Trailer, Sizzle, or Proof of Concept: Choose the Moving Material for a TV Pitch

By pitch.dogFoundation12 min read20 September 2026

Television

Someone asks you for a video and names none of the three. Or names one and means another. Either way, you are now expected to produce a moving object that will be judged, and the judgment will land on the show — including the parts of the show that do not exist yet.

The three names are less useful than they look. Trailer, sizzle, and proof of concept are not grades of finish and not a ladder from cheap to expensive. They answer different questions, and the questions belong to the person watching. Get the question wrong and you can spend real money on something handsome that settles nothing.

So before the timeline, before the music, before anyone asks whether there is budget for a colorist: what does this recipient need to be able to judge?

Four answers cover most requests. They need to see a show that already exists. They need to imagine one that does not. They need to know whether one hard thing will work on screen. Or they need to hear from the person making it.

Translate "send a video" into an assessment question

Before choosing anything, take inventory of what actually exists, in four separate piles, because these tend to merge in the retelling.

Completed work. A finished pilot, a locked scene, an episode with sound and color. Anything you could put in front of an audience tonight.
Development footage. Material you shot on purpose but not as an episode: location tests, tone studies, a night with the real workers, no cast.
Specially made tests. Footage you would shoot in order to answer a question. The expensive pile, and the one that needs the tightest brief.
References. Other people's work, brought in to point at a quality you want. These cause the most trouble, because a reference sitting inside a montage starts to look like your footage.

Keep the piles apart. Most bad pitch videos come from one pile quietly borrowing the authority of another — a borrowed shot that reads as yours, a location test that reads as a scene, a scene that reads as a series.

When the request is ambiguous, ask. "What do you need to be able to judge?" is a better question than "how long should it be," and it usually changes what you make. The file name on the last video they liked is not evidence either; a file called sizzle_v4 is frequently a trailer.

To keep the four jobs distinct, here is one invented show at four points in its life. Call it Nightbloom: a half-hour scripted series about the overnight staff of a botanical garden. In the earliest state, no scene has been shot — only the garden itself, filmed over two nights for development. In the next, a single scene has been made as a test. Later still, a finished pilot exists. And in a fourth scenario, someone asks to hear from the creator directly. Nothing about this show is real; the four states are the point.

Use a trailer to select from work that exists

The trailer is the only one of the four that requires the thing itself to exist. That is its advantage and its limit. You are cutting from footage of the show, so what appears is real: real cast, real set, real scene.

What a trailer establishes is tone, situation, and the invitation to keep watching. It is an argument made by exclusion — this moment and not that one, this beat and not the four hours it took to shoot.

What it cannot establish is the ordinary episode. A trailer is built from selected moments, which is what makes it persuasive and what makes it silent about everything else. It will not tell a recipient whether episode four holds together, whether the show can be made on schedule, or whether the lead is available in March.

For Nightbloom in the state where the pilot is finished, the trailer draws from episode one and nothing else. It can show Ines and Cal in the glasshouse at three in the morning, and the misting line coming on. It cannot show that any of this repeats, because there is no second episode to draw from.

One habit worth keeping: say what the footage is and where it comes from. "This is the finished pilot" and "this is selected from the pilot" are different claims. A recipient who believes they are watching a series sampler and then learns it is one episode will trust the next thing you send slightly less, and they will be right to.

Use a sizzle to make a proposed experience legible

Rewind to the version of Nightbloom with no pilot. No scene has been shot. What exists is the place — two nights of development footage inside the actual garden, no cast — and a description of what the show should feel like after closing.

A sizzle's job is to make the proposed experience legible. Not the plot. The experience. What it means to stand in a working botanical garden at 3 a.m., and what the show's relationship to that place is.

What it can carry is an encounter, a way of observing, an intended tonal relationship. The garden at night is indifferent and busy: pumps cycling, timers firing, condensation running. The people on shift are the only ones who notice any of it, because noticing is the job. Cut correctly, a sizzle can make a recipient feel that relationship without a single line of dialogue.

What it cannot carry is dialogue, casting, or episode construction. Nor can it settle the ambiguity the show actually depends on. That is a different object.

This is where the brief has to be honest about its material. If the only footage you have is the garden, and the show you are pitching is built on two-hander scenes, the sizzle is a partial portrait — and partial is fine as long as it is labeled. Footage from the proposed show and footage that merely illustrates an intention should not sit in the same montage unmarked. A borrowed shot suggesting a quality of darkness is a reference; it does not become your show by being cut in, and whether you may use it at all is a rights question that has to be answered before the edit, not after. A slate or a covering line reading "development footage, no cast attached" is not a confession. It is the reason the recipient can trust the rest.

There is a quieter failure too. A sizzle that is beautiful and silent can leave a recipient unable to tell whether the show is funny, sad, or a documentary about the garden. If the montage does not decide, they will decide for you.

Use a proof of concept to expose one uncertainty

The proof of concept is the smallest of the four in scope and usually the largest in risk, because it is the one you have to make. It is also the only one whose value depends on including the difficult part.

Define the question first. For Nightbloom: can a single scene hold two readings of the same event at once, without the script or the edit declaring which one is correct?

The scene would be this. At 3:14 a.m. the misting line in Glasshouse 4 switches on. Ines, a horticulturist, hears a drifted timer — routine, hers to fix before morning. Cal, the contract security guard, hears someone who has been in the building — a person, his to find. One room, one minute, one sound, two maps of what is normal.

