How to Record a TV Pitch Video That Explains Your Show

Somewhere in the instructions you agreed to is a sentence that says you owe someone a video. Find it before you touch a camera, because it decides almost everything that follows — and it is rarely the sentence people read it as.

The reflex, when the request arrives, is to make something that moves. Music. Cuts. A slow push in on an empty street. That reflex produces a trailer, and a trailer answers a question the reviewer has not asked yet. A recorded pitch is a stand-in for a conversation you don't get to have. No one can stop you to ask what "the team" means, who the person on screen is, or whether the place is real. Every clarification you would have supplied out loud has to already be in the recording. That single constraint — the missing interruption — explains most of the decisions below.

Read the request before writing the recording

Take the instructions apart for five things: what object is being requested, what questions it has to answer, how long it may run, how it gets delivered, and what else is submitted alongside it.

The 2020 Episodic Makers Lab FAQ, which covers the 2019 application cycle, is a useful historical example of how specific those answers can get. The page listed Series Overview, Video Pitch, and Samples of Previous Work as separate submissions, and it treated a series overview as a different document from a pilot synopsis. Read plainly, that is a recipient telling applicants that these are three different objects, and that they want the one they asked for (apply.sundance.org).

That FAQ is not a rule you are working under. It belonged to one program in one cycle; its length limits were that program's instructions to that year's applicants, not the length of a good pitch. Nobody has inspected the videos submitted under it. What it usefully demonstrates is the habit: the recipient's own terms tell you what to make, and the phrase they used — "Video Pitch" — is worth borrowing rather than replacing with something more impressive-sounding.

Then sort the request into one of three jobs.

• Explain an unmade series. They want to understand the project. Spoken explanation leads; visuals support it.
• Demonstrate existing work. They want evidence you can do the thing. Show produced material and be exact about what you did on it.
• Sell a mood. They want a sizzle, usually for an audience that already knows the show exists.

These are different objects, and none substitutes cleanly for another. The most common mistake at this stage is answering a request to explain with material designed to excite.

There is a fourth possibility worth naming, because it is not a mistake: sometimes the honest answer is to read the deck aloud. If the instructions say the reviewer will have the deck open, or that the recording accompanies a live presentation, then the deck is present and can carry context the recording doesn't have to. The failure is not reading the deck. The failure is assuming the deck travels with the file.

Give the listener a route through the show

Once you know you are explaining, work out a route. A route is not an outline of your show; it is the order in which a stranger can absorb it. Four moves do most of the work:

1. A situation someone can picture. A room, a job, an event that keeps happening.
2. The people and the format, in that order — names before pronouns, roles before the names are asked to carry meaning on their own.
3. Why there is another one. For a continuing series, what regenerates the premise. For a finite series, why the progression has to go where it goes.
4. The connection the request actually asked about. If the instructions ask what qualifies you, answer in your own terms.

One example should run through all four rather than four separate examples, because the listener has to hold the whole thing in working memory at once.

What follows is an invented show and two invented scripts. Nothing here has been recorded, and no reader has been tested on it. The scripts are drafts on a page, offered for the difference between them.

Night Returns is a half-hour continuing drama set in a city's central lost-property depot — the place where the things people leave on buses, in cabs, and on park benches end up. Staff at the claims counter resolve competing claims to ordinary objects. The claimants change. Two staff members don't.

Here is the deck read-aloud, the version that works only when someone is clicking through slides beside you:

Night Returns is a half-hour drama set in the city's central lost-property depot. Every week, two people come in claiming the same thing, and their conflict drives the season. Rue and Owen are the heart of it. The depot is where the city's strangers meet, and the stakes are high. These are the people we follow — [character portraits appear] — and over the season you'll watch them decide what a thing is worth to whom. Night Returns is about what we lose and what we refuse to let go of.

Here is the self-contained version, same show, same length of program:

A grey coat comes into the city's central lost-property depot on a Tuesday, in a bin of things left on the 12 bus. Two women come in the same week to claim it.

Rue Alvarez has worked the claims counter for twenty-two years. She asks each of them the same question: describe the inside of the left-hand pocket. One of them can. Her mother repaired it with grey thread that doesn't match the wool. Rue has already looked. She isn't testing them — she's deciding which of two true accounts the depot can act on.

