Hold for the Reaction—or Cut Before It?
Hold for the Reaction—or Cut Before It?
There's a sentence that survives every treatment review and means almost nothing: We hold on her face as the implication lands. Everyone nods. Nobody in the room knows how long the hold is, what happens inside it, or what it earns. And the sentence is doing something worse than being vague. It has quietly merged two events that deserve separate decisions.
The first event is the viewer's recognition. The gag is set up, the revealing action happens, and the assembled information adds up. Recognition is the audience's. The second event is the character's reaction, which belongs to the character. A reveal usually hands the viewer recognition instantly — that is what a reveal is for. The reaction is a separate invention with a separate cost, and the only question worth asking about it is whether it changes the reading or just repeats it.
So here is the answer to how you describe comic timing in a treatment: name the information the beat needs, name the behavior that spends it, and name the image or sound that ends it. Do not name seconds. A duration with no event attached isn't a timing instruction; it's a mood, and it will be resolved by whoever sits at the edit.
Everything below is about how to decide which of those two events carries the joke — and how to write that decision so a director, an editor and a client are all arguing about the same thing.
Find the moment the gag is already over
Before you can time a reaction, you have to locate the revealing action and the instant the viewer can connect it to the setup. In a physical gag, that instant is usually the frame the cover clears the plate, the box opens, the person turns around. If the audience cannot assemble the relation, stop. Timing cannot create knowledge; it can only spend knowledge the action already delivered. A hold on a face will not fix a cause the viewer never saw, or geography that puts the reaction in a different room from the event. Those are repairs to the action, and they change the gag.
Be strict here, because a vague setup produces the worst possible timing note: hold a little longer so people get it. Nobody gets anything from duration. They get it from the arrangement of objects and the moment one of them is revealed. Once they have it, you have a decision to make and a budget of attention to spend.
What the hold has to buy
The assumption behind most reaction holds is that time itself is funny, or that a face is funny, or that a pause is funny. None of those is true on its own. A hold is a slot in which something can happen, and if nothing happens, the audience reads the empty slot as an instruction: this was supposed to be funny here. That is worse than a cut.
So give the hold a behavioral development. The person tries to recover dignity. They check who noticed. They commit to pretending nothing happened. Each of those is new information: not the cake is tiny, which the viewer already has, but the presenter knows, or the presenter has decided not to know, or the presenter's composure is more important to them than the cake. The viewer's understanding changes, which is the only definition of comic timing that survives contact with an actual edit.
The failure mode is a second confirmation. A face that goes wide, a slow blink, a small intake of breath — those can all be legitimate, but they usually restate what the reveal already said. Widening at the tiny cake confirms. Reducing a grand flourish to a two-finger version of itself informs. One of those is a repeat and the other is a rewrite of the moment.
Note, too, that restraint is not the same as content. A still face during a hold can be magnificent, but stillness is a performance choice, not a plan. If the hold is only stillness, the audience will supply whatever they feel about the person, which means the film has handed its tone to the room.
What the early cut has to hand the joke to
Cutting before the reaction is not the fast version of holding. It is a different route, and it needs its own mechanism. The joke moves out of the person and into whatever comes next: another image, a sound, a witness, a consequence. If you cut without specifying that next beat, you have not made a comic choice — you have declined one, and an editor will make it for you with whatever's on the timeline.
There are two useful things an early exit can do. It can leave the implication with the audience, who finish the reaction themselves and often enjoy it more for having done so. Or the incoming image can change what the reveal meant. Both depend on the cut point and the next beat being written as one thing, because the cut point means nothing except as the first frame of what follows.
There are also two ways it goes wrong. The first is erasure: the story needed the person to have registered something — a cost, a wound, a decision — and the cut removes the only evidence that they did. The second is casual dismissal: a sincere character gets abandoned at the exact moment their sincerity would have been the point. If the character is earnest and the film cuts away before anything can be made of it, the audience can read that as contempt, whether or not anyone intended it. That is a real risk when the person on screen is meant to represent the brand's customer.
One more consequence worth naming: an early cut also decides what the character knows. Cut before the face changes and the presenter may never learn. That is not a loss, and it is frequently the more interesting position — the audience holds knowledge the character doesn't, which produces a specific kind of pleasure and a specific kind of sympathy. But it has to be chosen, not stumbled into.
Writing a cut point without a stopwatch
This is where treatments go soft. Hold two beats. A beat too long. Snap cut. These are gestures toward a rhythm the writer can feel and the reader cannot.
Bind the timing to an event instead. Hold until the cover is back on the plate. Cut on the lift, before the arm comes down. Cut before the hand finishes the flourish. Each of these gives an editor something executable and a client something to disagree with, which is what you want in a treatment. The hold ends when a behavior completes, not when a timer expires. And each one implies its own duration without claiming to know what it is.
The moment you type a number — hold 1.5 seconds — you are asserting something about how long recognition takes for this audience. You have not measured that, and it varies by image, sound, culture, and what the viewer was looking at three shots ago. Numbers belong in the timing study, as placeholders to be resolved by building the thing. In a treatment they mostly signal confidence you don't have.
Write the sound in the same passage as the timing, because a hold with nothing continuing inside it isn't held by the pause. It's held by what's still running underneath. If the note stops at the reveal and nothing replaces it, the silence plus the face is one long silence plus one long face, and you have invented dead air. If something continues — a room tone, a fork, a small mechanical hum — the hold has an environment and the behavior inside it reads.
A constructed pair
What follows is a made-up example for method. Nothing here has been shot, cut, timed, or shown to anyone. Treat it as a specification, not a result.
The setup is a presenter at a table with a serving cover on it. The presenter performs an elaborate flourish over the cover: a full arm sweep, a wrist turn, a small dip of the head. The flourish is the lead-in, and the sound is a single sustained note that runs underneath it and stops the instant the cover clears the plate.
