What each department actually needs out of a script breakdown
A breakdown is not one list. It is eleven lists that have to agree with each other, which is why the spreadsheet version holds together right up until the first revision.
7 min read
Breaking down a script is simple to describe. You read it once, slowly, tagging every physical thing each scene needs: who is in it, what they wear, what they touch, where it happens, what has to be built, broken, driven or removed in post. The tags become lists, the lists become counts, and the counts become a day count and a budget.
What gets underestimated is the output. It is not a document but eleven documents, for eleven readers who never read the same columns and all have to end up with numbers that match.
The 1st AD is reading it for shape, not detail
Six numbers decide the schedule: how many scenes, how many total page eighths, how many of them are night, how many are exterior, how many company moves the script implies, and how many cast days fall out of it. Those six decide the number of shoot days, which is most of the budget.
The AD does not need to know which lamp. They need to know that scenes 22, 23 and 31 share a set so they can be shot together, and that 23 is night so it cannot sit in the middle of a day block. On a commercial the page count stops being useful and setups take over: ten to fourteen setups is a day, whatever the pages say.
The line producer is reading it for five expensive words
Stunts. Animals. Minors. Special effects. Picture vehicles. Each of those turns a line of script into a different production.
One sentence, the car clips the barrier and keeps going, converts a fourteen person unit into something closer to thirty: a stunt coordinator, a precision driver, a safety officer, a doubled vehicle, an insurance rider and a location that will allow any of it. Finding that sentence in week one of prep costs an afternoon. Finding it in week three costs a shoot day.
Extras counts belong in the same read. Twelve background artists is a casting call; a hundred and twenty is a unit of its own, with transport, a catering line and a holding area that in summer has to be cooled.
Art needs a distinction the tag does not make: hero props versus dressing
“Lamp” is not a prop. “A practical lamp the actor switches off in 22, which has to be the same lamp in 23, 31 and 44” is a prop. The three questions art asks of every tag are: does an actor handle it, is it on camera doing nothing, and does it get destroyed.
The answers change the number. Handled props need doubles because they get dropped. Anything that breaks on camera needs three, because the second take exists. Anything branded or period needs clearance, and clearance runs on its own calendar.
A breakdown that gives art a list of nouns has given them a shopping list. What they needed was a count.
Costume and makeup read the script in story order, not scene order
This is the one most breakdowns get wrong, and the most expensive.
Scenes 22 and 23 might shoot in the same morning and be a week apart in the story. That is two looks, not one. Scenes 14, 22 and 44 might shoot across three weeks and all be the same story day, which means the shirt still has the same coffee stain in April that it had in March.
So the unit costume works in is not the scene, it is the story day. A breakdown that tells a costume supervisor which scenes an actor appears in is close to useless. What they need, per character, is the story day, the scenes inside it, the condition (clean, wet, dusty, bloodied, distressed) and therefore how many duplicates of the garment to buy. Hair and makeup read the same structure, plus lead times for anything prosthetic.
Camera and sound need to know what the scene does, not only what is in it
Camera is reading for method, because method is kit and kit is a day rate: handheld, dolly, crane, drone, car rig, high speed, macro insert, underwater. A scene tagged only with its cast and props tells camera nothing about whether tomorrow needs a grip truck.
Sound is reading for playback, for dialogue that comes through a phone or a radio, for crowd, and for the thing almost nobody tags: what the location sounds like. In the Gulf that list is specific. Construction that starts at 07:00 and runs all day. The call to prayer, five times, on speakers you do not control. Flight paths over parts of Jeddah and Riyadh. Wind that picks up across open desert in the afternoon and puts grit in every take. None of it is in the script. All of it belongs in the breakdown, because it decides what gets shot when.
Locations is reading for permission, which runs on calendar time
Interior or exterior is the easy half. The half that matters is controllable or public. A private villa, a studio or a warehouse you have hired is a scheduling question. A street, a mall, an airport or a government building is a lead time question, measured in weeks, with a drone adding its own approval on top.
The breakdown is the first document where you can see how many of those you have, and it is the last moment the answer is cheap.
Post and VFX need the method, not the word “VFX”
A scene tagged VFX tells you nothing you can act on. Clean plate, screen comp, wire removal, set extension and sky replacement are five different requests to make of the unit on the day. A screen comp means the phone needs tracking markers and someone has to shoot the plate. Wire removal means a locked off plate before the wires go on. If the breakdown does not say which, the plate does not get shot, and post pays for a roto artist to do by hand what thirty seconds on set would have given them.
Why the spreadsheet breaks, in four specific ways
- One row per scene cannot hold a many to many relationship. The practical lamp appears in nine scenes, so it is typed nine times, and by the ninth it is spelled three ways: practical lamp, Lamp (practical), lamp- practical. Filter the column and you get two props. Every cell reads correctly on its own, which is why nobody catches it.
- Revisions do not behave like edits. Revision colours exist because pages change under you: white, blue, pink, yellow, green, goldenrod. Scene 14 becomes 14 and 14A. Scene 31 is omitted but keeps its number so nothing renumbers. Pages gain A-pages so the page count stays stable. A spreadsheet knows none of this happened, so a person re-reads the new pages and hand merges at 23:00 on day nine of prep.
- Everyone forks it. Costume copies the master into their own tab and adds four columns. Art does the same. Two days later the master changes and there are three breakdowns, each of them real to somebody.
- The counts cannot be trusted, and the counts are the whole point. Cast days are derived by counting distinct actors per shoot day. One misspelling splits one actor into two, which makes a cast day appear or vanish, which moves the budget by a day rate plus per diem plus travel. That error is invisible on the page and visible in the cost report six weeks later.
A bilingual script breaks it one step further
Productions here are routinely mixed: an English script with Arabic dialogue, or an Arabic script with a client reading the breakdown in English. Spreadsheets handle mixed direction text badly, and a cell holding an Arabic character name next to a Latin scene code renders in an order that is technically correct and practically unreadable.
Then there is transliteration. Khalid, Khaled and Khālid are one actor to a human and three actors to a filter. Naming stops being cosmetic the moment your counts are derived from strings.
What good looks like: tag once, as a thing, not as a word
Every element exists once, with an identity, and scenes reference it. Rename it and it is renamed everywhere. Counts are derived from the references, so they cannot drift away from the tags. A new revision is compared against the previous one, so you read what changed instead of re-reading the script. And each department gets a view rather than a copy, which is the difference between eleven filters on one truth and eleven documents disagreeing politely.
That is the shape of what Klaket does with a script. It reads the pages, proposes the tagging, counts the eighths and the cast days, and hands you a breakdown to correct rather than one to build. The elements are entities, not strings, so the lamp is one lamp. When a revision lands, you get the diff. Departments read the same breakdown filtered to their own columns, in Arabic or English.
A breakdown is finished when two people get the same number
Tagging every scene is not the finish line. The finish line is the costume supervisor and the line producer, opening it on different days, counting the same number of cast days and the same number of looks. Until that happens you have a document, not a plan, and the difference will show up on the day.