Owen has been behind the counter six weeks. He asks why she doesn't just hold the coat up and let them sort it out. "Then it's a contest," Rue says. "I'd rather it be a question."

That's one episode. Night Returns is a continuing half-hour drama, and each episode turns on a single contested object. The claimants change; Rue and Owen don't. The depot's engine is the morning delivery — a new bin of found property arrives every day, so the problem in the room never resolves. It only changes hands. [Original sketch of the depot: the counter, the shelves behind it, the distance a claimant walks to see her coat again.]

The reason I know this room: I spent five years working a claims counter before I wrote anything, and I learned that two people can describe the same afternoon with total conviction and both be describing what they actually saw.

The whole script is invented, including that last line; a real version substitutes the connection you actually have. The useful part is the repair list.

Names arrive with their jobs attached. The read-aloud introduces Rue and Owen, but what tells you who they are is the slide heading underneath — senior claims clerk, new hire. Remove the slide and you have two names and one vague noun, "the heart of it." The recorded version gives Rue a duration instead: twenty-two years on the claims counter. That single fact does the work a heading did, and it does something a heading can't — it makes the job feel like a life rather than a credit.

Pronouns get resolved. "Their conflict drives the season" is ambiguous in speech in a way it isn't in a document, where you can look back. Whose conflict? Rue and Owen's? The claimants'? The recorded version settles it by showing the conflict instead of asserting it, which costs about the same number of seconds.

An abstraction is replaced by a pocket lining. "The stakes are high" is a claim about a feeling. The repaired pocket lining is the feeling, and it also demonstrates the show's method: Rue knows the answer before she asks.

The engine is stated rather than implied. The read-aloud promises a season without explaining why there's a next episode. The recorded version points at the morning bin. A reviewer who understands what regenerates the show can imagine episode forty.

The ending changes audience. The read-aloud closes on a theme. The recorded version closes on the person who would be making the thing, which is what a project pitch is usually for.

The recorded version runs about a hundred and fifty words longer. That is the budget, and it is worth spending.

Assign each visual insert a missing piece of understanding

The test for any insert is to cover it and ask what the viewer loses. If the answer is "nothing," cut it. If the answer is "the whole premise," the audio is thin and the fix belongs in the audio, not in a bigger insert.

Two kinds of insert earn their place. A diagram handles a format the ear struggles with — an anthology that recasts each season, a show that alternates between two timelines, anything where the shape of the thing is the point. A sketch handles an unusual setting: cheap, honestly labeled as a drawing, and it establishes a room without pretending to be footage from a show you haven't shot.

In the invented example, the sketch has exactly one job: the geometry of the depot. How close the shelves sit to the counter. How far a claimant walks to look at her own coat. That last detail is a whole scene's worth of feeling, and it lives in a measurement — which speech conveys slowly and a drawing conveys at a glance.

Four rules keep inserts from becoming decoration.

Say what it is before it appears, in roughly the words a viewer would use. "Here's the layout of the room, looking from the door." A drawing that arrives unannounced reads as an unfinished graphic.

Keep talking. Keep the spoken line running under or immediately after the insert. A held image with silence hands the reviewer a pause exactly when you most need their attention on the explanation.

Never let a fact live only in the frame. If the sketch is the only place the depot's layout exists, a viewer who can't see it loses the premise, and so does anyone listening in a car. The picture confirms; the words carry. This also means captions for the speech, accurate ones, and — if the request names any required accessibility support — supplying it rather than assuming the reviewer will sort something out.

Don't use portraits as evidence of character. A face establishes what someone looks like. It says nothing about what they want from the other person, which is the thing the reviewer needs.

Repair the rough recording before polishing it

Record a rough pass. Not a good one — a complete one, out loud, in order, in a single take, before you tune anything. Then do four things with it.

Listen without the pictures. Screen off, or audio only. Every place you would need to raise your hand and ask a question is a gap. This is the single most useful edit you can make, because it reproduces the reviewer's actual condition: they are hearing your explanation, not reading along.

Review the sequence with the rest of the application closed. Field labels, the neighboring answers, the synopsis on the next page — none of that is present when someone presses play. If the recording only makes sense next to the form, it isn't finished.

