A Reel is the hardest format to build from a website because a page is read and a Reel is watched. You cannot paste text into a video and expect it to hold. What you can do is mine your site for the idea, then script that idea as motion, voice and on-screen text. This guide is about writing Reel scripts from your pages, for people who are not video creators and do not want to be.
For static formats, see Instagram content from your website and the carousel workflow . For the same job on other video platforms, the TikTok script generator workflow and YouTube Shorts from website content cover the format differences.
The first 3 seconds are the whole game
On Reels, the algorithm decides almost everything by 3-second retention: did the viewer stay past the first three seconds. So the hook is not the caption, it is the first thing said and the first thing shown, at the same time. A weak open (“Hi everyone, today I want to talk about…”) loses half the audience before the topic arrives.
Pull the hook from the sharpest claim on your page and say it as the very first line, while the most interesting visual is already on screen:
- “This pricing mistake cost us 40% of trials.” (over: a screen recording of the old pricing page)
- “Everyone exports data wrong. Here’s the 4-click way.” (over: hands on a keyboard, the action already happening)
- “Three things our FAQ should’ve said louder.” (over: founder talking, fast cut)
The beat-by-beat script
A 20-to-30 second Reel has a tight shape. Write it as beats, with a spoken line, on-screen text, and a shot for each.
| Time | Beat | Spoken line | On-screen text | Shot |
|---|---|---|---|---|
| 0-3s | Hook | The sharpest claim | Big, 4-5 words | Action or face already moving |
| 3-8s | Problem | Why it matters | Short reinforce | Cut to the thing |
| 8-20s | Payoff | The 1-2 steps or the reveal | Step labels | Demo / point to camera |
| 20-25s | Proof | One number or result | The number | Screen or quote |
| 25-30s | CTA | “Save this” / “follow for more” | The ask | Face, direct |
Notice there is no slow intro. The page content lives in the 8-20 second payoff. Everything before it exists to earn those seconds, everything after it banks the result.
Source pages and the Reel they become
| Page | Reel idea | Format |
|---|---|---|
| How-to / docs | “Do X in 3 steps” | Screen recording + voiceover |
| Objection on a sales page | “You’re worried about Y. Here’s the honest answer.” | Founder to camera |
| Product page | “Watch this actually work” | Demo / screen capture |
| Case study | “How [customer] fixed Z” | Talking head + on-screen numbers |
| Customer review | React to a real quote on screen | Founder reading + reacting |
A founder talking to camera for 20 seconds about one objection will usually beat a polished graphic, because Reels reward a real person over a designed slide. You do not need a studio. You need one good line off a page and decent light.
On-screen text is non-negotiable
Most Reels are watched on mute. If the hook is only spoken, the muted viewer never gets it. Burn the hook into the first frame as text and reinforce the key step with captions throughout. This is also where the page’s exact wording can return: a precise step label, a real number, a customer’s words. Treat on-screen text as a second, silent script running in parallel.
The CTA and the link reality
Like the rest of Instagram, a Reel caption cannot link, so the CTA is “save,” “follow,” “comment a keyword,” or “link in bio.” Saves matter most here because a saved Reel signals the algorithm that the content was useful, not just entertaining. Pick one ask, say it out loud in the last beat and put it on screen. Anchor it to a deliberate social media CTA strategy rather than ending on a vague “follow for more.”
Measure watch-through, then rebuild what held
The signal that matters is average watch time / completion, then saves and shares. A Reel watched to the end twice gets pushed; a Reel abandoned at second four does not. If retention collapses early, the hook is the problem, not the topic. When a script holds attention, reuse its structure across other pages and feed the winner into your wider content repurposing workflow .
Utin is being built to draft these scripts from your site: it reads a page, proposes a 3-second hook, lays out the beat-by-beat script with on-screen text and shot notes, and keeps the source attached so claims can be checked before filming. You can register interest for the early pilot.