Write a Treatment for a Commercial on a Public Display

A treatment for a public screen begins with the sidewalk, not the opening shot. Before you decide what happens in the film, you decide where the screen is, who is in front of it, how far away they are, which way they are walking, and what they can hear. The film follows from those facts. When it is written the other way round — a nice sequence first, a screen second — the treatment usually ends up describing a short online film and hoping the setting will behave like a laptop.

The decision most treatments for this placement have to make is structural: is the piece one continuous progression that builds toward a reveal, or a set of self-contained actions that each stand on their own? The second is usually the safer answer when people arrive at random. The first can be the better answer when a real audience waits. Which one is right should come out of the site, not out of preference, and the rest of this article is about how to get there.

Describe the encounter before you describe the film

Open the treatment with a short, factual paragraph about the viewing situation. Where does the screen sit? How high off the ground? Is it behind glass, in daylight, next to other screens? How far is the nearest place a person could plausibly stand still? Is there sound, and is there any route by which sound could reach anyone?

Then sort what you have into what is known and what is not. If the brief does not supply a dwell-time figure, write "not supplied" and ask for the observation that would produce one. An invented average is worse than a blank. A number like "viewers stay about eight seconds" has the texture of rigor, so it quietly governs everything after it: the length, the amount of text, the number of beats. If nobody measured it, you will have shaped the film around a fiction that now looks like a specification.

Name the kinds of arriving people, because they are not one audience. There are people passing, who will not stop and did not come for this. There are people waiting — in a queue, on a bench, in a car, at a counter — who will be in front of the screen for reasons that have nothing to do with the film. And there are people who deliberately stop, who are the smallest group and the one most treatments are secretly written for. Each has a different amount of time and a different willingness to spend it.

The point sounds obvious and is constantly ignored: a public display is a physical encounter, not a differently sized export. Different aspect, different distance, different light, different sound — often no sound at all — and different company, since whatever is on the neighboring screen is competing with you in the same field of view.

A treatment for this placement can stay short if it keeps four things straight:

• The encounter — where the screen is and who ends up in front of it.
• The action — what happens, in order or in repetition.
• The entry test — what survives arriving late.
• The presentation — what has to be legible at that distance and in that light.

Two pages can do it. The order matters more than the length.

Find what a partial arrival can understand

Once the action is written down, test it the way the placement will. Take the sequence and mark three or four entry positions. Not your favorite moments — arbitrary ones, because that is how arrival works. Read forward from each and write down, in plain sentences, what a person who saw nothing before this moment would believe is happening. Then sort what you wrote into three piles.

Carried. Understood without any earlier material. The action explains itself.

Restored. Only makes sense if you saw the earlier moment. This is the category that decides most of these treatments.

Lost. Not recoverable within the time that remains after the entry point.

The repair options are limited and each has a cost. Move the premise earlier so fewer entries land before it. Repeat the premise physically rather than verbally — a second broken object, a second hand reaching for a tool. Make the visual action itself carry the meaning, so the viewer does not need to have been told. Or accept the dependency and cut what depends on it.

There is also an honest fourth option: decide that something is a bonus. Some information is genuinely rewarding for a person who saw the beginning and unnecessary for a person who did not. The failure is not having bonuses. The failure is not knowing which pile a piece of information is in, and treating required material as though it were a treat.

Two cautions. First, on-screen text is a dependency like any other. A product name that holds for half a second is required information at close range and may be required information that never arrives at ten meters. Whether a given size and hold time reads from a given distance is a site question and a test, not something to settle with confidence inside the treatment. Second, do not rely on a second pass to repair a first encounter the viewer may never complete. A second pass exists only if the person is still there. Many will not be.

Compare a progression with recurring complete actions

A continuous progression earns its shape. It builds, it can hold a genuine reveal, and it rewards a longer stay with something a late arrival cannot get. The cost is where it puts its meaning. If the reveal sits at the end of a forty-second film, then the people who see the end are the people who missed the middle, and the people who saw the middle never learn what the film was for. A progression concentrates its payoff at one position in time. On a public screen, that position is a gamble.

Repeated self-contained actions work differently. Each segment shows a complete unit: the problem, the visible change, the identity. A person can arrive at any of them and get a whole idea rather than the back half of one. The costs are real. There is less cumulative build, so the piece can feel like a set of variations rather than a story. Repeating the identity in every segment can feel insistent. And if the segments are near-identical, the screen risks becoming wallpaper — the second segment may teach nobody anything because the first said everything and the viewer has already looked away.

