Run a Director–Writer Session Around the Scene You Haven't Solved
Run a Director–Writer Session Around the Scene You Haven’t Solved
You have the brief. You have the director's voice written down somewhere. And one scene still doesn't resolve — not because the sentences are bad, but because two things want to happen in the same eight seconds and nobody has said which one wins.
That is the session. Not a workshop, not a tone-alignment meeting, not another pass at making the treatment prose read better. You take one passage, put two concrete versions of the action beside it, work out what each version makes a viewer understand, and leave with a decision or with a question that belongs to someone specific. Either of those is a good session. A more elegant description of the same uncertainty is not.
What follows uses an invented scene. It isn't a transcript, it doesn't belong to any real director, and it isn't evidence that this method works in the room — it's there to show the operation.
Bring the point where the scene stops making sense
Choose the smallest consequential passage, not the section. One beat. Who notices the object. What motivates the cut. When the joke becomes legible. The order of two actions that both currently happen first.
Arrive with three things written down: the scene's purpose in one line, the requirements nobody is reopening, and the exact spot where you stop being able to predict what the viewer understands.
The requirements matter more than they sound. In the example below they include: no dialogue anywhere in the film, the client's name appears once in the final five seconds, and there are two cuts, sixty seconds and thirty. Those aren't up for discussion. Naming them first means the session argues about the one thing that is open, instead of relitigating the brief from the top.
Then name the disagreement as an action problem, not a quality problem. "This section feels weak" gives the director nothing to respond to and invites a general conversation about taste. "The key is on screen for four seconds and I can't tell whether the audience has been told what it is yet" is a question with two answers.
Three different conditions get confused constantly, and they need different handling:
- An unclear sentence. A writing problem. Fix it before you get there; don't spend the director's attention on your grammar.
- An undecided action. A session item. Nobody has chosen yet, and choosing is the point of the meeting.
- An execution question. Needs a specialist or a small test — a prop, an editor, a UI designer, a location scout. Discussion won't settle it.
The most common way a session burns an hour is that both people agree the scene should feel intimate, and disagree about who moves first. Tone agreement is not action agreement. It can hide the gap for weeks.
Put two filmable descriptions on the table
Write each version as what the viewer sees and hears. Ordinary actions, in order. Not "warmer," not "more human," not "she should feel conflicted." If you can't say where the camera is and what happens in what sequence, the version isn't ready to compare.
Both versions start from the same conditions. This is where writers lose arguments they should have won: a new idea gets a fully imagined treatment while the existing idea gets three thin lines, and the new one wins on word count. If you're rewriting the option already on the table, write it as well as you can. Stacking the deck wins the meeting and costs the film.
Keep whatever the director has already committed to inside both versions. If you're introducing something new, mark it as a proposal — "proposal: move the reveal to the end" — rather than attributing it to the director. "The director wants the audience to be surprised" is a conviction you invented on someone's behalf.
Here's the invented scene.
Fixed: Nadia, mid-thirties, stands in a narrow hallway. Oz, who lives there, holds out a plain key on a plain ring. She hesitates before taking it. No dialogue. The client's name appears once, in the last five seconds. There are two cuts.
Version A — the viewer knows first. Three seconds of Nadia's thumb on a phone: a booking confirmation, a street address, the host's first name, a time. No client mark on the screen, and no client name — that one stays held for the end. Cut to the hallway. Oz extends the key. She hesitates, takes it, unlocks the door. The interior is plainly someone's home — a shelf of paperbacks, a coat on a hook. In the last five seconds the phone comes out again and the client's name appears on the screen.
Version B — the viewer learns with her. No phone. Cold open on the key in Oz's open hand, then her face, then the handoff. She hesitates. She takes it. She unlocks the door, and the interior is plainly someone's home — a shelf of paperbacks, a coat on a hook. Only at the end does the phone appear, with the client's name.
Both versions run the same actions in the same order. The only thing that moves is where the audience is standing when they happen. No adjectives were needed to write either one.
Compare consequences instead of trading adjectives
Follow each version through three channels and see where they separate.
Attention. In A, the viewer spends the first seconds reading information and arrives at the hesitation already oriented. In B, the viewer spends those seconds not knowing what the object is, and finds out by watching her decide. What arrives too early in A is the premise.
Performance. A asks the actor to evaluate a person — is this host trustworthy. B asks her to evaluate a situation without dialogue, which is a harder and more exposed piece of acting. That isn't an argument against B; it's a fact about what B costs.
Product role. In A the product is the premise. The viewer knows what kind of transaction this is before anything happens, so the film is about whether one particular transaction goes well. In B the product is the payoff. The viewer experiences the ordinary strangeness of taking a stranger's key, and then learns that this strangeness now has a system behind it.
Then ask what each version quietly asks production to establish. A needs a phone screen legible at small sizes, which means a designed asset and an approval path; the interior only has to be plausible. B has no screen until the last beat, so the interior carries the premise on its own — one unglamorous shot that reads as inhabited, doing the work that in A is done by the screen.