The sample, if you made it, would be those two actors, that room, that night, that dialogue, shot plainly. Not cut to music. Not intercut with b-roll of the garden. Because the uncertainty is whether two readings can coexist on screen, the sample has to let them coexist rather than smooth the problem away.

A favorable sample would establish three things: that the scene plays with two actors in one location; that the dialogue can carry both readings at once; and that an audience can follow without being told which one is correct. That is a real answer to a real question.

It would not establish that the device survives past episode one, that the series has enough incident to fill a season, that these actors are the eventual cast, or that the scene shoots on schedule. One scene in one room on one night is not a production plan.

The failure mode here is well known and easy to fall into: making the test as if it were a trailer. If your proof of concept is cut like a preview — fragments, music, no sustained scene — you have answered a different question and left the real one untouched. Fragments hide the difficult part, which is exactly why they are tempting. The test exists to include the part you otherwise could not prove.

Say in advance what a favorable result would not mean. Otherwise "the sample looked good" quietly becomes "the show works," and nobody notices the substitution until the sample is a memory and the schedule slips.

When the request is for a person: the recorded pitch

Sometimes the ask is not for the show at all. It is for you. "Send a video" meaning: explain it. Walk us through the season. Who are you, and what is this.

A recorded pitch establishes that you have the show in your head, that you can say it out loud in a level voice, and something about presence. It is evidence about the creator. It says nothing about execution, and a strong recorded pitch and an unmakeable show coexist all the time. That is not a flaw in the object; it is what the object is. Treating it as a fourth kind of trailer is how pitches get muddled.

The practical rule is not to smuggle one object inside another. If they asked to hear you explain the show, ninety seconds of mood footage with twenty seconds of you delivering half a sentence answers neither question. If they asked to see the world, a single locked-off shot of you talking into a laptop answers the wrong one. And if the on-camera route is the one they actually want, it has its own craft — framing, what to say in the first twenty seconds, where to stop — which is a separate conversation.

Write the smallest adequate video brief

Whichever route you take, the brief is one page and five lines.

The question. What the recipient needs to be able to judge, in one sentence.
The material. What exists, sorted into the four piles, with anything borrowed marked as borrowed.
The object. Which job the video is doing — and which single job it is doing primarily.
The necessary sequence. The smallest set of moments that answers the question. Not the most impressive set.
The remaining uncertainty. What this video will not settle, written down before you shoot rather than after.

Labels overlap, and they should. A sizzle with a forty-second proof-of-concept scene near the end is an ordinary file. What matters is that the brief names one primary job, because a video with two primary jobs gets judged on the wrong one.

If the recipient used a word — "sizzle" — you do not have to correct their vocabulary. You have to deliver the object that answers their question, and if that object is not what they named, say so in one line instead of arguing about definitions. "You asked for a sizzle; the useful thing we can show you is a single scene, because the question you raised about tonal control is a scene-level question." That sentence costs nothing and saves a great deal.

Then follow their delivery instructions. If they specified three minutes and a vertical frame, that is the specification. There is no universal length for a sizzle that survives contact with actual recipients, so do not invent one and do not treat a convention you half-remember as a rule.

And reject extra footage that does not help the assessment. Polish that does not help is the most expensive kind there is.

The comparison here is not a package. You do not owe anyone three videos, or four, or a trailer plus a sizzle plus a test. One adequate object beats three partial ones, and a creator who sends a single well-aimed scene is easier to work with than one who sends a showreel of intentions.

The brief I would want to end on has two columns. On the left: what this video will let you judge. On the right: what it will not. The left column gets you the meeting. The right column keeps the meeting useful, because a recipient who knows where your evidence stops can ask you for the next thing — and a recipient who was oversold finds out eventually, usually in the room where it costs the most.

Frequently asked questions

What should I ask before choosing among a trailer, a sizzle, a proof of concept, or a recorded pitch?

Ask what the recipient needs to be able to judge. Four common needs are seeing a show that already exists, imagining one that does not, knowing whether one hard thing will work on screen, and hearing from the person making it. A recorded pitch is evidence about the creator, not execution, and should not be mixed with mood footage.

What can a trailer establish, and what can it not?

A trailer requires footage of the show to exist. It establishes tone, situation, and the invitation to keep watching, and it should say where footage came from, such as a finished pilot or selections from the pilot. It cannot establish the ordinary episode, whether episode four holds together, the schedule, or cast availability.

What is a sizzle for, and what should be labeled?

A sizzle makes a proposed experience legible, such as an encounter, a way of observing, or an intended tonal relationship, not the plot. It cannot carry dialogue, casting, or episode construction. Development footage and borrowed references should be clearly marked, and rights questions must be answered before the edit.

How should a proof of concept be designed?

Define the single uncertainty first, then include the difficult part rather than hiding it in fragments. If the question is whether one scene can hold two readings at once, shoot that scene plainly with the actors, room, and dialogue. State in advance what a favorable result would not mean.

What belongs in the smallest adequate video brief?

The question the recipient needs judged; the material sorted into completed work, development footage, specially made tests, and references; the one primary job of the object; the smallest necessary sequence; and the remaining uncertainty. Follow delivery instructions, and reject polish that does not help the assessment.