Pitch a Commercial Built Entirely From Existing Footage

By pitch.dogFoundation13 min read20 September 2026

Advertising

A treatment for an existing-footage commercial looks almost exactly like a treatment for any other commercial: a logline, a beat structure, a music note, an end card. The difference is that every image in it has to have come from somewhere, and when a beat doesn't work you cannot shoot your way out of it. That is not a small inconvenience to note at the back of the deck. It is the shape of the pitch.

So the answer, up front: write the treatment from material you have actually watched and can describe, mark separately the material you want but have not found or cannot use, and put the argument where existing footage can genuinely carry it — in selection, order, sound, text, and contrast between shots. The idea has to survive the collection you have, not the collection you wish you had.

What follows is a worked example using an invented collection for an invented campaign, so the reasoning is visible without borrowing anyone's actual clips or permissions. It is a paper exercise, not a test of a real archive.

Establish what existing-only allows and excludes

Before the ideas arrive, read the assignment literally.

Must the footage come from a collection the client supplies, or are you searching? Is anything new permitted alongside it — a text card, a voiceover, an animated transition, a score? Do stills count as footage, or only moving images? What actually gets delivered: a sixty, a thirty, cutdowns, vertical versions?

Those answers decide what a treatment can promise. If the brief permits a newly composed score and new text but no new photography, then sound and typography are legitimate places to carry something the pictures can't. If the brief says the footage must come from the supplied collection, then a search result from elsewhere is a candidate to raise with the client, not a shot in your film.

This stage is where holes get papered over most easily. A line like "we cut to the moment she locks up" reads like an editorial decision. It is a request for a shot that may not exist. Nobody in the room flags it, because it sounds like craft. Three weeks later it has become a budget line for a half-day pickup, and the whole premise of the job has quietly changed.

A useful piece of boilerplate to borrow comes from the AICP Bidding Reference Guide (2026), a US production-association document of bidding questions, whose relevant pages were recorded at a check for this piece. It asks whether stock or other licensed material is needed, and who carries its budget. That is the right shape of question to ask early. It is not an answer: it establishes no rate, no entitlement to pitch payment, no actual budget or contract term, and nothing about whether any particular clip exists or may be used this way. Treat it as a prompt, not as support.

Keep two roles distinct. One person confirms that material exists and on what terms. A different judgment asks whether the proposed final use is acceptable — a clip that is fine as background texture may be a very different question when it is the image carrying your product's promise. Neither of those is a legal conclusion you write into your own treatment.

Inventory usable relationships rather than attractive clips

The basic unit of an existing-footage pitch is not the good shot. It is the relationship between two shots.

For each candidate, record the same handful of things: what actually happens in frame, the date and format as recorded, the location as recorded (or "not recorded"), what is visible that dates or places it, what the sound is doing, and the limitation. Watermark, resolution, partial action, a recognizable face, a brand, a performance — all of it belongs in the note.

Then make one distinction and keep making it: a collection's search description is a claim, and a clip you have watched is an observation. Write down which one you have. A vivid sentence in a catalog entry is not the same as a clip you have seen, and the difference matters most for the shot your whole treatment leans on.

With that noted, ask what the material can carry. Not "is this a nice clip," but:

a repeated gesture across unrelated footage
a change of space or scale
a contrast of viewpoint, era, or image quality
accumulation over time

A collection of people opening doors in five different decades can support "this keeps happening." It cannot support "this is the same room." Those are different pitches, and only one of them is available to you until someone confirms otherwise.

Write the edit's argument from those candidates

Here is the invented example. A fictional community-arts campaign — call it Room to Rehearse — for a fictional client. The stipulated brief: all images must come from the supplied collection; newly composed music, a voiceover, and text cards are permitted; no new photography; deliver a sixty and a thirty.

