How to format a script for a teleprompter.

A document is built for an eye that can stop, go back, and skim. A prompter script is built for an eye that has to stay near one spot while the mouth keeps going. Those are different jobs, and the formatting that serves one quietly sabotages the other.

Why a document reads badly off a prompter

Three things change the moment text leaves the page and goes on a prompter.

You cannot see ahead. On paper you see the whole paragraph at once, so you know a sentence is long before you start it and you plan your breath. On a prompter you see a few lines at a large size. A long sentence ambushes you halfway through.

Your eyes are on camera. Every sweep across a wide line and every jump back to find your place is visible. The layout decides how much your eyes move, which is why formatting is also an eye-contact problem.

Punctuation stops doing its job. In a document a comma is a gentle hint. Spoken under pressure, commas become the only places you breathe, so your rhythm ends up shaped by the grammar instead of the meaning. A script for speaking has to show you where the meaning breaks, not just where the clauses do.

So formatting a prompter script is not decoration. It is deciding, in advance and on purpose, the things you would otherwise decide badly in the middle of a take.

Short lines, broken by meaning

The single most useful change: break the text into short lines, one idea or one phrase per line, and break where you would naturally pause, not where the margin happens to fall.

Compare a document sentence:

When we moved the onboarding call from week three to day one, which most of the team resisted at first, churn in the first month dropped noticeably.

With the same words, laid out to be spoken:

We moved the onboarding call
from week three to day one.
Most of the team resisted.
And churn in the first month dropped.

The second version is not just shorter on screen. The aside became its own sentence, the payoff got its own line, and every line ends where a breath belongs. You can glance down, take in the whole line, and look back up without travelling across it.

Some practical rules:

• Aim for lines you can take in at a glance. If you have to read a line left to right, it is too long.
• Never split a name, a number, or a phrase that has to be said in one breath across two lines.
• Leave a blank line between blocks of thought. The white space is where your eye lands when you look back.
• One sentence with two ideas becomes two sentences. If you stumble on it reading aloud, it is built for the page.

The eye-contact side of this, one idea per line and white space between blocks, is already covered in stop losing eye contact while reading a script, along with why a narrow column matters. What this post adds is the rewrite itself: turning a document sentence into lines you can speak.

Sections you can restart from

A script is not one long take, even when you mean to record it as one. You will fluff a line, and when you do, you want to go back to a clean starting point, not to the top.

Split the script into sections at the places where a retake would naturally begin: the hook, each main point, the call to action. Each section should open with a sentence that works as a fresh start, so you can cut between takes without a jump in logic. A section that starts with "And that is why..." is hard to restart from, because it depends on what came just before.

Keep sections short enough that redoing one is cheap. If a mistake near the end of a section costs you three minutes of re-reading, the section is too long.

Mark pauses and emphasis before you record

Pauses you leave undecided get decided mid-take, usually badly. The fix is to put your intentions on the page in a form you cannot confuse with the words you are meant to say.

That last part matters. A stage direction written as plain text, like "(pause)" or "smile here", is easy to read out loud by accident, which is a retake in itself. Directions need to look different from speech.

Mark three kinds of things:

• Pauses. After a point you want to land, before a number, before the turn in a story. Mark the ones that matter, not every full stop.
• Tone shifts. Where you want to move from explaining to warm, or from relaxed to serious. One word is enough to remind you.
• Emphasis. The one word per sentence that carries the point. Only one. If everything is emphasised, nothing is.

In Mirrento, directions go in double brackets, such as [[pause 2s]] or [[warm]], and show up on the prompter as small pills rather than as words, so they read as instructions at a glance. You can define your own cues too, so the vocabulary is yours rather than a fixed list. These are reminders for you, not timers: the pause still happens because you take it.

If you want the reasoning behind where pauses should go, how to stop sounding like you're reading covers that in depth: why pauses at meaning sound human and pauses at punctuation sound read. This post only deals with marking them on the page.

What to strip out

The last pass before recording is deletion. A document carries a lot that only makes sense on a page, and every piece of it is something to trip on.

• Headings and bullet points. Nobody says "Point two" out loud unless they mean to. Turn bullets into sentences, or into section breaks.
• Parentheses. A parenthetical is an aside the eye can skip. The mouth cannot skip it. Make it its own sentence or cut it.
• Abbreviations and symbols. Write numbers, units and acronyms the way you will say them. "€1.2k" on screen invites a hesitation; "twelve hundred euros" does not.
• Links, references and footnotes. They belong in the description, not on the prompter.
• Long quotes. Trim to the part you actually need, or paraphrase. Reading a block quote aloud is where most people sound most like they are reading.
• Notes to yourself in plain text. Anything that is not meant to be spoken either becomes a marked cue or leaves the script.

Then read the whole thing out loud once, at speaking pace. Every stumble is a line still built for the page. Fix those, and only those.

Where you do this editing is up to you. Mirrento keeps each script as a plain JSON file in a folder you choose, and if you edit one in another editor, the script list in Mirrento updates on its own, so the final pass can happen in whatever tool you write in. If that workflow sounds like yours, the 7-day trial is enough to format one real script this way and record it.