Where the ninety seconds actually go

Most short films spend half their length on setup

Count it on any amateur brand film. Forty seconds of establishing: the exterior, the sign, the door opening, somebody walking in, the owner sitting down and being introduced. The actual subject arrives at 0:45 and there are fifteen seconds left for it.

This happens because the maker experiences the setup as necessary context and the viewer experiences it as a delay. Both are right about their own experience. The viewer wins, because they have the option of leaving.

A workable division for ninety seconds

This is a starting point to break deliberately, not a template.

• 0:00 to 0:03: the promise. What this is and why it concerns me. Usually the most surprising true statement you have, or the strongest image. Not the logo, not the establishing shot.
• 0:03 to 0:20: the situation and the problem. Who, and what is wrong. Seventeen seconds is enough for both if you stop describing and start stating.
• 0:20 to 0:60: the turn. Something is attempted; it goes wrong or it costs something; a second attempt works. This is the majority of the piece and it is where amateur films are thinnest.
• 0:60 to 0:80: the new state. What is different now, shown rather than claimed.
• 0:80 to 0:90: one ask. One. Named, specific, achievable in the next ten seconds.

The useful discipline here is that the turn gets the largest share. If your turn is eight seconds and your setup is forty, you have the proportions of a film that will be watched by your client and nobody else.

Two numbers that make scripting concrete

Speech runs at roughly 140 to 160 words per minute in natural, unhurried delivery. So ninety seconds is about 210 to 240 words of voiceover: roughly one side of A4 at normal type size. Write your script, count the words, and you know whether it fits before you record anything. Almost every over-long edit was over-long in the document.

Shot length: a comfortable average for this kind of piece is two to four seconds, which gives 25 to 40 shots in ninety seconds. If your shot list has twelve items, expect a slow film; if it has ninety, expect one nobody can follow. This is a rough guide rather than a rule: a single unbroken shot can carry a whole piece, and does, in the hands of somebody who planned it that way.

The same shape compresses and stretches

At fifteen seconds you get roughly: two seconds of promise, five of problem, five of turn, three of new state and ask combined. There is no room for a separate ask, so the ask has to be inside the new state: the phone number on screen while the result is shown.

At six minutes the shape does not change; the middle subdivides. The turn becomes three or four linked state changes, each with its own small reversal. What does not happen is that the setup grows proportionally. A six-minute film can still put its problem on screen in the first twenty seconds, and the good ones do.

Write the beats before the shots

One line per beat, in plain sentences, before any camera or software is involved:

1  Sign was made the same way for 40 years.
2  BUT the road widening cut the shopfront by two metres.
3  THEREFORE the old layout no longer fits and trade drops.
4  They try shrinking the same design. It looks cheap.
5  BUT the apprentice suggests splitting it across two panels.
6  THEREFORE the sign now works and reads from further away.
7  ASK: bring us your awkward shopfront.

Seven lines. You can rearrange, delete and test this in two minutes; you cannot do that with a rough cut. Every hour spent here saves several in the edit, and the beats document is also the thing you show the client before shooting, when a change is still free.

The limitation

This structure suits persuasion and explanation, which is most commercial work. It is not universal. Observational documentary, a mood piece, a piece of music promotion or an experimental short may deliberately withhold structure, and imposing a turn at 0:20 on material that is about stillness ruins it.

And a well-proportioned film about nothing is still about nothing. Structure allocates attention; it does not manufacture a reason for the attention to be given. If the beats document reads as boring in seven plain sentences, no amount of grading, music or camera movement will fix it. And knowing that on day one rather than in the edit is the entire point of writing it down first.