Time it against the requested duration, not against what feels right. Pitch videos feel shorter to the person who wrote them.

When you are over the limit, cut repetition or narrow the example. Do not accelerate. Rate is how you buy a listener comprehension; speeding up a route that already fails to stop just fails faster. If you have three examples and you need two minutes, keep the best one and give it room. One object understood is worth more than three objects named.

There is a specific trap here worth naming, because it is invisible from the inside: a fluent read-aloud can sound completely polished and depend entirely on a slide you can't see. Fluency reads as clarity to the person who wrote it. It is the audio-only listen, not your own confidence, that tells you whether the explanation travels.

The two scripts above have not been through any of this. They exist on the page. The rough pass is the next thing that would happen to them, and it is the step that finds the missing premise.

Check the submitted object, not just the local file

The explanation and the submission are two different jobs, and passing one doesn't demonstrate the other.

Listen on the device a reviewer will actually use. A phone speaker is unforgiving in ways headphones are not. If music or room tone sits under the voice, the mix that sounds intimate in your ears can bury half your consonants through a phone.

Check the request's duration and format, and the delivery method. A file and a link are different objects with different failure modes.

Supply captions and any required support, and check that they're corrected rather than auto-generated and approximate — especially with names, which is what a pitch is full of.

Open the submitted route in a browser you are not signed into, or have someone else open it. Your own logged-in view will happily play a link a reviewer cannot access. The two are not the same test, and this is the failure that wastes the most work, because everything upstream was fine.

The recordable outline

Fill each slot with your show. Every spoken unit gets a job and nothing else.

1. The situation. One concrete thing happening in a place someone can picture. One or two sentences.
2. The people. Names, roles, and the relationship between them, said out loud. Job: replace the slide heading.
3. The form. Continuing or finite, episode length, and the pattern a viewer would notice in the third episode.
4. The renewal. What generates the next episode, or why a closed series has to move in the direction it moves.
5. The one example. Runs through slots 1 through 4 rather than restating them.
6. The connection. The answer to whatever the instructions asked about you. Your answer, not a template's.
7. The insert. Optional, one only, with its job written in the margin and a spoken line that introduces it. Cut it if you cannot say what the viewer loses without it.
8. The close. On the specific, not the theme.

Then the checks, in this order: the requested length; the audio heard on the delivery device; captions corrected; the link opened signed out; the file or upload in the requested form; the sequence understood with the rest of the application closed.

The recording is a substitute for a conversation that didn't happen. It succeeds when the reviewer finishes it knowing what the show is and why there is another episode — not when you got through your copy without a stumble.

Frequently asked questions

What should I pin down from the instructions before recording a pitch video?

Identify five things: the object being requested, the questions it must answer, the allowed running time, the delivery method, and what else is submitted alongside it. The body uses a historical program FAQ to show how specific those terms can get, while noting that FAQ belonged to one program and one cycle and is not a universal rule.

How do I decide whether to explain, demonstrate, or sell?

Sort the request into one of three jobs: explain an unmade series, demonstrate existing work, or sell a mood. They are different objects, and none substitutes cleanly for another. A common mistake is answering a request to explain with material designed to excite.

When is reading a deck aloud acceptable, and what is the failure?

Reading the deck can be honest if the reviewer will have the deck open or the recording accompanies a live presentation. The failure is assuming the deck travels with the file; if a fact lives only in the deck, the recording may not carry it.

What makes a visual insert earn its place?

Cover it and ask what the viewer loses. If nothing is lost, cut it; if the whole premise is lost, fix the audio rather than enlarging the insert. Diagrams can carry format, sketches can establish an unusual setting, and inserts should be announced, keep speech running, never hold a fact only in the frame, and not be used as character evidence.

How should I cut if the recording runs over the requested length?

Cut repetition or narrow the example; do not accelerate. Rate buys comprehension, so speeding up a route that already fails just fails faster. One object understood is worth more than three objects named.

Further reading

• Rehearse a TV Pitch as a Conversation, Not a Recital — Read if you will also pitch the show live: it builds a three-anchor spoken route you can reach from more than one direction.