Screenwriting and television

How to finish a screenplay

August 16, 2026 · The StoryProp team

Screenplays are rarely abandoned at the beginning. The beginning is the easy part — an image you have carried around for months, a first ten pages that arrived almost fully formed, a midpoint you can see so clearly you assume the rest will fill itself in. Scripts die later, in a particular place, for reasons particular to the form. Recognizing that place is most of what finishing takes.

The page-seventy swamp

Somewhere between page sixty-five and page eighty. That is where the stall lands in the drafts writers bring me, and the page numbers are not arbitrary. Austin Film Festival's 2026 competition rules put a feature at roughly 90 to 120 pages, and the median script in Stephen Follows' March 2026 analysis of 2,520 US Letter feature screenplays runs 119; on a script that length, pages sixty-five to eighty are the back half of act two. Writers describe the stretch in some version of the same terms: the story stops pushing, the new scenes feel like variations of scenes already written, the protagonist turns passive and starts having conversations about the plot instead of pressing on it. You open the file, reread the last scene, cannot think of the next one, and close the file. Then you do that a few more times and stop opening it.

It is the hardest stretch of a screenplay to write, for reasons that have nothing to do with your talent. By page seventy the setup has been spent. The premise questions have been asked. The fun of the concept — the sequence you imagined first, the moments that would go in a trailer — is mostly behind you. What remains has to be built rather than discovered: escalation that costs the protagonist something, complications that grow out of earlier choices, a steady closing of options so the ending feels inevitable rather than scheduled.

A screenplay is a hundred-odd pages of almost nothing but structure. There is little room to be interesting in the margins. A novel can survive a slack chapter, because a reader will follow a voice through slack terrain; a script that goes slack goes dead, because what keeps the pages turning is the machinery of want and obstacle. That makes a screenplay brutal to lose your way inside and, oddly, easier to finish once you know what the failure looks like.

The swamp is usually a want problem. If you cannot think of the next scene, it is generally because you cannot say in one sentence what your protagonist is trying to get right now and who or what is in the way. Stop writing and answer that on paper — not the want of the movie, the want of the scene. When a writer tells me the second act drags and I ask what the protagonist wants on page seventy-two, the length of the pause is the diagnosis.

Two other causes account for most of the rest. One is a midpoint that did not turn anything. A midpoint should change the terms of the story: new information, a reversal, a commitment that cannot be walked back. If yours was an event rather than a turn, the back half has nothing to run on, and adding scenes will not help — you are building the top of a staircase on a landing. The other is that you settled on a climax months ago and the script you actually wrote no longer leads there. Reread pages one through sixty as though a stranger wrote them and ask where this version is heading. Often a draft has found a better ending and is refusing to walk toward the old one.

Write the vomit draft, then admit it is one

The form has one real advantage here: the rough draft is legitimate and universally understood. Nobody expects a first pass to be shootable. You can write notes to yourself in the action lines. You can write a scene that is two characters saying the plot out loud, purely so you know what happens, and fix it later. You can write THEY ARGUE. HE LEAVES. FIX THIS. and move to the next slug line. Let the layout sort itself out too — pagination and elements are mechanical, and screenwriting software can carry them while you are still deciding what happens.

The trick is being honest about which draft you are in. Writers stall by trying to write a readable draft on the first pass — polishing dialogue in a scene whose function has not been decided, tuning an action line in a sequence that will be cut. That is not craft, it is avoidance wearing the clothes of craft. It feels productive because words keep changing. Nothing is being decided.

So make the passes explicit. The vomit pass exists to answer one question: does this story go all the way through? Its only obligation is to reach the end with a shape. The readable pass answers a different question: would a stranger keep turning the pages? Those two passes want opposite behavior from you, and running them at once is how a script stalls at seventy.

The readable pass has concrete work in it, which makes it less frightening than it sounds. Cut the first line and the last line of every scene, then count how many you do not miss. Find every place a character explains something the audience already understood. Find the scenes that exist only to move someone from one location to another, and either give them a want or delete them. Check that each scene ends somewhere other than where it began — a decision, a discovery, a shift in who holds the power.

The table read as a finishing tool

The one thing available to a screenwriter and unavailable to a novelist is that a script can be performed by a handful of people in a living room in an afternoon. Most writers treat the table read as a late event, something you do once the draft is good. Used that way it is a validation ritual. Used early, on a rough draft, it is the most efficient finishing tool the form has.

Cast whoever you can get. Actors are ideal; friends who read flatly are fine, and flat readings are diagnostic in their own right — a line that works only when performed brilliantly is a line that does not work. Read the whole thing in one sitting, aloud, with a clock running. Do not stop to explain. Do not apologize in advance for the parts you know are broken. Sit with a printed copy and make marks.

You are not listening for whether people like it. You are listening for where the room goes quiet in the wrong way, where a reader stumbles because a line is unsayable, where two characters sound identical in two different mouths, where a laugh lands that you did not plan and one dies that you were counting on. You are noticing which pages made you want to skip ahead — information you cannot get reading silently, because silently you supply the performance you imagine.

The swamp announces itself at a table read. Around page seventy the attention in the room visibly goes. That is not a verdict on you; it is a map coordinate. It tells you which dozen pages to rebuild, and rebuilding a dozen identified pages is a job you can actually finish. Vague dissatisfaction with an entire draft is not a job at all.

Define “done” before you need it

A draft with no definition of done will absorb unlimited work, and unlimited work is hard to tell apart from abandonment. Here is a definition that holds up: a screenplay is done when it is the draft you would hand to a stranger — someone with no affection for you, no context, no patience — and let them read it without sitting in the room to explain.

That standard does more than it appears to. It rules out the draft with placeholder scenes, because you would explain those. It rules out the thin third act, because you would warn them first. It also rules out endless polishing, because eventually the honest answer becomes yes, I would hand this over, and the right response to that answer is to stop. The stranger test is a quality bar and a stopping rule in the same sentence, which is exactly what a stalled writer is missing.

Scripts and novels fail differently at the finish, which is worth saying plainly. Novels tend to die of accumulated drift across hundreds of pages, a slow loss of the thread with its own remedies — why can't I finish my novel covers that ground. Scripts die of structural starvation inside a narrow, predictable band of pages. And if your problem is speed rather than direction — you know where it goes, you simply cannot get the pages out — that is a third thing, handled in how to finish a first draft fast.

What finishing teaches

The first script you finish will be worse than the first script you abandoned, in the sense that the abandoned one was perfect in your head. It is worth far more anyway, and not because you will sell it. It is worth more because you now have a body's knowledge of the shape: you know what page seventy feels like from inside, you know the feeling passes, you know roughly how many weeks the back half takes you specifically. That does not transfer from reading about structure. The next script goes faster, not because you got more talented, but because you stop reading difficulty as evidence of failure.

Writers who have finished several scripts are not people who stopped hitting the wall; they are people who stopped believing the wall means stop. So choose the ending you can defend today rather than the one you might invent in six months. Write the swamp badly on purpose. Reach the last page, read it aloud with a room full of people, rebuild the dozen pages where they reached for their phones, and hand it to a stranger.

Rebuilding a dozen pages in the middle of act two is exactly the work that goes wrong when a tool quietly rewrites more than you asked for. With StoryProp you select the sequence, say what it needs to do, and the surrounding draft stays as you left it — and every earlier pass is kept, so a rebuild that fails is not a loss.

← All posts