the ending. Most pieces end and then continue for twenty seconds. Find the last moment that matters and cut there.

Why it is hard, specifically

Two mechanisms, both worth naming because naming them helps.

Sunk cost. You spent four hours on that sequence. Those hours are gone whether you keep it or not, and keeping a weak sequence does not recover them — it spends the viewer's time as well.

You are the only person who will ever see the alternative. Nobody watching the finished piece knows what was removed. There is no version of the audience that misses your drone shot. The only person who experiences the loss is you, and you are not the audience.

The parking file

The practical defence against the pain: never delete, park.

Make a timeline called offcuts or parked and drop everything you remove into it. It costs nothing, it takes five seconds, and it removes the finality that makes people fight to keep things.

Twice in a career you will go back and retrieve something from it. The rest of the time it simply makes the removal easy, which is its actual job.

The parked material is also genuinely useful afterwards: it is where the short-form cuts, the social clips and the trailer come from.

When length is not the problem

Two honest exceptions.

When the piece is already short. A ninety-second video cut to seventy-two seconds is a different kind of decision, and at that length the risk is under-explaining rather than over-running.

When the slowness is deliberate and load-bearing. A long take of somebody thinking, a landscape held past comfort, a silence after bad news. These are slow on purpose, and the twenty per cent exercise will suggest cutting them because it cannot tell intention from padding. Protect them consciously — by which is meant: decide, in words, why each one is there, and if you cannot, it goes.

Applying it to somebody else's cut

When you are asked to shorten a piece you did not make, the same order applies, with one addition: ask what the spine is before you start. Without it, the shortening is guesswork and you will remove the thing the piece was for.

Today

Duplicate a finished timeline, remove twenty per cent of it by whole sections, and watch both versions back to back. Whichever wins, you have learnt something you cannot learn by reading.