Write an Editing Treatment From the Commercial Rushes You Actually Have
Write an Editing Treatment From the Commercial Rushes You Actually Have
An editing treatment is a proposal about footage, not a restatement of the script. The shooting treatment describes the film everyone intended to make. The editing treatment describes the film the takes can be assembled into, names which takes carry which beats, and admits what is still absent. Those are different documents, and confusing them is how a commercial reaches the edit with a confident plan and no shot to support it.
The reason this matters is structural. A board or a shot list records intent: reaction, hero product, spill. The rushes record what happened on the day, take by take, with whatever continuity the day actually produced. An editing treatment works from the second list. Where the first list still governs, it does so as a request for new material, kept visibly separate from what exists.
A note before the demonstration. The rushes inventory below is invented for this exercise. No footage has been shot, viewed, logged, selected, cleared or judged usable. I have inspected nothing; the reasoning is paper reasoning, and I mark it as such where it matters.
Start by separating the brief from the capture
Two documents open the job. First, the brief and script: what experience is being promised to the viewer, and in what order the commercial expects to deliver it. Second, the rushes: the recorded takes available for review. Rushes are what came back from the day, not the selects. A log can mark every file; only inspection tells you which take holds the performance you need.
Confirm which files and versions are in scope before proposing anything. A camera card, a proxy set, a second unit's material and a re-encode are not interchangeable. If a version is unaccounted for, that is a scope question, not a footnote.
The discipline here is narrow. Because a close-up appears in the board, it does not follow that a close-up exists in the rushes. Because the schedule allotted a reaction shot, the reaction may have been logged under a different name, or shot, or missed. The editing treatment begins with that uncertainty stated plainly, then resolves what it can from the material at hand. Everything after this point rests on that resolution.
Log moments, not labels
A shot label tells you what the crew meant to capture. A log entry tells the editor what a take can actually do. The useful unit is the moment: where an action starts and where it finishes, whether the person's response lands, whether the product reads at the size needed, whether the sound is usable, and how this take differs in continuity from its neighbours.
Identifiers and time references matter for the same reason. "Take 4, close, 00:12–00:19, hand enters frame at :13, label faces camera" lets another editor find the same moment and disagree with you about it. "Reaction shot" does not.
Two traps sit close together here. The first is treating a file named for a reaction as if it contains the realization the scene needs. The name records intent, not outcome: the take may hold a flinch, a blink, or an actor waiting for a cue. The second is blurring an observed limitation with an editorial judgment. "The spill was not captured" is an observation, verifiable against the log. "A cutaway to the glass would cover it" is an interpretation, and should be labelled as one, because it may not be true: masking the cause of an event the viewer is meant to understand leaves the viewer confused, not satisfied.
Keep the two kinds of statement in separate columns or separate sentences. The log earns trust by staying factual; the proposal earns a hearing by being visibly a proposal.
Build two sequences the footage can hold
An editing treatment becomes useful when it stops describing and starts choosing. The move is to construct two bounded sequence proposals around different available strengths, then say what each one costs.
Take a worked case. A short drink-serving commercial. The intended dramatic beat is a spill: someone reaches for the glass, the glass goes over, the moment turns into an apology and a shared laugh. The board has the spill. The rushes, in our invented inventory, do not.
Invented rushes inventory
Every time below is a source timecode read off the file, not a position in an assembled sequence. Where a proposal enters a take, it names the source timecode it enters at.
- T1 — complete handoff. 00:03–00:14. Server brings the glass, sets it down, hands it across; recipient takes it, nods, drinks. Continuous action, one take, no cut points hidden inside. Product visible at the start of the handoff, mostly occluded by the hand at the moment of transfer. Light and dress not recorded in the log.
- T2 — product close-up. 00:31–00:35. Locked-off, glass centred, condensation and pour visible. Strong, clean, no people in frame. Nothing happens except the product.
- T3a — reaction. 00:47–00:51. Recipient looks down at the glass, small smile, gaze holds on the glass.
- T3b — reaction. 01:02–01:06. Same person, same setup, but the jacket collar is different and the light has moved warm.
- T4 — hand enter, no spill. 01:19–01:22. A hand reaches toward the glass and stops. No contact, no topple.
- Intended spill: not present in the inventory. No take records contact, topple, or aftermath.
Proposal A — the available exchange. Open on T2 for two beats so the product is established cleanly. Cut to T1 at 00:05, entering after the handoff has begun, letting T1 carry the transfer, the nod, and the first drink. Close on T3a. T1 is internally continuous — within the take, the viewer sees one action completed. The join to the response is only provisionally continuous, because nothing in the inventory yet establishes that T3a's collar and light match T1's; until the frames are set side by side, that cut rests on an assumption. Within those limits, the sequence retains the human exchange and the product, and sacrifices the spill. The absence is not hidden; the sequence simply never promises an event it cannot show. The risk: T2 to T1 is a jump from locked-off product to active handoff, so the edit must let T2's last frame sit a beat longer than feels natural to make the cut read as a deliberate introduction rather than a stumble.
Proposal B — product-led. Open on T2 and stay longer, letting the pour and the condensation carry the opening. Cut to T4 at 01:19 — the hand approaching, the near-contact — and hold on the approach rather than the topple. Cut back to T2 for a short closing beat. This sequence is product-forward and survives without any interpersonal continuity, because it does not depend on a reaction that has to match. It sacrifices the exchange and the emotional turn. The risk is different in kind: holding on an approach that never resolves may read as a tease the viewer expects to pay off. On paper, this is the cleaner assembly to cut; whether it works is a question only an assembly can answer.
