Use a Table Read to Improve the Pitch for a Dialogue-Driven Series

A page can be reread. That is the whole reason a dialogue-driven scene can look solid in a pitch document and fall apart in a room.

On the page you can hold two speeches side by side, notice that one "fine" is answering a different question than the other, and decide the ambiguity is deliberate. A listener gets one pass and no margin. If they lose track of who owes what to whom, they don't lose a line — they lose the scene, and they often can't tell you where it went.

A table read is a cheap way to find that difference before someone with a development budget finds it for you. Not because a read proves the show works. It doesn't, and it can't. It produces something a pitch conversation rarely gives you: other people's accounts of what they heard, in their words, before you explain what you meant.

Run it with one question and a plan for what each kind of answer would change. There are three things that might need revising, and they are not interchangeable: the scene, the sample, or the sentence in your pitch that describes them.

Name the promise before you book the room

Start with the claim your pitch already makes. Not the premise — the claim about how people talk to each other.

For a fictional example, suppose you're pitching Ferndale, a half-hour series set in a small bookshop, and your pitch says this:

Everyone here is fond of everyone else, and everyone is keeping score. The staff meeting is where the affection and the leverage happen in the same breath.

That sentence is a promise about dialogue. It says the show's pleasure comes from watching warmth and pressure occupy the same exchange, repeatedly, episode after episode. It is also, usefully, a promise that can fail in a specific way. If a listener hears only the pressure, you have not written the show you described.

Ask one question, and make it about obligations

The obvious question — "is it funny?" — wastes a read. A yes tells you nothing about where to cut, and a no tells you even less. Laughter is also poor evidence on its own: a room can laugh at a line it didn't understand, or in relief at a line that was strange, or out of politeness to the person who brought snacks.

Ask instead about the thing dialogue actually has to do: build a change of state that a listener can follow in one pass.

For the bookshop scene, the bounded question is:

By the last line, what has Owen agreed to do, and when?

That question is narrow enough to answer, has nothing to do with whether anyone liked it, and sits directly under the pitch's promise. If listeners can name the obligation but can't hear any fondness in it, the scene and the description disagree. If they can't name the obligation at all, the exchange may be charming and unintelligible at once.

Pick the question that matches your own promise. If your pitch sells "politeness concealing rivalry," ask who was being polite and who was losing. If it sells a running joke, ask what the joke's setup was and what the punchline changed. One question. Written down before anyone reads.

Choose a scene where the dynamic costs something

A recurring dynamic needs a scene where it does work — where someone's position moves, or doesn't, because of how the conversation goes. A witty exchange demonstrates a register. That's not the same thing as demonstrating a series.

Also check that the scene's dynamic lives in the words. If the moment you most want a listener to feel is a look across the shop, or a cut from the till to the stockroom, a read will remove exactly the thing you're selling, and you'll spend the session learning about your own staging rather than your dialogue. Put a different scene on the table first.

Here is the invented scene for Ferndale. It has not been read as part of writing this article — there is no listener result below, and no sheet filled in. Write your own version of the last section only after you've actually run the exercise.

FERNDALE BOOKS, BACK ROOM. 8:40 A.M.

A standing meeting before opening. NADIA has the rota. BEA is by the noticeboard.

NADIA: Owen. Dacre. OWEN: Forty-one. NADIA: Forty-one copies in one evening. Nobody in this shop has moved forty-one of anything. OWEN: The chair helped. NADIA: The chair was Bea's. OWEN: Then the chair helped. NADIA: You never take the credit. I've noticed that about you. OWEN: I took the credit for the chair. NADIA: I've put you down for Sunday. The rep's coming in with the spring list. OWEN: Sunday. NADIA: Elaine's away, and I'd rather have you in the room than anyone. You know I mean that. BEA: (reading the rota on the wall) The rota says ten till six. NADIA: The rota's old. OWEN: It's fine. BEA: Which one's fine? BEAT. NADIA: We open in twenty minutes.

Two things to notice about the construction, because they affect what the read can tell you.

First, the promise is genuinely in the text. "I'd rather have you in the room than anyone" is warmth and a maneuver in one line, which is what the pitch claims the show does. A read can fairly test that.

