On this page
The template
Copy this template into your script document. Fill one block per section; every block has the three columns an AI production needs.
UGC SCRIPT – [brand] – [product] – [angle] – [length]s – [format]
Angle: [the one reason this ad should make someone buy]
Character: [who is speaking: age range, look, setting, tone]
Allowed claims: [list, word for word, from the brand]
1. HOOK ([seconds]s)
Voice line: [first sentence, said straight away]
On-screen text: [short line that repeats or sharpens the hook]
Shot / prompt: [framing, action, setting, light]
2. PROBLEM ([seconds]s)
Voice line: [the pain, in the speaker's own words]
On-screen text: [optional]
Shot / prompt: [the problem shown, not only said]
3. DISCOVERY ([seconds]s)
Voice line: [how the speaker came across the product]
On-screen text: [optional]
Shot / prompt: [first look at the product]
4. PROOF ([seconds]s)
Voice line: [the specific moment it worked for them]
On-screen text: [the proof point in a few words]
Shot / prompt: [the proof shown on screen]
5. PRODUCT ([seconds]s)
Voice line: [what it is and the features that matter to the angle]
On-screen text: [product name or key feature]
Shot / prompt: [close-ups: label, use, detail]
6. CTA ([seconds]s)
Voice line: [what to do next, plus the offer if there is one]
On-screen text: [offer or action]
Shot / prompt: [speaker to camera, product in frame]
The three columns let one script feed every tool in the workflow. The voice line goes to voice and lip sync, the on-screen text goes to the edit, and the shot description becomes the prompt.
How to fill each section
Each section has one job. If a line does not do that job, cut it.
| Section | Job | Write it like this | Avoid |
|---|---|---|---|
| Hook | Stop the scroll and set up the angle | Start mid-thought, with a claim, a question or an admission | Greetings, names, "hey guys" |
| Problem | Make the viewer recognise their own pain | A specific moment, not a general complaint | Listing several problems |
| Discovery | Make the product's arrival believable | One sentence on how they found it | A long backstory |
| Proof | Give a reason to believe | One concrete moment with a detail | Numbers or results the brand cannot support |
| Product | Show what it is and why it fits | Two or three features tied to the angle | Every feature on the product page |
| CTA | Tell the viewer what to do | One action, in plain words | Two competing actions |
Write the voice lines the way people talk. Short sentences, contractions and the odd unfinished thought sound more natural when an AI voice reads them.
Timing for 15, 30 and 45-second versions
Split the length across the sections like this. The numbers are seconds per section and add up to the full length.
| Section | 15-second | 30-second | 45-second |
|---|---|---|---|
| Hook | 2 | 3 | 3 |
| Problem | 2 | 5 | 7 |
| Discovery | 2 | 4 | 6 |
| Proof | 3 | 7 | 12 |
| Product | 4 | 7 | 11 |
| CTA | 2 | 4 | 6 |
| Total | 15 | 30 | 45 |
Write the 45-second version first. Cut it down to 30 by trimming discovery and the problem, then to 15 by keeping only the hook, one proof line, the product and the CTA.
Time each voice line by reading it aloud at the pace of the character. If it runs long, cut words before you speed up the voice.
A filled example, 30 seconds
This example is for a made-up foldable laptop stand. It shows how the three columns read when filled in.
| Section | Voice line | On-screen text | Shot / prompt description |
|---|---|---|---|
| Hook (3s) | "My neck stopped hurting the week I changed one thing on my desk." | One change, no more neck pain | Woman in her thirties, home office, rubs her neck, looks into a phone camera held at arm's length |
| Problem (5s) | "I work on a laptop all day, and by four I'm hunched over like a question mark." | Hunched by 4pm | Side view of her bent over a laptop on a kitchen table, late-afternoon light |
| Discovery (4s) | "A friend who works from home had this on her desk, so I tried it." | — | She unfolds a slim metal stand from a pouch, handheld close-up |
| Proof (7s) | "Screen at eye level, separate keyboard, and I got through Friday without stretching once." | Screen at eye level | Laptop raised on the stand, she types upright, camera at desk height |
| Product (7s) | "It folds flat, fits in my bag, and has six height settings." | Folds flat | Close-ups: folding, sliding into a bag, adjusting height |
| CTA (4s) | "If you're on a laptop all day, the link's below." | Link below | She sits back, stand in frame, points down |
The proof line uses a moment, not a medical claim. "Got through Friday without stretching" is a personal experience; "cures back pain" is a health claim that needs support from the brand.