Watch the moment the conversation turns into preference. "I like B better" is a creative vote. "B depends on an interior that reads as lived-in in one shot" is an execution question with a specialist attached. Both are legitimate, but they get settled differently, and neither gets settled by a show of hands. A director's preference is a decision. Feasibility is not.
One more thing to check: which relationship disappears. In A, Nadia and Oz are two people completing an arrangement. In B, they're two strangers negotiating one, and the interior is a third presence. If the brief cares about the host, A spends its first three seconds somewhere else entirely.
Make a decision that the writer can actually use
Record four things, not one: the chosen action, the reason it serves this film, the rejected alternative, and who holds the decision.
The reason is what protects the choice later. Without it, a note in three weeks — "should we see the booking up front?" — reopens a settled question at full cost, because nobody remembers why it closed. With it, the answer is one line and the conversation is over.
In the invented scene, the director chooses B and gives this reason: the film's claim is that a room in a stranger's home is an ordinary thing to do, and the audience's own hesitation is the story. Showing the booking first replaces the audience's hesitation with Nadia's, and then the film has nothing to do but confirm that her host is fine.
The rejected alternative goes in the note too, with its reason attached: A, because it spends the audience's uncertainty before the film can use it.
Note what the writer did and didn't do here. Bringing B, or bringing A, or inventing the phone-screen opening in the first place — all of that is contribution, and it's welcome. Staging the hallway is the director's call. Contribution and authority are separate, and it helps everyone if the note says which is which rather than leaving it to be inferred later.
If the decision turns out to belong to someone else — a client rule, a producer's constraint — write down whose it is. Inventing agreement is worse than leaving a gap, because the gap gets discovered during production.
Leave a useful non-decision when evidence is missing
Sometimes neither version can be chosen yet. That's not a failed session, provided you leave with the unanswered question named, its owner named, and a description of what would resolve it.
Don't manufacture a decision to end the meeting on a good note. And don't send the writer away with "make it work" — that's not an instruction, it's an absence of one.
In the invented case, the decision is B for the sixty-second cut, and the open question is whether the withheld information survives the thirty. The hesitation has to read as anticipation rather than confusion, and that depends on a performance that doesn't exist yet. So:
Decision: Version B. The hallway and the interior can be drafted now. The opening three seconds stay marked provisional.
Open question: In the thirty-second cut, does the pause read as anticipation or as confusion? Owner: the director, once a cut exists. What would resolve it: assembling the short version and watching it with someone who hasn't read the treatment.
If the short cut fails, a second question becomes live — whether a fragment of the booking screen may appear before the final beat without the client's name attached, which is a client call and not a writer's. That question stays dormant until the short cut gives it a reason.
Keep further exploration inside the scope you agreed. An unresolved beat is a reason to work on that beat, not a licence to re-outline the film.
You end with a scene that can be drafted except for three seconds of it, and a single question with a name attached to it. That's a real result. The version where every disagreement evaporates by the end of the hour is the one to be suspicious of.
Frequently asked questions
What is a director-writer session actually for?
It is for one unresolved scene, not a workshop, tone-alignment meeting, or another polish of treatment prose. Take one passage, put two concrete versions of the action beside it, and work out what each version makes a viewer understand. A good result is either a decision or a question that belongs to someone specific; a more elegant description of the same uncertainty is not a result.
What should I bring to the session?
Bring the smallest consequential passage, one beat, and three things written down: the scene's purpose in one line, the requirements nobody is reopening, and the exact spot where you stop being able to predict what the viewer understands. In the example, the requirements include no dialogue, the client's name once in the final five seconds, and two cuts, sixty seconds and thirty. Naming them first keeps the session on what is open.
How do I tell a writing problem from a session item or an execution question?
An unclear sentence is a writing problem: fix it before the session. An undecided action is a session item: nobody has chosen yet, and choosing is the point of the meeting. An execution question needs a specialist or a small test, such as a prop, editor, UI designer, or location scout. Tone agreement is not action agreement; it can hide the gap for weeks.
What is the difference between the two example versions?
In Version A, the viewer sees the booking confirmation first and arrives at the hesitation already oriented; the product is the premise, and it needs a phone screen legible at small sizes, with a designed asset and approval path. In Version B, there is no phone until the end; the viewer learns with Nadia, and the product is the payoff. B asks the actor to evaluate a situation without dialogue and makes the interior carry the premise in one shot. Both versions run the same actions in the same order; only where the audience stands changes.
What should a usable decision note contain?
Record the chosen action, the reason it serves this film, the rejected alternative, and who holds the decision. In the invented case, the director chooses B because the film's claim is that a room in a stranger's home is ordinary and the audience's own hesitation is the story; A is rejected because it spends the audience's uncertainty before the film can use it. If the decision belongs to someone else, write down whose it is. If evidence is missing, leave a named question, its owner, and what would resolve it rather than manufacturing a decision.