Second, the obligation is deliberately unresolved. Owen says the situation is fine without naming which hours he means, Bea asks the precise question, and Nadia closes the meeting instead of answering. You should know whether that reads as productive ambiguity or as a dropped thread. Your listener accounts are how you find out.

Prepare a read, not an audition

A read of one scene, with three or four people and a stated twenty minutes, is a diagnostic. It is not an audition, and nothing you learn from it should be recorded as a casting judgment — a colleague reading cold off a screen tells you nothing about who should play the part.

What matters is that you can describe the conditions afterward, because the conditions are going to be your leading alternative explanation for anything that goes wrong. Write them down before you start:

• Who read which part, and whether anyone read two.
• What each person already knew about the show. "Nothing" is a useful answer and a different test from "heard the premise once."
• The context you gave aloud before the first line. Say where we are, what happened yesterday, and what the standing meeting is. Do not say what the scene is about, and do not say that the warmth is supposed to undercut the pressure. That's the answer you're testing for.
• Which action lines you voiced and which you skipped. A listener cannot see who is looking at the rota. Decide in advance whether "beat" gets read aloud, because a voiced pause and a silent one produce noticeably different accounts, and you need to know which version you ran.
• Whether you're recording, and what you've told people the recording is for. Ask first, agree on it plainly, and don't later repurpose a rough diagnostic into a demo.

Then read it once, straight through, no stopping. The whole value of the exercise is that the first pass is unrepeatable, and the moment you pause to explain something, you've spent it.

Collect accounts before you explain the effect

Ask open questions first. The order is the method.

Let each listener describe what happened in their own words, uninterrupted. Only then ask your bounded question — who agreed to what, and when. Only after that should you say what you intended, and even then, keep your explanation in a separate column. A listener who has just heard your intention will often adjust their memory of what they heard, and that adjustment is not evidence about the scene.

Keep laughter in a separate row. Record that it happened and where. Do not let it stand in for comprehension, and do not let its absence stand in for failure.

A sheet like this stays blank until the read actually happens:

Conditions: date / who read / what they knew going in / context given aloud / action lines voiced or skipped / recorded, and what you said it was for.

Notice that the third row is the bounded question and the first row is not. That order is what keeps a free account from collapsing into a yes or no.

Separate a scene problem from a reading condition

A confused listener is not automatically a broken script. Before you rewrite, walk through the competing explanations, because at least one of them will fit almost every bad account:

• Unclear intention in the text. Owen answers "it's fine" without naming which hours he means, and nobody on the page ever settles the hours. That's a property of the script.
• Missing action. The person reading may have skipped the rota line, or read it flatly, leaving the disagreement between the wall rota's ten till six and the hours Nadia never names invisible to the ear.
• Delivery. A reader can turn Bea's question into a joke, a sneer, or a genuine inquiry. Those are three different scenes, and the room heard one of them.
• Unfamiliarity. Someone hearing Ferndale for the first time has to learn three names and a workplace in forty seconds. Some of the confusion belongs to the first thirty lines of any scene in any show.
• Omitted visual cue. If the scene quietly depends on Nadia looking up before she says "Sunday," the read has thrown that away.
• Room conditions. Crosstalk, a reader doing two parts, a phone buzzing.

The practical test is whether the confusion survives a change of conditions. If you run the scene again with different readers and voice the action lines this time, and the same question goes unanswered, the text is the likelier culprit. If the confusion moves or disappears, you've learned something about your own preparation.

Two cautions on a second read. Use listeners who weren't in the first room, because the ones who were now know the answer you're looking for. And accept that a repeat is a different exercise: the scene is no longer arriving cold for anyone, including you, and that changes what gets heard.

Then decide which of the three things changes

The read isn't finished until it produces a decision. There are four exits. Three revise a document; the fourth sends you back to the read.

Revise the scene when listeners track the obligation but not the relationship, or when the turn you intended exists in your head and nowhere in the lines. The usual fix is concrete rather than decorative: let someone name the hours, or let Owen answer the question Bea asked. That isn't dumbing down a subtle scene. It's making the stakes audible in a medium that only gets one pass. Subtlety that no one can hear is just a private note.

