Write in a Director’s Voice Without Inventing Their Convictions
Write in a Director’s Voice Without Inventing Their Convictions
The paragraph that gets a treatment into trouble is rarely the boring one. It’s the one that sounds most like the director. It has her bluntness, her habit of ending a thought with “and we’re out,” the way she says “the thing is” before she says the thing. It also commits her to a memory she never had.
That’s the trap worth naming early, because it doesn’t feel like a trap. It feels like craft. But a director’s voice on the page isn’t an accent. It’s the pattern of what she noticed, chose, refused and could explain about this film — attached to a scene, a reference, a constraint. Cadence is a delivery system. It carries a position beautifully. It cannot supply one that was never there.
So the work splits into three parts, and they are easy to blur: carrying her judgments accurately, keeping track of which sentences are yours, and leaving the genuinely unresolved things visibly unresolved. The hard case — the interesting one — is the sentence you can already see is stronger than the thinking available to support it. There are honest ways to hold that sentence and one dishonest way. Most of what follows is about telling them apart.
Gather judgments rather than a bag of verbal habits
Before style, inventory. For one film you should be able to write down: what the director noticed when she read the board; which of two options she picked, and what she said against the other; what she ruled out and on what grounds; which references she reached for and how far she meant them to go; and where she said “I don’t know yet.” Those are the facts a voice is built from. If one is missing, no amount of diction fills the gap, because the gap is where the voice was supposed to be.
Notice that rejections count. “No voiceover” is a decision with consequences: it tells the editor, the composer and the client where this spot’s meaning comes from. A draft that quietly omits a rejection invites one back. If the treatment doesn’t say the voiceover was ruled out, the next reader has no reason not to suggest it.
Take a constructed case. It is invented, it stays invented, and it is not evidence about how any real collaboration has gone.
The notes. A 30-second spot for a sparkling water brand. One location, one product. From a call with the director:
- One location: the table by the window. She rejected the kitchen — “a second room makes it a story about a house.” Later in the same call she said something about maybe catching a kettle at the end. The two notes don’t agree, and you don’t know whether the rejection softened or she was thinking out loud.
- No voiceover. Ruled out.
- The pour comes first. Then the hold. Then the pack.
- The hold lasts “about two seconds.” She said it twice, which is how you know it’s a quantity and not a mood.
- Her reason for the hold: the product should arrive after the viewer wants it, not while they’re being told to want it.
- The camera during the hold: undecided. A slow push in, or locked off. She said she’d know when she saw it move.
- Sound during the hold: never came up. Nobody asked whether the fizz is audible.
That isn’t a voice yet. It’s the material a voice has to be built from, and everything below has to stay inside it. Note how much of it is decision rather than description — and note that two items are open, in two different ways. The camera question was left open on purpose. The sound question was never asked.
Separate articulation from contribution
Draft material falls into three kinds, and the labels matter more than they sound.
Her idea, made clearer. You moved her reason so it sits next to the choice it explains. You killed a hedge she’d want killed. It’s still hers. This needs no ceremony; it is the job.
Your addition. A transition, a reason, a beat, an organizing principle, an ending. Once you supply one, part of the material is yours. Pretending otherwise isn’t modesty, it’s a paperwork failure.
Nobody’s. A gap neither of you noticed, or one you both noticed and deferred. Not discussed is not the same as doesn’t matter, and it isn’t evidence that a decision was made earlier and forgotten.
Three short versions of the same ending paragraph make the difference visible. Each describes the pour, the hold and the pack, in two to four sentences.
A — articulation.
After the pour, the film holds on the glass for about two seconds before the pack enters. No voiceover. The pause is what makes the product feel earned rather than announced.
[OPEN — camera during the hold: push in or lock off. Not decided.]
B — invention.
After the pour, I hold. I know this pause from the inside; I’ve spent whole afternoons waiting for water to get cold enough to want it. The push in is slow because wanting is slow. That’s the film: two seconds of wanting, then the product.
C — proposal.
Proposal, for a decision. During the hold, the camera pushes in slowly over the two seconds, so the pack arrives at the end of the move rather than after it. This would replace the lock-off. What it costs: it tells the viewer the pause is going somewhere. If the hold is meant to be time they fill themselves, the lock-off is the stronger version. Your call.
