Six shapes, and what each one costs
Shape is a choice, not a default
Most unstructured material gets cut chronologically because that is the order it arrived in. Chronology is one option out of several, and it is frequently the weakest. Here are the shapes that recur, what each is good at, and the specific way each fails.
Chronological
Events in the order they happened.
Good for: process films, events with a genuine arc, anything where causation runs forward and the viewer needs to see it.
Fails when: the interesting thing happens at the end. An hour of setup for a two-minute payoff loses its audience before the payoff. The standard repair is a cold open — put the payoff at the front, then "three months earlier".
Thematic
Grouped by subject: here is the money problem, here is the staffing problem, here is the regulation problem.
Good for: explanatory work, anything with several independent strands, interviews where several people address the same topics.
Fails when: nothing connects the sections, so each transition is a full stop and the viewer can leave cleanly at every one of them. Repair: end each section on a question the next one answers.
Problem and solution
Establish something wrong, then work toward the fix.
Good for: tutorials, marketing, case studies, anything instructional. It is the most reliable structure for teaching because it supplies motivation before information.
Fails when: the problem is manufactured. A viewer who does not recognise the problem as theirs disengages immediately, and an exaggerated problem reads as a sales pitch. Repair: state the problem in the audience's own words, not the product's.
The essay
A claim, then evidence, then a conclusion. Often narrated.
Good for: argument, analysis, opinion, video essays.
Fails when: it becomes a lecture with pictures. The remedy is that every claim needs something on screen that is the evidence, not a stock shot of somebody typing.
The profile
A person, through what they do.
Good for: almost anything, which is why documentary reaches for it constantly. People are interesting and processes are not, so attaching a process to a person is the most reliable move in the form.
Fails when: the subject is not candid, or the film becomes a tribute. A profile with no difficulty in it is an advertisement.
The list
Five tips, seven mistakes, ten things.
Good for: search-driven and feed-driven work. The viewer knows what they are getting and how long it will take, and they can leave and return.
Fails when: it is the default rather than a choice. A list has no momentum — each item is independent, so there is never a reason to watch the next one. It also flattens material that had a real structure in it. Repair, if you must list: order the items so each builds on the last, and say so.
Choosing
Three questions:
- Does the material have a hinge? If yes, chronological or profile will probably use it. If no, thematic or essay will hide its absence.
- What does the viewer want? Somebody searching "how to fix a dripping tap" wants problem-and-solution starting immediately. Somebody who followed a link about a printing workshop wants a profile.
- How long is it? Under two minutes, almost everything works as a single unbroken idea. Over eight minutes you need sections, and sections need a shape.
Sections, and the joins between them
Whatever shape you choose, a piece over about five minutes breaks into sections of two to four minutes. The joins are where you lose people, so they need work.
Things that make a section join hold:
- A question at the end of the outgoing section, answered at the start of the incoming one.
- A change of texture — from interview to sequence, from voice to music, from close to wide. Sameness across a join makes the join feel like more of the same.
- A hard reset, used sparingly: black, a title, a beat of silence. This works precisely because it is a full stop, so use it where you want the viewer to reorient.
What does not work: a slow fade to black and up again, which reads as the end of the piece, and a narrated announcement of what the next section will contain.
Restructure on cards, not in the timeline
When a structure is not working, do not push clips around. Write each section on a card or a line in a document — one line, what it is, how long it is — and rearrange those. A twelve-minute piece is fifteen lines. You can try six structures in twenty minutes on paper and none in twenty minutes on a timeline.
Then implement the winner once.
Today
Take a piece you have made and write its sections as one line each. Then try re-ordering them into a different one of the six shapes above. You will usually find at least one arrangement better than the one you shipped.
The one thing to keep
Chronology is a choice rather than a default, and each shape has a specific failure — thematic sections give the viewer a clean exit at every join, lists have no momentum, and essays become lectures unless every claim has evidence on screen.
Before you move on
A ten-minute explainer is cut thematically into four sections. Retention shows a sharp drop at each section boundary but is healthy within sections. What does that pattern indicate, and what is the structural repair?
Pick the one you would defend. Nobody sees your answer.