Then the revealing action: the cover comes off, and the cake on the plate is the size of a thumb, sitting on a full-size plate. The disproportion does the work. The viewer has the joke at that frame — the reveal is the recognition moment, and no reaction is needed to produce it.
Both routes now share everything up to this point: the same flourish, the same sound, the same lift, the same recognition frame. They differ only in where the film exits.
Route A — hold. The camera does not cut. The arm that did the flourish has nowhere to go. The presenter looks at the cake. Then looks past the lens toward the diners we can hear and don't see. Then the recovery: the arm comes down into a much smaller version of the same flourish — two fingers, a compact wrist turn — and the cover goes back on the plate with a deliberate click. The full-size sound does not return. What the hold buys is awareness, a check, and a recovery performed at the correct scale. The audience watches a person's dignity get rebuilt in front of them, and the joke stops being about a small cake and starts being about someone who cares what the room saw.
Route B — cut. We leave at the reveal, before the presenter's expression forms. The next image is a guest's plate holding an equally tiny cake, already half eaten, no ceremony. The cut delivers a reinterpretation rather than a reaction: the scale is not this presenter's private catastrophe, it is the room. And it denies the presenter recognition entirely — the flourish stays sincere, the presenter never learns, and the audience carries the knowledge for them. What it costs is everything Route A bought: the awareness, the check, the recovery. If the film needed the audience to like this person for surviving the moment, that affection now has to come from somewhere else in the spot.
The two routes are not slow and fast, and they are not good and bad. Route B's next image is part of its mechanism, not a neutral control, and any honest comparison has to say so: you cannot hold the following shot constant across both versions, because the following shot is the second route.
Which route the commercial wants
The choice is close to the ad's whole argument about people, so make it that way.
The hold keeps the audience inside the presenter's composure failing and reassembling. It is affectionate and a little cruel in the way that watching someone recover always is. If the presenter is the product's user or the brand's face, you are asking the viewer to watch that person get it wrong and then handle it, which is a warm position and also an exposed one.
The early cut protects the presenter's composure by never inspecting it. The joke moves to the surroundings, the audience is in on something the presenter isn't, and the person stays earnest. That is a cooler and often more durable relationship, and it survives a client who is nervous about the spokesperson looking foolish — but it can also read as the film losing interest in its own character.
Write down what each version loses. A recommendation that only lists gains is not a recommendation; it is a preference wearing a suit.
When the client asks which one is funnier
Answer with an editorial judgment and label it as one. The reasons are legitimate: one version gives the audience a person to care about and the other gives them a system that is absurd; one requires the recovery to play out to its end and the other exits at the reveal and hands the rest to the next image. Those are reasons, and they can be argued.
What you cannot do is convert a preference into a finding. A timed study tells you whether the passage reads, whether the hold is covered by sound, whether the cut point lands where you thought it did, and whether the incoming image is legible in the time it gets. It does not tell you what people find funny. Neither does a preferred edit, and neither does a room's reaction at a pitch — a laugh in a conference room is a fact about the conference room, stated as such. If a reaction is ever reported, report the conditions alongside it: who was there, what they saw, whether sound was playing. Otherwise you have laundered an anecdote into evidence.
Build the pair when you build it: same flourish, same sustained note, same lift, same recognition frame, differing only at the exit and downstream of it. Watch with sound, then without, and note which version depended on the audio to survive the hold. If you find that a hold needs a specific duration to work, say that you found it in this passage under these conditions — not that you found the length of a comic beat.
The exit sentence
Finish the treatment with the exit point and one sentence about what it keeps or leaves out. For the hold: We leave the cover back on the plate, the flourish reduced to fit it; the room has seen the recovery. For the cut: We leave the guest's plate, the same tiny cake eaten without comment; the presenter never finds out.
Both sentences are honest, both leave the trade-off standing, and either one can be defended in the room. The only thing the two routes are actually arguing about is whether the audience gets to know something the presenter doesn't. Decide that, and the timing describes itself.
Frequently asked questions
What two events are merged by 'hold on her face as the implication lands'?
The viewer's recognition and the character's reaction. A reveal usually hands the viewer recognition at the revealing frame; the reaction is a separate invention with its own cost. The useful question is whether the reaction changes the reading or just repeats it.
What does a reaction hold have to buy?
A hold is a slot in which something can happen. It should give a behavioral development that changes the viewer's understanding, such as the presenter checking who noticed, trying to recover dignity or committing to pretend nothing happened. A wide-eyed second confirmation often just restates the reveal. Stillness alone is a performance choice, not a plan, and if only stillness is there the audience supplies the tone.
What must an early cut specify?
It must specify the next beat: another image, a sound, a witness or a consequence, written with the cut point as one thing. The cut can leave the implication with the audience or let the incoming image reinterpret the reveal. Without a next beat, the film has declined a comic choice and an editor will make it. Risks include erasure of the only evidence that the character registered something, and casual dismissal of a sincere character.
How do you write timing without naming seconds?
Bind the timing to an event: hold until the cover is back on the plate, cut on the lift before the arm comes down, cut before the hand finishes the flourish. The hold ends when a behavior completes, not when a timer expires. Typing a number asserts how long recognition takes for this audience, which has not been measured and varies; numbers belong in a timing study, not as confident treatment copy.
Can a timed study or a room's laugh settle which version is funnier?
No. A timed study can tell you whether the passage reads, whether the hold is covered by sound, whether the cut point lands where you thought and whether the incoming image is legible in its time. It does not tell you what people find funny. A laugh in a conference room is a fact about the conference room; if a reaction is reported, report who was there, what they saw and whether sound was playing.