How to write a TV pilot that can carry a season
· The StoryProp team
A feature script has one job: tell a story that ends. A pilot has a stranger job: tell a story that ends tonight while proving the show can keep going without you propping it up. Most advice on how to write a TV pilot treats it as a short film with act breaks. It is not. A pilot is a working sample of a machine, and the reader's real question is never whether this hour is good — it is whether they can imagine hour nine.
Take our recurring example, The Quiet House. As a feature it works: Mara comes back to the small town she left at nineteen to clear out her late mother's house, spends a week among boxes and casseroles, discovers what Daniel has been hiding, and drives away changed, a chipped cup on the passenger seat. Satisfying — and completely closed. Filmed as written, it would make a fine movie and an unwatchable series, because the premise burns itself as fuel. The house gets cleared. The secret gets found. Nothing is left to want next week.
This piece is about that difference: engines, promises, and the bookkeeping that keeps a season honest. It assumes you already have the grammar of the page. If you are still learning how to write a TV show script at the level of sluglines, action lines, and dialogue blocks, start with writing a screenplay when you've never written one — the format is the same grammar, and this article won't reteach it.
Premise engines: where episode nine comes from
A premise engine is a situation that produces stories on its own, without the writer forcing it. Engines have recognizable shapes. A job with intake: cases arrive, patients arrive, clients arrive, each one a guest story with a beginning and an end. A place people pass through: a bar, a hospital, a border town. A lie that must be maintained in public, generating fresh danger every time someone gets close. A long goal that breaks into stages, each stage an episode's worth of trying. What these share is renewal: the situation resets enough to keep generating, but the characters never reset at all.
Here is a blunt test. Set a timer for ten minutes and write one-line stories for episodes four, nine, and thirteen. Not good ones — plausible ones. If they come easily, you have an engine. If you find yourself inventing new premises to make each episode possible — a cousin arrives, a storm hits, someone inherits a hotel — you do not have a show yet; you have a pilot-shaped film. Better to learn this before you write sixty pages, because no amount of polish on episode one fixes a machine that only turns once.
The Quiet House becomes a series the moment Mara stays. Say she takes over the estate-clearing business her mother ran: every week a house, a family, a life told in objects — and every house is a closed story a new viewer can walk into. Underneath runs the serialized spine: the chipped cup matches a set Mara has never seen, Daniel knows more about their mother's clients than he admits, and the town's houses keep yielding things that should not be in them. The engine is the job. The reason to keep watching is the spine.
The cold open teaches the audience how to watch
The cold open's job is not to shock; it is to calibrate. In its opening minutes, viewers learn what kind of show this is: how fast it moves, what it finds funny, what counts as danger here, whether death is possible, whether strangeness will be explained. Every choice in the open is a promise about every episode that follows, which is why a misleading cold open is worse than a dull one. A dull open loses viewers; a misleading one loses them angry, three episodes later.
Two candidate opens for Mara's show. One: a body in a crawl space, police tape, a scream. Two: an estate sale, a stranger haggling over a dead man's coat, Mara watching, and one box she quietly refuses to sell. The first promises a corpse a week — a promise this show cannot keep. The second promises exactly what the series delivers: objects as evidence of lives, and Mara as the person who reads them. Same world, same character; only the second teaches the audience how to watch.
Episode one makes the season's promises
A pilot runs three clocks at once. The episode question is opened and answered tonight — this house, this family, this small mystery resolved by the credits. The season question is opened tonight and answered in the finale — what was their mother actually doing for the people whose houses she cleared? The series question is opened tonight and never fully answered — can Mara live in a town that watched her leave? Weak pilots run only the first clock. Overloaded pilots try to answer the second. Episode one's job is to start all three and settle only one.
Every promise is a debt. Plant the cup in the cold open and you owe the audience its meaning — not necessarily in episode two, but visibly on the books. The practical discipline is a ledger: what did I plant, in which scene, and what payoff does it owe? Write the entry when you plant the thread, because six weeks from now you will not remember whether Daniel's hesitation at the kitchen door was a setup or just a mood. Threads you track are suspense; threads you forget are continuity errors wearing suspense's clothes.
Serialized or episodic: decide what carries over
The real decision is what a viewer loses by missing a week. In a fully episodic show, almost nothing: the situation resets and the guest story stands alone. In a fully serialized show, everything: each episode is a chapter, and joining late means starting over. Most working television lives between the poles — a closed guest story each week riding a serialized spine — and the pilot must demonstrate the actual ratio, because the pilot is the only evidence anyone has.
Getting the ratio wrong in episode one misleads everyone downstream. A pilot that spends most of its pages on the mythology of the cup promises a heavily serialized show, and viewers who came for a weekly house-and-family story will drift when the spine takes over — or the reverse. For Mara's show, the honest ratio is visible in page count: most of the hour on the house of the week, a scene or two on the spine, the spine touching the weekly story wherever it can. Decide the ratio before you draft, and let the pilot be a fair sample of it.
The television series bible is living memory
Two documents share the name. The pitch bible is a brochure — tone, world, character sketches, season arcs — written once to sell the show. The working bible, the one that determines whether season two contradicts season one, is a ledger that never stops changing: canon facts as of the latest draft; a timeline; every planted thread and the payoff it is owed; and, the entry television uniquely needs, who knows what as of which episode. Mara learning Daniel's secret in episode four changes what every later scene between them can be. If the bible does not record when she learned it, someone will eventually write her ignorant of it.
Television strains memory harder than any other form. A season is written over months, sometimes by many hands, sometimes out of order; the person drafting episode eight is functionally a stranger to the person who wrote the pilot, even when they share a body. A continuity failure in a novel embarrasses you once, in the reprint. In television it compounds: an episode that contradicts the pilot becomes canon too, and now the ledger has to hold both facts and the excuse. The habit that prevents this is unglamorous — update the bible the day the decision is made, not the week you need it.
This is also the honest way to judge an AI story structure tool for screenwriters: not whether it can generate scenes, but whether it remembers what you decided. It is what StoryProp does for television — decisions you confirm get promoted to records of canon, characters, and story shape, and the agent reads those records before it drafts, so the thing you settled about Daniel while working on episode two is still binding when you are deep in episode eight. Pages stay format-native, with professional layout and FDX and Fountain interchange, so keeping the bible never comes at the cost of a readable script.
The last scene sells episode two
A feature ends on resolution; a pilot ends on ignition. The last scene's job is to make episode two feel less like an option than an inevitability — not necessarily with a cliffhanger, but with the engine visibly turning over. The best pilot endings are quiet confirmations of the premise: the case closes and the next one is already in the doorway; the lie survives the week and the person most likely to expose it moves in next door. Resolution for the episode, ignition for the series, in the same image if you can manage it.
For Mara: the house of the week is cleared, the family drives off, and in the empty kitchen she unpacks her own box to find a second cup, chipped in the same place as the first — from a different house. Cut to black. A reader who finishes that page can pitch episode two back to you without prompting, and that is the test that matters. A pilot has done its job not when a reader says the hour was good, but when they start telling you, unasked, what happens next week.
A pilot asks you to keep a season's worth of promises in your head before the season exists — who knows what, which threads are owed, what the cup means. StoryProp keeps that ledger with you: what you agree to gets written down, the agent reads it before drafting, and the promise you plant in the cold open is still on the books when you reach the finale.