Notice what the two routes share. Both depend on T2 and on T1 or T4; both are honest about the missing spill. The difference is which strength each is built around, and each has its own cost.
The reaction takes complicate Proposal A, and they are where this demonstration's own rule gets tested. A log records action, not performance, so any statement about what a face is doing is an interpretation and belongs outside the inventory. T3a and T3b differ in collar and light, and nothing in the inventory establishes which one matches T1. T3a's small smile is on the record; that it plays as pleasure — and that T3b plays as mild surprise rather than pleasure — are editorial reads, provisional, and on paper they stay provisional, because no frame here has been viewed. A provisionally selects T3a for the pleasure read, subject to checking that match. B uses neither reaction. Their different dependencies belong in the treatment: A needs a convincing human join, meaning a reaction that matches T1 in collar and light; B needs the unresolved reach to feel intentional.
Name the missing material, then keep it separate
If neither route serves the brief, the honest move is to name the precise absent action rather than reach for a fix and call it an answer. Here, the absent action is the spill itself: contact between hand and glass, the topple, and the aftermath. That is a specific missing moment, not a general shortage of coverage.
Three remedies are available, and each is a new decision with its own dependencies, not footage already in the edit.
An agreed pickup. Shoot the spill: hand enters, contacts the glass, glass tips, liquid spreads. This is a new filming requirement, with its own scope — a location, a person, a prop, a take of the spill that matches the light and dress of T1. Since the inventory does not record T1's light or dress, "matches T1" is not yet a specification but a check to be run against the frames; until that check is done the pickup cannot be scoped or costed. It is not executable until approved and scheduled. Whether pickups are even possible is a production question I cannot answer from the material.
A graphic or sound solution. A graphic or a sound effect can be a legitimate element, but they are decisions with costs. Sound alone does not create the absent visual event: a crash of glass over T4 tells the viewer something happened but does not show the hand meeting the glass, and the viewer may feel cheated rather than served. A graphic that announces the product is a different proposition from a graphic that fabricates an action the footage does not contain. Both need to be described accurately, then judged on their own terms.
A narrower creative ambition. Rewrite the beat the footage can carry: the moment of near-contact, the almost-spill, or simply the pleasure of the drink. This keeps the film traceable to takes, because it asks nothing new of the rushes. Its cost is the emotional turn the spill was meant to deliver, and a filmmaker should be able to say whether that cost is acceptable rather than discovering it in the edit.
Keep those three in a separate section of the treatment, under a heading that says plainly they require approval. The end of the document should let a reader distinguish what exists from what needs to be made, and each proposal's beats should point back to a stated take or forward to a stated requirement. A treatment that cannot be traced that way is a wish list.
Let the material change the plan
The test of an editing treatment is not whether it sounds confident. It is whether an editor, given the same rushes, could execute every beat, and whether the reader can see which beats would break if a take turns out to be marked differently than the log suggests.
So the treatment's last page should do two things, and the first one has to be conditional. It should offer an edit proposition supported by identifiable takes — but the proposition turns on a single deciding criterion: whether the brief can survive without the spill. If it can, propose A, built on T2, T1 and T3a, and say why A rather than B: A keeps the human exchange the brief was written around, while B, cleaner to cut on paper, gives that exchange up and delivers a product film instead. If the brief cannot survive without the spill, A and B stop being propositions and become fallbacks; the primary route is the pickup or the rewrite, named as a requirement with its approval still outstanding and its match to T1 still unverified, and neither fallback is worth assembling until that call is made. Second, a separate, clearly labelled list of what it still needs: the spill pickup or the rewrite if the brief requires one, a decision on whether the near-contact beat is enough, and a review of T3a and T3b against the actual frames for collar, light and the match to T1. Those are the open questions the assessment leaves, and naming them is part of the work.
The footage gets to rewrite the plan. A treatment that respects that gives the next person at the desk something they can actually cut—and names the moment they cannot.
Frequently asked questions
How is an editing treatment different from a shooting treatment?
A shooting treatment describes the film everyone intended to make. An editing treatment describes the film the takes can actually be assembled into, names which takes carry which beats, and admits what is absent. It works from rushes, not from the board or script's intentions.
Why should rushes be logged by moments rather than shot labels?
A label records what the crew meant to capture; a moment records what a take can do: where an action starts and finishes, whether a response lands, whether the product reads at the needed size, whether sound is usable, and how continuity differs. A file named for a reaction may hold a flinch, a blink, or an actor waiting for a cue.
In the drink-serving example, why can't Proposal A simply hide the missing spill?
The spill itself, meaning contact, topple, and aftermath, is not in the rushes. Proposal A can retain the human exchange and the product by building on the complete handoff and a reaction take, but it sacrifices the spill. The reaction join is only provisionally continuous because nothing establishes that the reaction's collar and light match the handoff's; that has to be checked.
What are the three remedies when the missing spill is required?
An agreed pickup: a new filming requirement with its own scope, matching the handoff take's light and dress to be checked, and not executable until approved and scheduled. A graphic or sound solution: legitimate but costly, and sound alone does not create the absent visual event. A narrower creative ambition: rewrite the beat the footage can carry, such as near-contact or the pleasure of the drink, accepting the loss of the emotional turn. All belong in a separate section marked as requiring approval.
What should decide whether Proposal A or Proposal B leads?
Whether the brief can survive without the spill. If it can, propose A and say why over B: A keeps the human exchange, while B is cleaner to cut on paper but gives that exchange up for a product film. If it cannot, A and B become fallbacks; the primary route is the pickup or rewrite, named as a requirement with approval outstanding and its match to the handoff still unverified. A separate check should also flag the near-contact decision and the review of the reaction takes against the frames.