It’s tempting to settle all this with a rule about who writes what — either the director authors and the writer transcribes, or the writer authors and the director approves. Real collaborations are messier and usually more interesting. Jordan Brady, in a 2025 Soapbox column published at LBBonline, describes writing his own creative approach. That’s one director describing his own practice in a first-person piece. It isn’t a survey, and it doesn’t predict how your project is arranged.
Neither extreme helps. Draft A rewrites the wording of her reasons, and if every rewording required sign-off, no treatment would ever be finished. The test isn’t how many words you touched. It’s whether the film — or the reason given for it — comes out different on the other side.
Test a strong sentence against its source
Two questions per sentence, in this order. Which note supports this? And does the sentence make the claim more certain than the note did? Most problems are one of the two.
Certainty creeps in through a short list. Invented experience and personal history. Confidence about execution — “the push lands on the beat” asserts something nobody knows yet. Claims about how the product performs. And the common one: a tentative preference promoted to a declaration. “I might want it quieter” becomes “she wants silence.” Those are not the same sentence, and only one of them is true.
Now take Draft B apart.
“I know this pause from the inside; I’ve spent whole afternoons waiting for water to get cold enough to want it.” That isn’t in the notes. She said nothing like it. It’s an invented memory in the first person of a person who hasn’t told you a memory — and it’s the best-written sentence in the paragraph, which is precisely the problem. The better it reads, the more likely it survives review, and by the time it’s approved it reads as her history.
“The push in is slow because wanting is slow.” Two faults stacked. It states a general conviction as though she’d offered one. And it resolves the camera question — the one she explicitly left open — inside a subordinate clause. That’s where most of this damage happens: not in a bold sentence, but in a clause that treats an open question as settled scenery.
“That’s the film: two seconds of wanting, then the product.” This keeps her quantity and recasts her reason. It’s close enough to what she actually said that it might survive a conversation. It still shouldn’t survive by default. If she reads it and confirms it, it becomes hers. Until then it’s an unlabeled proposal.
And a quieter fault: Draft B drops the voiceover rejection. Not because the writer disagreed, but because it didn’t fit the rhythm. Omissions are decisions too; they just don’t announce themselves.
The “I” isn’t the problem. Treatments are often written in a first person that’s closer to a register than a confession — “I want the light coming off the glass” is a normal way to state an intention. What has no author is a first-person memory. Keep the pronoun. Lose the history.
One more thing from the notes: the kitchen rejection and the kettle. Two lines from the same call that don’t sit together. The instinct is to keep whichever reads better. Don’t. Put both back to her: “You ruled out a second room; later you mentioned a kettle. Which is live?” A contradiction you resolve silently isn’t a resolution. It’s a substitution, and she may find out on set.
Present the better idea without laundering its authorship
Draft C is the same push Draft B smuggled in sideways, with two differences: it’s labeled, and it says what it costs.
Both matter. The failure mode isn’t having ideas. It’s proposing them in the indicative mood, as received fact. A proposal written as description stops being an offer the moment it leaves your hands; it becomes a claim about her intention, and the producer, editor or client downstream has no way to tell it came from you. The label isn’t a formality. It’s the part that makes the idea usable.
A good proposal carries three things: what would change, what it would replace, and the consequence. Draft C changes the character of the ending. It replaces the lock-off. The consequence is that the hold stops being static and becomes a move — the pack arrives at the end of something rather than after a stillness. Then it hands her a criterion she can answer: is the hold time the viewer fills, or time the camera spends? That’s the difference between an offer and a preference.
Should you send it, or wait? Three things you can usually check: how much latitude you’ve been given, how consequential the change is, and whether there’s time for a round trip. A paraphrase or a paragraph break doesn’t need a meeting. A changed ending, a changed reason, a new beat at the product moment, anything with budget attached — those go back.
When the discussion should continue but the draft is due, a clearly provisional version beats an invented certainty. Mark it where the reader will look. And when the decision lands, don’t delete the option that lost. Record it with the reason, so it isn’t re-proposed in three weeks by someone who wasn’t on the call.
Review the meaning before polishing the first person
Order of operations: does this say the film she intends, and only then, does it say it well.
