Leave Room for Improvisation Without Losing the Commercial’s Required Beats
Leave Room for Improvisation Without Losing the Commercial’s Required Beats
“The dialogue will be largely improvised” is one of the least useful sentences you can put in a treatment, because it does two entirely different jobs depending on what sits around it.
In one version, it means: the words aren’t written, the destination is. The two people in the kitchen have to end up at the same place every take, and how they get there is theirs.
In the other, it means: we’re hoping something good happens.
Only the first one is a plan, and only the first one gives a client something to approve. The whole craft question is how to write the boundary so clearly that everyone can see which sentence they’re reading.
What “fixed” actually means here
Before you can describe freedom, you have to know what it’s freedom from. On a commercial, the genuinely fixed material tends to fall into four groups, and they aren’t fixed the same way.
Product action. The physical thing the product has to do on camera, in a form that reads. Not “she uses the machine” but “the lever is pressed once and the brew starts.” This is the most concrete and usually the least negotiable.
Wording. Exact copy, mandatory claims, a tagline, a closing line. The important question is whether it’s fixed verbatim or fixed in sense. “One press. That’s the whole morning” delivered word-for-word is a different constraint from “something in this spirit.” Both are legitimate, and they produce completely different amounts of room.
Endpoint. Where the film has to arrive, and what has to be true on the last frame before the product shot. This is the one people forget, because it feels like a storytelling matter rather than a requirement. It isn’t. If the machine has to be running when the beauty shot begins, that’s an obligation, and improvisation has to end somewhere upstream of it.
Story information. Audience has to understand these two people are housemates, that it’s morning, that this is the first time the thing has come out of the box. Nothing about that is dialogue, and all of it can be handled by behavior rather than script.
Two questions cut through the list faster than anything else.
First: who can actually authorize a change to each item? If legal signed the claim, the claim is not the director’s to loosen in the rehearsal room. If the client approved the closing line as written, then “we’ll explore it” quietly relocates authority into a place the client can’t see. A treatment that expands its own permission is a treatment that will be renegotiated on set, which is the worst possible moment for that conversation.
Second: could this requirement be met again on take four? Improvisation means variation, and a fixed beat has to survive variation. If the required moment can only happen once — the accidental overlap, the perfectly timed interruption, the one laugh that made the line land — then it isn’t a requirement. It’s a wish, and the client should be told which one they’re buying. Reproducibility is the honest test of fixedness, and it’s the test that makes everything downstream plannable.
There’s a compliance dimension here that’s easy to miss. Every unscripted line is an unreviewed line. The moment dialogue is unwritten, the scene acquires a surface where someone can mention a price, a competitor, a health outcome, or a performance claim that nobody approved. This isn’t a reason to avoid improvisation. It’s a reason the treatment should name a short, specific exclusion rather than saying “we’ll keep it on brand,” which is advice rather than a boundary.
Where the freedom can actually live
Variation on a commercial set comes in roughly three kinds, and they cost different things.
Phrasing. The same intention, said a different way. Cheapest to allow and easiest to control, but only where the wording is genuinely open.
Approach. How someone goes at the task: bickering about it, doing it while talking about something else, doing it absent-mindedly. This is usually where the useful discovery lives, and it doesn’t require a single unapproved word.
Response. What one person does because of what the other one just did. This is the thing you can’t script and the main reason to leave room at all.
Notice that all three are about how the scene is played, and none of them is a change to what the scene is for. Unscripted dialogue and unplanned scene purpose are different conditions, and treatments collapse them constantly. A scene can have entirely unwritten words and a purpose that is fully decided. It can also have a script and no idea what it’s for, which is not an improvement.
The practical failure mode is telling performers to improvise “around” constraints that were never written down. Nobody can play inside a boundary they haven’t been shown. But there’s a limit on the other side too: the treatment is not the place to write performance notes. The document states which region is open and what has to be true at each edge. How someone approaches it is a conversation, and it belongs in the room.
A contained passage of freedom
Here is a worked comparison. It’s fictional — an invented spot with the ordinary shape of a thirty-second product commercial — and both routes are proposed plans, not descriptions of anything that has been rehearsed or shot.
Two housemates, Ines and Dev, in a kitchen, early morning. The product is a small brewer. What the brief genuinely requires: the box is opened on camera and the brewer is lifted out and set on the counter; the lever is pressed once and the brew starts; and the spot ends on an approved closing line, delivered verbatim as the last spoken thing. Thirty seconds total, which gives the scene maybe twenty.
Route A — improvisation confined to the unpacking.
The free passage sits between two physical landmarks. Entry: both pairs of hands on the closed flaps, box on the counter. Exit: brewer standing on the counter, right side up, label legible, both hands clear. Everything in between is unscripted, with a short prohibition list — no price, no competitor, no product claims, nobody comments on what the machine does, hands stay on the box and the brewer.
The actors are trying to get the thing out of its packaging without looking like they’re performing a task. The director is trying to get an unforced moment between two people who plausibly live together. The press and the closing line sit outside the free zone, untouched.
What this route buys is structural reliability. The exit state is physical, which is better than a line as a return point, because it can’t be interrupted mid-sentence. Any take that reaches the counter is usable for the required action.
What it costs is visible in the same breath. The exchange may be building toward something at exactly the moment the machine has to land, and then the take either loses the discovery or loses the beat. A mechanical return makes a scene feel segmented, and the audience reads segmentation as performance. There’s a second, subtler cost, and it grows with every item added: a prohibition list long enough stops being freedom. It’s a script with the lines removed, and performers can feel the difference.