Change the sample when this scene's effect depends on effects a read removes — a look, a piece of business, a cut, a piece of timing that comes from performance. An excellent scene can be a bad demonstration. Choosing which sample to send is a decision of its own, and a read is one input into it, never the decision itself.

Narrow the claim when the scene does something real that isn't what the pitch says. If the read comes back with leverage and no warmth, your scene may be working and your sentence overstating. "Affection and leverage in the same breath" is a large promise. If the material only carries the leverage, the honest pitch says so, and it's a better pitch for it.

Investigate a specific condition when the confusion has an obvious suspect you haven't ruled out yet — the missing action lines, the reader doing two parts, the listener who arrived midway. Then run the read again rather than touching the script.

What the exercise can't establish

A table read removes camera, blocking, editing, physical performance, sound, and the accumulated familiarity of an audience that has watched four episodes. It cannot tell you whether the finished show lands. It cannot measure demand, and it cannot tell you whether a joke is funny in the way it will be funny with a face behind it.

It also can't validate your series. The most a clean read earns you is a modest, defensible sentence: in this stripped condition, with these people and this context, a listener could follow the turn and hear the fondness in it. That is a real thing to know and a small thing to claim, and keeping the claim small is what makes it usable in a room where someone will ask how you know.

Nothing in the method above depends on a published account of somebody else's development process. If you go looking for those, they're worth reading for one purpose: noticing how much of what a professional read surfaces sits in staging, casting and pace rather than in lines on a page. Those rooms have actors, a director and rehearsal. Yours has a kitchen table. Check the conditions before you borrow the conclusion.

One sentence to end on

After the read, write a single sentence that ties a recorded observation to an action, and let that be the whole output. Something in this shape:

Because [listener] could not say which hours Owen had accepted, and [listener] heard Nadia's praise as the setup of a trap rather than warmth, the scene changes: Owen names the shift he's agreeing to, and Nadia's line about the room keeps its fondness without the maneuver stacked on top of it.

Four exits are available to you instead, and the sentence has to be honest about which one you're taking. Change the scene, change the sample, narrow the claim — or admit that you haven't isolated the cause and say which condition you're testing next. One decision, one piece of evidence, one document revised. Then stop, because a read that produces five revisions has produced none you can defend.

Frequently asked questions

Why can a dialogue-driven scene look solid on the page but fail in a table read?

A page can be reread, but a listener gets one pass and no margin. If they lose track of who owes what to whom, they lose the scene. A table read is a cheap way to find that difference before someone with a development budget finds it for you.

What question should a table read ask instead of 'Is it funny?'

Ask one bounded question about the obligation the dialogue must make audible, such as: by the last line, what has Owen agreed to do, and when? Laughter is poor evidence on its own because a room can laugh at a line it did not understand, in relief, or out of politeness.

How do I tell whether confusion comes from the scene or from the reading conditions?

Walk through competing explanations: unclear intention in the text, missing action, delivery, unfamiliarity, omitted visual cue, and room conditions. Then test whether the confusion survives a change of conditions; if the same question goes unanswered with different readers, the text is the likelier culprit.

What decisions can follow a table read?

There are four exits: revise the scene when listeners track the obligation but not the relationship; change the sample when the scene depends on effects a read removes; narrow the claim when the scene does something real that is not what the pitch says; or investigate a specific condition and rerun.

What can a table read not prove?

It removes camera, blocking, editing, physical performance, sound, and the accumulated familiarity of an audience. It cannot tell you whether the finished show lands, cannot measure demand, and cannot tell whether a joke will be funny with a face behind it. A clean read earns only a modest, defensible sentence.

Further reading

• What Should a Film Proof of Concept Actually Prove? — If you're unsure whether a read is the right test, this frames the cheaper checks first and how to choose the smallest object containing your doubt.
• Your Sizzle Makes a Different Show From Your Written Pitch. Which One Changes? — For the case where a recording or sizzle is more convincing than the written pitch, this asks which document actually changes and whether the moment was reproducible.