The questions to put to her are blunt and few. Is this the film? Are these reasons you’d stand behind? Is there anything here you don’t recognize as yours? After that comes “this sentence has three clauses and I want two.”
And a rule about the fabricated memory, because it’s the one people try to fix the wrong way: approval doesn’t repair it. If she reads Draft B and says “sure, keep it,” she has endorsed an afternoon she never spent. The fix isn’t a smoother rewrite. It’s a question: “Do you want a personal opening here? If so, tell me one. I won’t write it for you.”
Credit, scope, who signs off, who may change what after approval — those are agreed per project, and getting a paragraph approved settles none of them. That’s fine. It just means you can’t infer the arrangement from the draft, and neither should anyone else.
Now suppose she comes back. She picks the lock-off, and gives a reason: a push “taps the viewer on the shoulder and tells them when to feel something.” She’d rather leave the hold for them to fill. She says nothing about the kettle, so that question is still open, and it goes back with the kitchen rejection. The pour–hold–pack order isn’t in doubt — it’s fixed, and the paragraph below can be written against it. What a live kettle would move is the ending: a kettle implies a second room, and a room after the pack means the film doesn’t end on the pack. If she comes back and says the kettle is live, the paragraph comes back with it. The sound gap stays open separately, as its own item: nobody has asked whether the fizz is audible.
After the pour, the camera is locked off. The film holds for about two seconds — the glass, the light off it, nothing else — before the pack enters. No voiceover. The hold is the viewer’s to fill; a push would tell them when to feel something.
[OPEN — sound: does the fizz carry into the hold? Not raised yet.]
Sentence by sentence, the sources are checkable. The two seconds, the pour–hold–pack order, the no-voiceover rule and the reason for the lock-off are hers. The arrangement — reason after shot description, open question visible at the bottom — is the writer’s. The bracket belongs to neither of you yet, which is what it’s for.
Notice what happened to the writer’s proposal. The push is dead, but its reasoning survived. The criterion offered in Draft C — is the hold time the viewer fills, or time the camera spends? — is now close to the reason she gives. That isn’t the writer winning. It’s the writer asking a question in a form that could be answered.
The temptation at the end is to close the bracket. An open question looks like unfinished work, and a treatment with a bracket in it can feel like an admission. But the bracket is the most honest line in the document. It’s the one place where a reader can see that nobody has decided, and that the person who wrote the rest knows the difference between what the director chose and what the page sounds like it should say. Close it early and you haven’t finished the treatment. You’ve made a decision that nobody made.
Frequently asked questions
What should an inventory of a director’s judgments include before drafting?
Write down what she noticed when she read the board; which of two options she picked and what she said against the other; what she ruled out and on what grounds; which references she reached for and how far she meant them to go; and where she said “I don’t know yet.” Rejections count too, because a draft that omits one can invite it back.
How can you tell an articulation of her idea from your own addition?
An articulation moves her reason next to the choice it explains or kills a hedge she would want killed; it is still hers. An addition is a transition, reason, beat, organizing principle, or ending you supply, and once you supply it part of the material is yours. A “nobody’s” gap is one neither of you noticed or both deferred; “not discussed” is not the same as “doesn’t matter.”
What makes a proposal usable rather than a smuggled decision?
A good proposal is labeled and says what would change, what it would replace, and the consequence. In the sparkling-water case, Draft C changes the ending’s character, replaces the lock-off, and hands back a criterion: is the hold time the viewer fills, or time the camera spends? An idea written as description or in the indicative mood becomes a claim about her intention, and downstream readers cannot tell it came from you.
What should you do with a contradiction or an open question in the notes?
Do not silently resolve it. The kitchen rejection and later kettle mention should go back to her: “You ruled out a second room; later you mentioned a kettle. Which is live?” A contradiction you resolve silently is a substitution. Keep open questions visible, because closing a bracket early does not finish the treatment; it makes a decision nobody made.
What if a director approves a fabricated first-person memory?
Approval does not repair it. If she endorses an afternoon she never spent, the fix is not a smoother rewrite but a question: “Do you want a personal opening here? If so, tell me one. I won’t write it for you.” The “I” is not the problem; a first-person memory with no author is.