One clarification that saves a lot of confused conversation: joining a loop seam cleanly and being enterable at an arbitrary moment are different problems. A loop that runs end to start without a visible jump has solved the join. It has not solved the person who looked up thirty-one seconds into a forty-second cycle. That person still missed thirty-one seconds. Seam craft and entry craft are separate, and this article is about the second.

Neither route guarantees attention. Repetition does not compel watching; a reveal does not compel staying. What the structure decides is how much is lost when someone arrives late, and whether the loss is recoverable in the next few seconds or gone for good. That question is answerable from the encounter, and the answer often changes when the site shows a genuinely stationary audience. If people really do wait — a counter, a queue, a waiting room — a longer build has somewhere to land, and the progression becomes the better choice. If everyone is walking, the entry test governs.

Carry the encounter into the proposed presentation

Describe the action at the scale it will actually be seen. What reads from several meters away: a silhouette, a change of state, one moving object against a still field. What does not: two simultaneous actions, small type, subtle gradations, a face doing something quiet. A treatment that specifies a glance between two people at distance from a moving viewer has specified something that will not survive the trip.

Words need their own decision. If sound is unavailable, on-screen text is the only language channel — and it is also the least reliable element at distance and the hardest thing to catch mid-stride. That argues for fewer words, held longer, each with a job. It also argues for checking whether the proposition can be carried without words at all, with text as confirmation rather than as the thing that makes the film make sense.

Do not reach for legibility numbers. There is no universal text height that works from every distance; it depends on the screen, the resolution, the distance, the light, and the viewer's sight. Name the requirement — "the service name must be legible from the far side of the sidewalk" — and then get it checked on site rather than resolved in the document.

Before anyone approves the design, ask for a representative site visit and a playback review. A desktop preview is a rehearsal of sequence and timing, and a useful one, but it does not reproduce scale or viewing distance, and it cannot establish physical visibility, glare, or accessibility compliance. Keep three judgments separate: whether the idea reads (creative), whether the screen meets the venue's requirements (technical acceptance), and whether the work reaches people who cannot see or hear it fully (accessibility). One approval does not stand in for the other two.

A worked example: the workshop window

The following brief is invented, to work through the method. Nothing here has been installed, measured, or shown to anyone.

The brief. A screen sits beside the window of a repair workshop, facing a sidewalk. Two groups pass in front of it: people walking by, who will not stop, and customers waiting nearby, who are in the area for other reasons and may glance up. Sound is unavailable.

Version A — one continuous progression, single delayed reveal, 40 seconds, silent.

• 0:00–0:07 A benchtop. A kettle, lid off, turned over. No words.
• 0:07–0:24 Close work on a small switch assembly: a burnt contact cleaned, a new part seated. Tight enough that only hands and mechanism are visible. The workshop is not yet shown.
• 0:24–0:33 The kettle is filled and switched on. Steam.
• 0:33–0:40 Pull back to the workshop window. The shop's name on the glass, the workshop visible inside.

Version B — three self-contained actions, 13 seconds each, silent, each showing a broken object, a visible change, and the service identity. The three run in order and repeat as a 39-second loop.

• 0:00–0:13 A kettle with a dead switch. The switch is replaced. The kettle boils. The workshop sign is visible in frame throughout.
• 0:13–0:26 A chair with a split rung. A new rung is fitted and tapped home. The chair takes weight. Sign in frame.
• 0:26–0:39 A desk lamp with a frayed cord. The cord is replaced. The lamp comes on. Sign in frame.

Now enter both versions cold, at positions defined on the film's own clock.

Version A

Version B

The comparison turns on one difference. In both versions, an arrival just after a change sees a working object and cannot reconstruct what it was before. That loss is unavoidable and identical. What differs is how long it lasts. In Version B the next broken object is never more than a few seconds away: the 0:11 arrival meets the chair two seconds later, and the 0:37 arrival meets the kettle two seconds later, when the 39-second loop reaches its end and starts again at the dead switch. In Version A the same loss at 0:27 runs to the end of the film, because the only change in the piece has already happened by the time they look up. Version A is not badly made; it is well made for someone who was there from the start.

