The Teleprompter Removed the Last Excuse.

Instagram brought the teleprompter from its Edits app into the main Instagram camera, so a script scrolls under the lens while you record.

Adam Mosseri’s framing was simple: add a script that scrolls while you film, and stay on message without doing a ton of takes. The text sits below the front-facing camera and the speed is adjustable.

That removes the last practical objection to a scripted piece to camera. It does not touch the reason most brand video goes out unscripted, which is that nobody in the team actually owns writing the script. A tool that displays words cannot supply them.

What the teleprompter does

The facts are narrow and worth stating plainly.

• It sits in the main Instagram camera, having started life in the Edits app.
• You load a script and it scrolls while you record.
• The text renders below the front-facing camera, so your eyeline stays close to the lens.
• The scroll speed is adjustable, so the script follows your pace rather than the reverse.
• Instagram’s stated reason for building it was that creators found it hard to read scripts off their notes while filming.

What has not been spelled out is rollout. There is no published detail on phases, markets or account types, so treat it as a capability to check for on each device your team films on, not one you can assume everyone has.

The excuse it removes

Every team that has tried scripted pieces to camera knows the failure. Someone props a second phone sideways next to the tripod, the presenter’s eyes track left, and the finished video looks exactly like a person reading off a second phone propped sideways.

So the team quietly stops scripting. Bullet points on a hand, a rough idea of the argument, and eleven takes to get one usable minute.

Text under the lens closes the eyeline gap. The presenter looks like a presenter, and the number of takes falls because nobody is trying to remember paragraph three. That is a real gain in filming time, and it is the whole of what the feature is for.

The reason brand video goes out unscripted

It was almost never a hardware problem. It is an ownership problem.

Look at what already has an owner in a typical brand video: the shot list, the location, the edit, the cover frame, the caption, the publish slot. Every one of those is somebody’s job with a date attached.

The script is the exception. It is assumed to arrive with the person on camera, and that person is usually the founder or the marketing lead, which is to say the busiest person in the building. They write it in the ten minutes before filming, or they do not write it at all.

The result is a video with no argument in it. No edit rescues that, and no teleprompter does either.

A script is a production artefact

Treat the script the way you already treat a thumbnail or a caption. It is a named deliverable with an owner, a due date, a format and a length, and it exists in the plan before anyone books a filming slot.

That means the script is due before the camera comes out, not on the morning of the shoot. A filming session with no approved script should be moved, in the same way you would move a shoot with no location.

Length is decided by the format, not by whoever is writing. If the output is a Reel, the script is written to the length the format actually rewards, and the same judgement that decides how long an Instagram Reel should be decides the word count before the first sentence is typed.

Who writes it

In a small team, the honest answer is one named person per format, not a committee and not the presenter by default.

1. The person who owns the idea writes the first draft. They already know the point, so they are the cheapest writer in the building.
2. The person on camera edits for their own mouth. They may change wording, phrasing and rhythm freely. They may not change the argument, because the argument was signed off.
3. One reviewer approves, and only one. Two reviewers on a ninety second script produces a fortnight of comments.
4. If the presenter is also the founder, someone else writes the draft. A founder editing a draft takes ten minutes. A founder writing from nothing takes a week of good intentions.

The rule that matters most is the second one. A script the presenter cannot say in their own voice will be read, not performed, and the teleprompter makes reading very easy to hide behind.

How long the script should be

Count words, not minutes. NBK plans a piece to camera at roughly two to two and a half words a second, which is a comfortable pace for someone who is not rushing.

• A fifteen second hook piece: around 35 to 40 words. One idea, one line of proof.
• A thirty second explainer: around 70 to 80 words. A claim, a reason, a consequence.
• A sixty second piece: around 140 to 160 words. Three beats, and the third one has to earn its place.
• Anything past two minutes: write it as a section list first, because prose that long always overruns.

Write to the low end. Speech expands when someone is nervous, and every script that runs long gets fixed by speeding up the delivery, which is the fastest way to sound like a hostage video.

Write for the ear, not for the page

Written prose read aloud sounds written. The teleprompter is faithful, so whatever you wrote is what the audience hears.

• Short sentences. If you need a comma to hold it together, cut it in two.
• One clause per breath. Mark the breaks in the script itself with line breaks, because the scroll respects them.
• Say the number, do not write it in figures and hope. “Around a third” beats “34.7%” out loud.
• No brackets, no asides, no “which, incidentally”. Spoken language has no parentheses.
• Open on the sentence you would actually say to someone. Not “in this video we look at”, just the point.
• Read the whole thing aloud once before it goes near approval. Every unsayable sentence announces itself in the first pass.

Approve the script, not the edit

The cheapest place to change a video is the document. The most expensive is the edit, and the worst is after publishing.

Most approval pain in video comes from reviewing the wrong artefact. A stakeholder sees the cut, realises the claim is wrong, and now you are reshooting a person, a location and a lighting setup to fix a sentence. Approve the words while they are still just words, and the edit becomes a technical review rather than an editorial one.

That is also what makes scripted video schedulable. When the script is signed off in advance, a filming block produces four finished pieces instead of one, which is the whole economics of running video production as a repeatable step rather than an event.

What a teleprompter makes easy to get wrong

A scrolling script rewards the presenter for keeping up with it. That is the trap.

The signs are consistent and easy to spot in a review:

• Flat, even emphasis, with every sentence landing at the same pitch.
• Eyes tracking horizontally at a steady rate rather than moving with the thought.
• No pauses anywhere, because the text never stops moving.
• A last line delivered while still reading, so the video ends on a downward inflection instead of a point.

None of that is a scripting failure. It is a delivery failure that unscripted filming used to hide, because a person improvising at least sounds like a person thinking.

How to keep it sounding spoken

Small mechanical fixes, all of which take less time than another take.

1. Set the scroll slower than feels right. You can always pause, you cannot un-rush.
2. Learn the first line and the last line by heart. Open and close looking straight down the lens with nothing scrolling.
3. Record in chunks of two or three sentences and let the edit join them. Nobody is filming a monologue in one pass.
4. Put a deliberate pause in the script as its own line, so the scroll gives you the beat.
5. Deliver it once to a person before you deliver it to a camera, even over a call. Emphasis you find talking to a human survives into the take.

If you are unsure whether a scripted version genuinely outperforms your usual unscripted one, run the comparison properly and test the Reel before you publish it rather than deciding from the edit suite.

What good looks like

A brand with scripting as an owned step looks noticeably boring from the inside, which is the point.

Scripts for the month’s pieces to camera exist by a fixed day, written by a named person, at the word count the format takes. One reviewer signs them off in a single pass, with comments on the argument rather than on commas.

The filming block is booked against approved scripts only, and it produces several finished pieces because nobody is inventing the point between takes. The presenter has read every script aloud once, so the wording is already in their mouth before the scroll starts.

The teleprompter, in that setup, is what it should be. A small piece of production kit that saves takes, not a strategy.

How NBK thinks about scripting

NBK treats social as an operating system rather than a content problem, and scripting is a clean case of the difference. The capability is free and sitting in the app. The output is still weak, because the missing part was never the tool.

Every input to a published video should have a name, an owner, a deadline and a definition of done. Captions and covers usually have all four. Scripts almost never do, and that single gap explains most of the distance between the video a brand meant to make and the one it published.

Fix the ownership and the tool becomes useful immediately. Skip it and you have bought a faster way to record an unfocused thought.

Next step

If your team is filming regularly and the finished pieces still feel unfocused, the constraint is almost certainly upstream of the camera. NBK can help find where the process is losing the idea, and rebuild the step that is missing.