Teleprompter for Sales and Outreach Videos: How to Read a Script That Still Sounds Personal

A sales video has one job that a tutorial or a Reel does not: the person watching has to believe it was made for them. That is also exactly what a teleprompter threatens. Read a prospecting script word for word and the viewer can tell within three seconds that they are number forty-one on a list. Skip the script and you ramble, forget the one number that mattered, and record it four times. The craft is in the middle: a script that holds the structure and the facts, and a delivery that leaves room to be a person.

This post is the practical version of that balance for founders, account executives, recruiters, freelancers and anyone who sends short videos to specific people. It assumes you are recording on an iPhone, probably in a hurry, probably more than one in a sitting. Where a teleprompter app helps, we say how; where it hurts, we say that too.

Why outreach videos fail (it is not the camera)

Most bad outreach videos share three faults. They open with the sender instead of the recipient ("Hi, I'm Priya from..."). They are too long, because without a script people pad. And they look away from the lens at the key moment, because that is when the sender glances at notes. A teleprompter that sits beside the front camera fixes the third problem outright, and a good script fixes the first two. What it cannot fix is a video that says nothing specific to the recipient; no amount of eye contact rescues a generic pitch.

The four-part script that stays personal

Write the script as four short blocks and mark which ones are fixed and which ones change per person. The fixed parts go on the prompter verbatim. The variable parts go on the prompter as a prompt in brackets, which you fill in live.

• The hook, variable (5 to 8 seconds). One sentence about them, not you: something you noticed on their site, their hiring page, their last post, their product. On the prompter it reads as [Their thing: what I noticed]. You say it in your own words, looking at the lens, because you just read it a moment ago and it is fresh.
• The bridge, fixed (5 seconds). One sentence connecting their thing to what you do. This can be word-for-word; it is the same for everyone and it needs to be tight.
• The value, fixed (15 to 25 seconds). The one claim, the one number, the one example. This is the part people get wrong unscripted, so script it fully and read it. It is also where a prompter earns its keep: numbers and names are exactly what you forget under mild pressure.
• The ask, variable (5 to 8 seconds). A specific next step with a specific time. On the prompter: [Ask: 15 min Thursday?]. Say it to them, not to the script.

That structure lands at roughly 40 to 50 seconds spoken, which is the right length for a first touch. If you want the arithmetic on words per second, our post on the ideal script length for a 60-second video has it; the short version is about 110 to 130 words.

Setting up the prompter so it reads as a conversation

In Teleprompter: Camera Overlay the script floats over the camera preview next to the front lens, so your eyes stay close enough to the camera that the viewer reads it as eye contact. Three settings matter for outreach specifically:

• Use voice-follow, not auto-scroll. Voice-follow scrolls as you speak, using on-device speech recognition, and finds your place again if you stumble. That is what lets you leave the script for the variable parts: when you ad-lib the hook, the text waits; when you return to the fixed bridge, it picks you up. Auto-scroll at a fixed speed will run away from you during the personal bits. We compared the two modes in voice-follow vs auto-scroll.
• Large text, short lines. Set the text size up so a line holds five or six words. Your eyes move less, which keeps the gaze steady, and the brackets for variable parts are easier to spot.
• Put the bracketed prompts in capitals. [THEIR THING] stands out in a wall of text and tells your brain: stop reading, start talking.

The script never appears in the saved video, and recording is portrait 4K, which matters when the video is going to be watched on a phone inside an email or a LinkedIn message.

Recording ten personalised videos in a row without sounding tired

Outreach is a volume game, so you will batch. Do the research first, not between takes: for each person write the hook line and the ask line into a copy of the script before you sit down. Teleprompter: Camera Overlay keeps unlimited scripts, so one script per prospect, named after them, is the easiest way to avoid saying the wrong company name on camera, which is the single most embarrassing outreach mistake and more common than you would think.

• Record in one setting, one outfit, one light. Consistency reads as reliability.
• Do the fixed value block exactly the same every time; vary only what is meant to vary. Trying to improvise the core claim ten times produces nine worse versions.
• Two takes per person maximum. The second is usually the better one; the fifth never is.
• Stand up between every three videos and re-read the next hook out loud once off camera. Thirty seconds of silent read-through removes most first-take stumbles.
• Stop when your energy drops. A flat delivery in the hook is worse than sending the video tomorrow.

The longer version of this workflow is in batch-recording a week of content in one sitting; the same rules apply, with the added twist that every video here has a different name in it.

When not to use a teleprompter for outreach

Two honest exceptions. First, replies. If someone has written back and you are answering a specific question, do not script it; your reply should sound like a person thinking, and a prompter makes it sound like a person reading. Jot three words on a sticky note and talk. Second, very senior or very small audiences. If you are sending one video to one CEO, the 45-second structure still applies, but write it as bullet prompts only, not sentences, so nothing about it feels produced. The prompter is for the twentieth video of the day, when your structure would otherwise collapse; it is not for the one video that only ever needed to be good once.

A related trade-off: voice-follow needs to hear you clearly. In an open office or a car it may lag on the ad-libbed parts. In that case switch to auto-scroll for the fixed blocks and pause it manually, or record somewhere quieter.

Compliance, claims and the words you cannot improvise

If you sell anything regulated (financial products, health services, anything with legally required wording), the fixed value block is where the approved language lives, and the prompter is how you guarantee you say it the same way every time. Keep the compliance sentence as its own paragraph on the script, read it exactly, and do not bracket it. This is one of the few situations where reading verbatim is the point rather than the problem.

A one-minute checklist before you hit record

• Correct prospect's script loaded, their name and company checked.
• Hook and ask lines rewritten for them, in brackets, in capitals.
• Voice-follow on, text size large, phone at eye level, front lens next to the first line of text.
• Value block is word-for-word and under 25 seconds when read aloud.
• One specific time proposed in the ask.
• Quiet enough for speech recognition to keep up.

Teleprompter: Camera Overlay is free on the App Store with occasional ads, a small one-time purchase removes them, there is no subscription and no watermark, and everything, including the speech recognition behind voice-follow, runs on the phone and works offline. It is iPhone-only. Next in this series: overcoming camera anxiety, for the people who have the script, the setup and the plan, and still freeze when the red light comes on.