Five candidates have been inspected, all described in the collection index:

1978, 16mm, colour, faded. A hall with tall windows and a parquet floor. Two people pull a pair of double doors inward; a third leaves a wooden wedge under one leaf. Room tone, no sync dialogue. Building not identified. No key visible.
1996, Betacam SP, low light. The threshold of a converted storefront with a roll-up shutter. A person unlocks it with a key on a looped cord, rolls the shutter up, then hands the cord to a second person who ducks under and takes it. Location recorded as a storefront, no street name.
2014, vertical phone video, handheld, fluorescent flicker. A basement room. Someone flips a switch, tubes stagger on, two people drag a folding table to the wall. Clipped audio, laughter off-mic.
2021, DSLR, static. A room, lights off, chairs stacked, blinds half-open. No people. The description does not identify the building.
1969, 16mm, silent, black and white. An exterior street, brick facade, pedestrians passing in coats. No door is opened; nobody enters a building. Location not recorded.

The argument you want is simple: people make room for others. The first version of the sequence ran 1978 doors and wedge → 2014 lights and table → 1996 key handed on → 2021 empty room. The handover was the climax. Someone receives the means to open the space next time, and then we see a room at rest.

Notice what makes that work: the shots are not claiming to be one event. Four decades, four formats, and locations the treatment does not claim are the same. The card at the top sets those terms, and the differences in grain, aspect and camera behaviour carry the meaning rather than fighting it. The phone video is rough — vertical, handheld, clipped audio — and that roughness can read as content rather than a flaw to hide. The note does not record who shot it, so participant filming is an inference to confirm, not a property of the clip. Mixed eras can serve an idea, but only when the edit is choosing them. An apparently continuous story built out of unrelated material is a different thing, and it is usually a lie of omission.

One more check inside this beat: the 2021 room. The description does not say which building it is. If you want the audience to read it as the 2014 basement emptied out, that is a claim needing confirmation, not a mood to imply with a slow crossfade of room tone.

Expose the shot the proposal does not yet have

Now take the 1996 handover out. In this example, stipulate that the clip sits in an external collection rather than the one the client supplied, and its availability and terms for this use are unconfirmed. That is enough to disqualify it from the sequence you are promising.

The revised sequence, using only the four confirmed candidates: 1969 exterior → 1978 doors and wedge → 2014 lights and table → 2021 empty stacked room. It is chronological, so it reads as recurrence rather than story.

And the argument changes. It is no longer about a depicted handover. It is about traces of one, and the traces are what the footage records: the wedge in the doorway, the chairs stacked rather than scattered. The reading we put on them — that someone could have taken the wedge and didn't, that the room was left ready for whoever comes next — is an interpretation the edit and the card supply, not something the clips state. That is a slightly quieter idea, and it is honest about which part is which. The 1969 exterior works at the top precisely because a card declares the terms up front: different rooms, different decades. With that framing, a street where nobody enters a building makes a fair opening statement. Without it, the same shot is a destination nobody earned.

Which brings us to the edit you should be able to describe as a mistake. Keep the handover, cut on the movement of the hand, and land on the 1969 street with street ambience carried over. It will feel good. It is also three fabrications in one cut: that the 1996 threshold and the 1969 street have any relationship; that the exterior moment comes after the handover, when the stock dates it twenty-seven years earlier; and that anyone in that shot enters a building, which it never shows. Matched motion and a sound bridge will paper over all of it so smoothly that the audience never notices the leap — which is exactly why craft working against accuracy is worth naming in the treatment rather than leaving to the timeline.

Keep the beat as a contingent proposal instead. Say what the film looks like with it, say what it looks like without, and put the condition in writing: "The 1996 handover adds a beat to the sixty. If it cannot be confirmed for this use, the sixty runs without it, with the 1978 wedge carrying the same idea at greater length."

Three labels do most of the work here, and they should be visibly different in the document, not blended into one hopeful paragraph:

Inspected and available — watched, described, in the supplied collection.
Desired but unfound — the beat exists in your head and not in any archive you can point to.
Found but unresolved — real, seen, and not yet usable. A brand in frame, a minor, a performance, a private moment, a resolution too low to finish with.

Then guard the boundary between them, because it erodes. A frame grab pasted into a deck becomes "our shot" by the second meeting. A proxy that looks beautiful at eight seconds gets remembered as final. A personal-archive clip used as a placeholder stops being a placeholder. Label it on the image, in the caption, and in the section heading — or, better, describe it in words instead of showing the picture. A description cannot be mistaken for a delivered asset.