Format versions of the template
The same six sections work in every format. What changes is who speaks, where the scene is set and how the shots are described.
Selfie UGC
One character speaks to a front camera in one location. Write handheld framing into every shot line.
SELFIE – [product] – [length]s
Character: [description]; location: [one room]; camera: front camera, arm's length
HOOK: [voice line] | [on-screen text] | handheld, mid-action, eye contact
PROBLEM: [voice line] | [text] | cutaway to the problem, same location
DISCOVERY: [voice line] | [text] | product enters frame in hand
PROOF: [voice line] | [text] | proof shown on camera
PRODUCT: [voice line] | [text] | close-ups held by the character
CTA: [voice line] | [text] | back to face, product in frame
See AI story-time ads for public examples of selfie storytelling.
Podcast, two speakers
Split the script between a storyteller and a questioner. The questioner's lines voice what a viewer would ask.
PODCAST – [product] – [length]s
Host A: [description, voice] Host B: [description, voice]
Set: [studio, mics, lighting]; shots: two-shot, A single, B single
HOOK: A: [admission or claim] | [text] | A single
PROBLEM: B: [question] / A: [the problem] | [text] | two-shot
DISCOVERY: A: [how they found it] | [text] | A single
PROOF: B: [follow-up question] / A: [proof] | [text] | A single, B reacts
PRODUCT: A: [features] | [text] | product B-roll
CTA: B: [recommendation and action] | [text] | B single
See AI podcast ads.
Drama, scene list
Turn each section into a scene. Dialogue is short; the problem and proof are shown, not explained.
DRAMA – [product] – [length]s
Cast: [character 1], [character 2] (reference image for each)
Subtitles: burned in
SCENE 1 – HOOK: [location] – [action that shows tension] – [line or text]
SCENE 2 – PROBLEM: [location] – [problem happens] – [line]
SCENE 3 – DISCOVERY: [location] – [character 2 introduces product] – [line]
SCENE 4 – PROOF: [location] – [the turn: problem solved on screen] – [line]
SCENE 5 – PRODUCT: [insert shots of product in use]
SCENE 6 – CTA: [end card] – [offer text]
See AI drama ads.
Animation, narrator and scenes
A narrator carries the voice lines while scenes show the action. Animation can show proof visually in ways live action cannot, such as time passing in a few frames.
ANIMATION – [product] – [length]s
Style: [3D / clay / paper / other]; character: [design, reference image]
Narrator: [voice, tone]; product: keep real shape, colour and label
HOOK: Narrator: [line] | Scene: [character, exaggerated reaction]
PROBLEM: Narrator: [line] | Scene: [problem shown as a visual gag]
DISCOVERY: Narrator: [line] | Scene: [product appears]
PROOF: Narrator: [line] | Scene: [visual proof, e.g. time passing]
PRODUCT: Narrator: [line] | Scene: [hero shot of product]
CTA: Narrator: [line] | Scene: [character points to end card]
See Pixar style AI ads and AI claymation and craft-style ads.
Street interview
A presenter asks a passer-by a question, and the passer-by's answer carries the script. Write the presenter's question so it sets up the hook.
