How to Write a Reel Script: A Guide by Fraggell Productions

Most advice on writing reel scripts tells you to “grab attention” and “add value” without ever showing you how. This guide is different. At Fraggell we write scripts for 700+ paid-social ads every month for 7 and 8 figure DTC brands, and over $450M has been spent running the ads those scripts became. The same mechanics that make a paid ad convert make an organic reel get watched: a hook that earns the first three seconds, a body that keeps people from swiping, and an ending that tells them exactly what to do next. Below is the actual process our writers use daily, including the framework and a worked example you can copy. Whether you are scripting an organic reel for your brand account or a UGC ad you plan to put spend behind, the craft is identical.

Why do the first three seconds decide everything?

The single most important metric in short-form video is what we call hook rate: the number of 3-second views divided by impressions. If your hook rate is low, nothing else in the script matters, because nobody saw it. On paid ads we consider anything under about 25% a rewrite, and the best hooks we ship clear 40%. The mechanism is simple. Meta and Instagram show your video to a test audience, watch how many people stop scrolling, and distribute accordingly. A weak first line does not just lose those viewers, it tells the algorithm to stop showing the video. That is why we write the hook last, after we know exactly what the script argues, and why we write five to ten hook options for every one script. When we rebuilt ALOHA’s creative around stronger hooks and customer-evidence scripts, their CAC dropped 40% across three rounds of testing. The script did not get longer or fancier. The first three seconds got sharper.

If you want to understand how view counts actually translate into reach, we broke that down in how many views it takes to go viral.

Where do good script ideas come from?

Not from a trends folder. The scripts that perform come from customer evidence: the words real buyers use when they describe the problem your product solves. Before we write a line, we pull three sources. First, reviews — yours and your competitors’. Sort by critical reviews on the competitor and you will find the exact frustration your product should open with. Second, community language — Reddit threads, Facebook groups, TikTok comments, anywhere your customer complains in their own words. A hook lifted nearly verbatim from a one-star review of a rival product will outperform anything a copywriter invents at a desk. Third, your ad account — if you are already running ads, your best-performing hooks, angles, and comments are a map of what your audience responds to. One idea per script is the rule that follows from this. A script that argues “hydration and energy and skin and sleep” argues nothing. Pick the single sharpest piece of evidence and build the whole script on it.

What is the script framework?

Every script we ship follows the same four-part spine. The labels matter less than the jobs each part does.

1. Hook (0-3 seconds). One line, spoken and shown on screen, that names the viewer’s problem or makes a claim they cannot ignore. Write it as the viewer’s own thought, not the brand’s announcement.

2. Problem and agitate (3-15 seconds). Prove you understand the problem better than they do. Describe the specific, felt experience of it — the 2pm crash, the third snooze, the shirt that never fits right — and the cost of leaving it unsolved.

3. Demo and proof (15-40 seconds). Show the product doing the thing. Then stack one or two pieces of proof: a review on screen, a before-and-after, a stat, a founder explaining the mechanism. Show, then substantiate.

4. CTA (final 3-5 seconds). One instruction, tied to the one idea. “Tap the link and try it for 30 days” beats “follow for more, like, comment, and check the link.”

What does a finished script look like?

Here is the framework applied to a hypothetical electrolyte drink, built from a real pattern we use constantly. Hook options first, because you always write several:

• “I stopped drinking plain water on my runs and my afternoon headaches disappeared.”
• “If you drink 2+ litres of water a day and still feel tired, watch this.”
• “My trainer told me I wasn’t dehydrated — I was under-mineralised.”

Then the body. Problem/agitate: “I was doing everything right. Two litres a day, cutting caffeine, going to bed early. And I still hit a wall at 2pm, every single day. Turns out when you sweat, you’re not just losing water — you’re losing sodium, potassium, magnesium. Plain water just dilutes what’s left.” Demo/proof: Talent tears a sachet into a glass, stirs, drinks on camera. “One of these in my first glass of water every morning. No sugar, actually tastes good. I’ve done it for six weeks and the 2pm crash is just — gone.” Cut to a screenshot of a five-star review saying the same thing in a customer’s words. CTA: “There’s a starter pack linked below. If it doesn’t fix your afternoons, they’ll refund it. Try it.” That full script runs about 40 seconds spoken. One idea — minerals, not water — carried from hook to CTA.

How do you keep people watching past the hook?

The metric here is hold rate: the share of people who hooked in and are still watching at the midpoint or through-play. A great hook with a saggy body produces a video that spikes and dies, and you will see it in the retention graph as a cliff at second four. The fix is almost always cutting, not adding. Read your script aloud and delete every sentence that does not either advance the one idea or add proof. Kill throat-clearing (“so basically what happened was”), kill the second example when one lands, and kill any sentence the viewer could predict. Every 5-7 seconds, something should change — a new visual, a new claim, an on-screen text beat — because the viewer is deciding whether to keep watching on that same rhythm. When we rebuilt Humantra’s script and hook system this way, the account produced a 72% higher click-through rate across 600+ ads. You can see how that engagement work compounded in the Humantra case study.

Does the same script work across different formats?

Yes, and this is the part most creators miss: a script is an argument, and the same argument can ship as five different assets. We take one validated script and produce it as UGC (a creator delivering it to camera), a founder ad (the founder telling the origin version of the same idea), a studio ad (polished b-roll with voiceover), statics (the hook line as a headline over a product image), and mashups (the best 2-3 second beats from several videos recut into one). This matters because platforms reward variety and audiences fatigue fast. It also makes testing honest — when the same script wins as UGC and loses as studio, you have learned something about the format, not the idea. Write once, ship five ways, and let the account tell you where the idea lives best. You can see the range of formats on our work page.

How do you know if your script worked?

Judge the script with numbers, in order. Hook rate first: 3-second views divided by impressions tells you whether the opening line earned attention. Hold rate second: retention through the body tells you whether the argument kept it. Then click-through rate tells you whether the CTA and the idea created intent, and — on paid — cost per acquisition tells you whether the whole thing made money. The order matters because each metric diagnoses a different part of the script. Low hook rate: rewrite the first line, keep the body. Good hook, bad hold: the body is bloated or the hook wrote a cheque the video does not cash. Good watch time, no clicks: the CTA is vague or the idea does not connect to buying. Working this loop across hundreds of scripts a month is how our clients see average CPA drop around 30%. Fraggell’s founder Fraser Cottrell teaches this exact process weekly on YouTube — including a recent breakdown called “How I Write Winning Meta Ad Scripts (without AI)” — if you want to watch it done live.

Where should you start today?

Do not start with a blank page. Start with twenty minutes in your reviews and your competitors’ one-star reviews, pull the three most emotionally loaded customer phrases you find, and write five hooks from them. Pick the strongest, build the four-part spine underneath it, read it aloud, and cut a third of the words. Film it simply — a phone and decent light beat a production truck when the script is right. Then post it, check hook rate and retention honestly, and rewrite the weakest section rather than starting over. Scripts are not written so much as iterated, and the writers who improve fastest are the ones who let the retention graph tell them where they lost the room. One idea, one piece of customer evidence, one instruction at the end. That is the whole craft, and it is learnable.

If you would rather have this done for you, script-to-screen creative for paid social is what Fraggell does all day — 700+ ads a month for DTC brands that need creative to hit a CPA target, not just look good. Have a look at the work and see if the approach fits how you think about creative.