The open version, and what it actually costs
Route B — the press and the closing line are fixed in place; the unpacking is required but unplaced.
The unpacking still has to happen. It can occur anywhere in the exchange, or late, or off to the side while the two are talking about something else. The exchange is allowed to run. Nothing marks the entry.
The gain is real and specific: the press stops being a step and becomes a response. If the two are arguing about whose turn it is to make coffee, the press can land as the answer to the argument. That’s the kind of behavior you cannot write, and it’s the honest argument for the open route.
The uncertainties are equally specific, and a treatment that doesn’t name them is asking the client to approve a feeling rather than a plan.
Duration. An unplanned exchange can run forty seconds or four. The thirty doesn’t stretch, and an open scene that overruns has to lose something, usually the thing that was working.
Continuity. At the top of every take, the box has to still read as closed, the brewer has to end up right side up, and the label has to be legible at some point in the film. Those are cheap constraints individually and easy to break collectively when nobody is holding them.
Endpoint. The closing line has to be the last spoken thing, delivered by an actor who is in a state to land it. If it arrives on the tail of an exchange that has wandered into a different register — loose and funny, then suddenly sincere copy — it doesn’t land, and no cut fixes that.
And one distinction worth being stubborn about: unwritten lines do not make a moment spontaneous. The circumstance and the performers still decide what can happen. A treatment that promises “authentic, unscripted reactions” is promising a result it has no way to schedule. Describe what’s open. Don’t describe what it will feel like. The second claim can only be settled after the fact, which means it can’t be approved before the shoot.
The return point, and what coverage can and can’t do
Both routes arrive at the same place, and that’s the useful part. Whatever happens in the middle, the film needs a take in which the box is open, the brewer is on the counter with the label visible, the lever has been pressed once, the machine is running, and the closing line has been delivered as the last line. The route changes how likely that is per take. It doesn’t change what has to be true.
That’s the sentence to put in front of a client. Not “we’ll shoot a lot and find it,” but “whichever route we take, this is the state every usable take has to reach, and here’s what we’ve planned to make sure it’s available.”
Coverage deserves the same precision. Coverage produces optional versions of the required beat — a better angle on the press, a better face while it happens, a better reaction after. It cannot produce a press that never occurred, a line nobody said, or a reveal that got skipped. Coverage funds variation. It does not fund absence. The distinction matters because “we’ll find it in the edit” is a claim about the edit, not about the shoot, and the two are not interchangeable. If the client wants the press and the closing line continuous in one image, then no amount of cutting assembles a take that didn’t exist. If they can be separated, that’s a decision the client should make while reading the treatment, not one an editor should be forced to make in a suite.
The same discipline applies to how much you ask of the people in front of the camera. A contained passage and an open scene are different amounts of work, and any expansion of someone’s scope is a conversation with them, not a line item in a document. The treatment can state a scope. It shouldn’t quietly enlarge one.
What the treatment should actually say
A workable version names five things, in plain language:
The anchors — the action, the wording that’s fixed (verbatim or in sense), the endpoint, and the story information that has to read. The open zone — specifically which passage is free and what kind of variation is welcome there. The exclusions — short and concrete. The return — what every usable take has to arrive at, and how coverage is planned around it. And the condition under which you’d choose the open route over the contained one, plus what would tell you the scene can’t sustain it.
That last item is where the document has to be honest about its own limits. What actually tells you whether an open exchange will hold is a room, two performers, and twenty minutes. A treatment can name the test. It can’t run it, and it shouldn’t imply that it has.
Both routes are plans on paper, which is what they are.
If the improvisation would change the approved proposition, the product claim, or the closing line, label that as what it is: a separate creative decision requiring the authority of whoever approved the original. The document should not move that authority into the rehearsal room by implication.
Which is the right place to end up: an explicit open zone, anchors that are named and held, and a route back to the required action that doesn’t depend on anything good happening. Write that, and the freedom is real. So is the commercial.
Frequently asked questions
What makes 'the dialogue will be largely improvised' a plan rather than a hope?
If it means the words are not written but the destination is, and the fixed anchors are named, it is a plan. If it means hoping something good happens, it is not something a client can approve. The treatment must make clear which sentence is being read.
Which commercial elements are genuinely fixed, and how do they differ?
Product action, wording, endpoint and story information. Product action is the physical thing that must read on camera. Wording may be fixed verbatim or only in sense. Endpoint is what must be true on the last frame before the product shot. Story information, such as the pair being housemates or the machine being new, can be carried by behavior rather than dialogue.
Why does reproducibility matter when deciding whether a beat is required?
Improvisation means variation, so a fixed beat has to survive variation. If the required moment can only happen once, such as an accidental overlap, a perfectly timed interruption or a one-off laugh, it is a wish, not a requirement. The client should know which one they are buying.
What can coverage not do?
Coverage gives optional versions of a required beat that occurred: a better angle on the press, a better face, a better reaction after. It cannot produce a press that never occurred, a line nobody said, or a reveal that was skipped. If the client wants the press and closing line continuous in one image, no cutting assembles a take that did not exist; if they can be separated, that choice belongs in the treatment, not the edit suite.
What are the trade-offs between confining improvisation to the unpacking and leaving it unplaced?
The confined route buys structural reliability because its exit is physical, with the brewer on the counter, right side up, label legible and hands clear, so it cannot be interrupted mid-sentence. It can cost a discovery if the exchange builds at the moment the machine must land, and a long prohibition list stops feeling free. The open route lets the press become a response, such as an answer to an argument, but brings uncertainties about duration, continuity and whether the closing line can land after an exchange in a different register. Unwritten lines do not by themselves make a moment spontaneous.