STREET INTERVIEW – [product] – [length]s
Presenter: [description, handheld mic]; passer-by: [description]
Location: [street type, time of day]; camera: handheld, shoulder height
HOOK: Presenter: [question that sets up the angle]
PROBLEM: Passer-by: [answer that names the problem]
DISCOVERY: Presenter: [follow-up] / Passer-by: [how they found it]
PROOF: Passer-by: [specific moment it worked]
PRODUCT: Passer-by: [shows product from bag]; insert close-ups
CTA: Presenter: [line to camera with action]
See AI street interview ads. Characters in this format must be generated people, not lookalikes of real ones.
Compliance notes for claims
Check every script against the brand's approved claims before it goes to production. AI makes risky visuals easy to generate, so the script is where you stop them.
- Claims. Use only claims on the brand's approved list, word for word. Personal experience lines ("I got through Friday") are safer than outcome claims ("fixes back pain").
- Numbers. Do not add figures, percentages or timeframes the brand has not supplied.
- Experts. Do not cast an AI character as a doctor, scientist or other authority unless the brand has approved it. TikTok's help center says it does not allow AI-generated content showing fake authoritative sources, even when labelled.
- Real people. Do not write scripts for characters that resemble public figures or real private people. TikTok's same page lists public figures shown making an endorsement among content it does not allow.
- AI labels. TikTok's advertising policy allows AI-generated content in ads only if it is labelled. AI-generated content labels on Meta and TikTok ads covers both platforms.
- Before-and-after. Avoid generated before-and-after shots unless the brand supplies real ones and has approved their use.
Add a "claims checked by" line to every script so someone signs off before production starts.
Common mistakes
Most weak UGC scripts fail in the same few ways. Check yours against this list.
- The hook is a greeting or the product name.
- The problem is general ("I was always tired") instead of a moment.
- The proof is a superlative ("the best stand ever") instead of something that happened.
- The product section lists every feature instead of the ones that fit the angle.
- The CTA asks for two things.
- The shot column says what to show only in the voice line, so the AI shot does not match.
From one script to many versions
A script written with this template is ready to turn into every format. Keep the angle, proof and claims fixed, and change the hook, setting and speakers.
One winning script, made in every ad format walks through that method beat by beat. Ad hooks: examples and patterns gives more ways to open the same script.
If you already make full AI ads that perform, Anglr provides winning scripts and angles and pays on performance for every version that runs. AI video editor jobs, paid on performance explains how to apply.
Who this is for
- Independent AI creators and editors. You make the ads yourself. You know your work performs, so you'd rather be paid on what it does than on a flat fee.
- AI agencies and studios. You run a team of editors. You write the brief, your team makes the versions, and you're paid on performance because you know your team's ads scale.
Not for you if
- You only want a flat fee.
- You're new to editing and not yet confident your ads will perform.
- You want every shot spelled out for you.
- You make one style and won't try others.
FAQ
What should a UGC script include?
A hook, the problem, how the speaker found the product, proof, the product itself and a call to action. For AI production, add a shot or prompt description and on-screen text for every line.
How long should a UGC script be?
Write to the ad length. Read each voice line aloud with a timer and cut until the full script fits, leaving a short pause after the hook. The timing table on this page splits 15, 30 and 45 seconds by section.
Can I use one UGC script for 15, 30 and 45-second ads?
Yes. Write the 45-second version first, then cut discovery and secondary proof for the shorter versions. Keep the hook, the main proof line and the call to action in every cut.
Can I reuse a UGC script across creators and formats?
Yes, as long as the angle, proof and claims stay the same. One winning script, made in every ad format shows how to adapt one script to podcast, drama, animation and brand film.
What is the most common UGC script mistake?
A slow opening. If the first line is a greeting or a name, cut it and start with the hook.
Do AI UGC scripts need different compliance checks?
They need the same claim checks as any ad, plus checks on AI labels and likeness. Never write a script for an AI character presented as a real doctor, expert or public figure.
Sources
- TikTok Help Center – About AI-generated content
- TikTok Ads – Advertising policies: Misleading and false content
Checked: 12 Oct 2026