For this brief, the proposal is Version B. Each of its three segments has the same shape, and at every entry sampled above the arrival gets a service identity and, within about two seconds, either a repair in progress or the broken object that begins one; nothing in the piece requires the viewer to have seen the beginning. Those are the three entries examined here, not a survey of every moment in the loop. This is a proposal for the specified encounter, not a tested site design.

What could change the selection:

• A measured distance from sidewalk to screen, and whether anyone naturally stands still in front of it.
• Screen size, orientation, resolution, brightness, and whether the unit sits behind glass or in direct sun.
• Whether the workshop's own window sign competes with the screen for the same glance.
• Whether sound is unavailable permanently or only during certain hours.
• Observed behavior of the waiting customers — whether they look at the screen at all, and for how long. This needs watching, not assuming.
• Whether neighboring screens or other advertising share the display's rotation, and what share of the time this film holds.
• Venue content rules, permissions, and accessibility requirements.
• Whether any on-screen text would need translation or an alternative delivery.

Until those come back, this is a written proposition with a plausible structure and a short list of unknowns. Screen measurements, visibility checks, actual playback, and audience response all remain outstanding — and an orderly desktop layout showing three tidy frames is not evidence that any of them has been addressed.

Frequently asked questions

What should a treatment for a public display establish before describing the film?

The viewing situation: where the screen sits, how high it is, whether it is behind glass or in daylight, how far away someone could stand still, whether there is sound and any route for sound to reach anyone, and what nearby screens compete for the same glance. Sort what is known from what is not. If no dwell-time figure is supplied, write 'not supplied' and ask for the observation that would produce one; an invented average quietly shapes length, text, and beats.

How do you test a sequence for partial arrival?

Take the action and mark three or four arbitrary entry positions, not favorite moments. Read forward from each and write in plain sentences what a person who saw nothing before that moment would believe is happening. Sort the results into carried, restored, or lost. Repair options are limited: move the premise earlier, repeat the premise physically, make the visual action itself carry the meaning, accept the dependency and cut what depends on it, or decide something is a bonus. The failure is not having bonuses; it is not knowing which pile required information is in. Do not rely on a second pass to repair a first encounter the viewer may never complete.

When is a continuous progression better than repeated self-contained actions?

A progression builds and can hold a genuine reveal, and it rewards a longer stay; it is the better choice when people really wait, such as at a counter, in a queue, or in a waiting room, so the build has somewhere to land. On a public screen, however, a progression concentrates its payoff at one position in time, so people who see the end missed the middle and people who saw the middle never learn what the film was for. Repeated self-contained actions let a person arrive at any segment and get a whole idea, though they can feel like variations rather than a story, repeating identity can feel insistent, and near-identical segments risk becoming wallpaper. The site decides.

What presentation constraints should be carried from the encounter?

Describe the action at the scale it will be seen. What reads from several meters: a silhouette, a change of state, one moving object against a still field. What does not: two simultaneous actions, small type, subtle gradations, a face doing something quiet. If sound is unavailable, on-screen text is the only language channel, but it is also the least reliable element at distance and the hardest to catch mid-stride, so argue for fewer words held longer and check whether the proposition can be carried without words. Do not reach for universal legibility numbers; name the requirement and get it checked on site. Keep creative, technical acceptance, and accessibility judgments separate.

What did the worked example propose, and what remains outstanding?

For the invented repair-workshop screen with walkers and waiting customers and no sound, the proposal is Version B: three 13-second self-contained actions, a kettle, a chair, and a desk lamp, each showing a broken object, a visible change, and the service identity, running in order as a 39-second loop. Each sampled entry gets a service identity and, within about two seconds, either a repair in progress or the broken object that begins one, so nothing requires the viewer to have seen the beginning. That is a proposal for the specified encounter, not a tested site design. Screen measurements, visibility checks, actual playback, and audience response remain outstanding; an orderly desktop layout is not evidence they have been addressed. What could change the selection includes measured distance and standing behavior, screen size, orientation, resolution, brightness, glass or direct sun, a competing window sign, whether sound is unavailable permanently or only at certain hours, observed waiting-customer behavior, neighboring screens and rotation, venue rules, permissions, accessibility requirements, and translation or alternative delivery for on-screen text.

Further reading

• Pitch a Series Viewers Can Enter at Almost Any Episode — Useful as a cross-lane method for testing entry: it shows how to audit what a stranger needs from an ordinary middle episode.