Plan substitution without pretending every clip is interchangeable

For every beat, decide what a replacement has to preserve, and whether losing the original changes the idea or only its surface.

In the example, the 1978 wedge shot is load-bearing. A replacement needs a person who could have removed something, and that something left in place, in the same frame as the doorway. A clip of a door being pulled shut would invert the argument. There is no substitute for this one, and saying so is more useful to the client than offering four options that quietly don't work.

The 2014 basement beat is scenic. Any clip from the collection showing a light switched on in an unused room would do the job — the specifics of the folding table carry no argument. This is the beat where you can be generous with alternatives.

The 1996 handover is contingent. It strengthens the sixty and is not required by anything.

That is the shape of a substitution table worth showing: two structural beats, one flexible beat, one dependency. A flat list where every shot has an "or similar" column tells the client that footage is interchangeable, which is the assumption that gets pitches into trouble in the first place.

Carry the sourcing and use questions to whoever owns them. The treatment's job is to make the dependency visible — "Clip 2 depends on confirmation of availability and terms; if unresolved by the date we lock, the sixty runs without it" — not to supply a clearance or call something safe because a search page let you preview it.

A sequence you can defend

Here is the thirty, using only inspected and available material:

0:00–0:05 — 1969, silent 16mm, exterior street, pedestrians, nobody enters. Card: "Different rooms. Different decades."
0:05–0:13 — 1978, doors pulled open, wedge left under the leaf.
0:13–0:22 — 2014, switch flicked, tubes stutter on, table dragged to the wall, laughter off-mic.
0:22–0:28 — 2021, empty room, chairs stacked, blinds half-open, lights off.
0:28–0:30 — Card: "Leave the room ready."

The music is newly composed and carries across all four; the phone audio stays in under the 2014 beat and nowhere else, so the texture break is a decision rather than an accident.

Three things remain unresolved, and they belong in the pitch as open items rather than footnotes. Whether the 1996 handover can be confirmed for this use, which affects the sixty only. Whether the 1969 exterior was shot in the same city as any of the interiors — the location is not recorded, and nothing in the sequence should imply it either way. And whether the 2021 room is related to any other interior in the film, which the current sequence assumes it is not.

The substitutions we would offer: the 2014 beat can be a different segment of the same phone clip. The 1996 handover can be dropped. The 1978 wedge shot has no replacement that preserves the argument, and we would rather say that now than discover it in an edit suite.

A pitch built this way is not a weaker pitch than one with a fictional shot list. It is the version that survives contact with a production meeting, because every image in it is either something you have watched and can describe, or something you have clearly marked as not yet there. The idea lives in the selection and the order and the sound, in the space between a wedge in a doorway and a stack of chairs, and none of that requires inventing a history the footage cannot support.

Frequently asked questions

What should an existing-footage-only treatment treat as its basic unit?

The relationship between two shots, not an isolated attractive clip. The treatment should argue through selection, order, sound, text, and contrast, and the idea must survive the collection actually available.

How should material be labeled so a pitch does not drift into fiction?

Keep three visibly separate labels: inspected and available, desired but unfound, and found but unresolved. Guard the boundary, because a frame grab or proxy can start being treated as a delivered asset.

In the invented Room to Rehearse example, why does losing the 1996 handover change the idea?

The handover sits outside the supplied collection and its availability and terms for the use are unconfirmed, so it cannot be part of the promised sequence. The revised chronological sequence becomes about traces of care rather than a depicted handover, and that reading must come from the edit and card, not from the clips.

How should substitutions be planned for an existing-footage film?

For each beat, decide what a replacement must preserve and whether losing the original changes the idea or only its surface. The 1978 wedge shot is load-bearing, the 2014 basement beat is scenic, and the 1996 handover is contingent; a flat list where every shot has an equivalent tells the client footage is interchangeable.

What is wrong with using matched motion or a sound bridge to join unrelated archive shots?

It can hide three fabrications at once: a relationship between shots that does not exist, a chronology the stock dates contradict, and an action the footage never shows. The article says to name that kind of craft working against accuracy rather than leave it to the timeline.