Ad Video Script: The 3-Act Structure That Converts on TikTok and Reels

The 3-act narrative structure applied to short-form video advertising. Discover how to write an ad video script that converts on TikTok Ads and Instagram Reels: with templates, examples and the mistakes to avoid.

SociaLover Team · Updated · 15 min read

A TikTok ad script that converts has three acts in 15 to 30 seconds: a hook that names a pain in 3 seconds, a demonstration that shows the transformation, and one call to action. Below: the timing per act, a fill-in template, five complete scripts by industry and the mistakes that kill a script.

Why narrative structure converts

An ad with no narrative structure gets processed as noise: there is nothing to resolve, so there is no reason to stay. An ad with a clear structure (even in 20 seconds) opens a question in Act 1 and pays it off in Act 3, which is what carries the viewer all the way to the CTA. The window to open that question is short: about 71% of viewers decide within the first 3 seconds whether they keep watching, according to Stratboost (2026). The video ads that hold attention to the end tend to share one trait: they tell a mini-story instead of listing features.

The three acts explained

The three acts split the runtime 20 / 55 / 25: the hook and the problem in the first fifth, the solution and its proof in the middle, one call to action at the end.

1. Act 1: Hook and problem
0 to 6 s, 20%
2. Act 2: Solution and proof
6 to 22 s, 55%
3. Act 3: CTA
22 to 30 s, 25%

The three acts mapped onto a 30-second format. Act 1 has to be over by the 6-second mark, Act 2 owns the 16 seconds in the middle, and the CTA gets the last 8. When a script runs long, cut inside Act 2, never from the hook or the CTA.

Act 10-20% of the runtime

Hook and problem

Grab attention by naming a problem the audience feels. No brand intro, no logo. Go straight to the problem.

Act 220-75% of the runtime

Solution and proof

Show your solution in action. Show, don't tell. Use numbers and social proof. The product should visibly solve the problem.

Act 375-100% of the runtime

CTA and urgency

One single action, asked clearly. Add a touch of urgency or a guarantee to reduce friction. Finish strong. The final shot is the one people remember most.

Three acts, five beats
Act 1
Hook
Act 1
Problem named
Act 2
Solution shown
Act 2
Proof
Act 3
One CTA
Three acts, but five beats to write, and that is why Act 2 is the longest: it carries both the demonstration and the proof, which are two different jobs. Act 1 has two beats to fit into the first 20% of the runtime, and Act 3 has exactly one. Reading a script against this chain is the fastest way to spot what is missing: most weak ads have four beats and skip the proof.
Which template for which product: the three structures detailed below, side by side.
TemplateIndustryDurationCampaign objective
Problem, Solution, ProofE-commerce, beauty, fitness15-30 sPurchase, closed with an offer: free shipping, 15% off, 30-day guarantee
Before, During, AfterPhysical transformation, organization, productivity20-45 sPurchase after a visible transformation: link in bio, limited offer
Shock Hook, Intrigue, ResolutionSaaS, courses, innovative products15-25 sFree trial or sign-up
Every act narrows toward one action
Act 1: a problem the viewer recognizes
No logo, no warm-up: go straight to the problem
20%
Act 2: the solution, with proof
Show, do not tell: numbers and social proof
55%
Act 3: one action
Urgency or a guarantee, to reduce friction
25%
The three acts as shares of the runtime, and the shape they should produce. The narrowing is the whole discipline: an ad that ends with five asks (subscribe, like, comment, share, click) gets none of them, while the three templates below all close on a single one, whether that is an offer, a limited promotion or a free trial.

Fill-in template (copy and paste)

The fastest way to write a first draft is to fill the brackets below, read the result aloud with a timer, and cut inside Act 2 until it fits the runtime. Everything in square brackets is yours to replace; the sentences around them are the structure.

Act 1: hook and problem (first 20%)

"[PAIN POINT, phrased the way the viewer feels it]? [ONE LINE THAT PROVES YOU HAD IT TOO: "Same here" / "For [TIME PERIOD] I..."]." [VISUAL: the problem on screen, no logo, no packshot]

Act 2: solution and proof (20% to 75%)

"Then I tried [PRODUCT]. [WHAT IT DOES, in one sentence a friend would use]. [REALISTIC TIMEFRAME]: [VISIBLE RESULT]. [ONE NUMBER OR PROOF: reviews, customers, before/after]." [VISUAL: the product in use, a cut every 4 to 5 seconds]

Act 3: one CTA (last 25%)

"[SINGLE ACTION: tap the link / try it free / use code X]. [ONE REASON TO ACT NOW: guarantee, offer, deadline]." [VISUAL: the action written on screen, identical to the spoken line]

The "Problem, Solution, Proof" template

Industry: E-commerce, beauty, fitnessDuration: 15-30 s

Act 1: Problem (0-5 s)

""Sick of [SPECIFIC PROBLEM]? Same here. For [TIME PERIOD], I tried everything out there. Nothing worked.""

Name the problem PRECISELY. Not "dull skin" but "the dark circles at 7 a.m. before work."

Act 2: Solution (5-20 s)

""Then I discovered [PRODUCT]. In [REALISTIC TIMEFRAME], I saw a real difference. Here's what it changes: [VISUAL DEMONSTRATION or LIST OF 3 BENEFITS].""

Show the product in action. The visuals should back up the words.

Act 3: Proof + CTA (20-30 s)

""[NUMBER/SOCIAL PROOF: "3,000 five-star reviews", "92% of customers satisfied"]. Try [PRODUCT] with [OFFER: free shipping / 15% off / 30-day guarantee].""

The CTA should be one simple action. One message, one direction.

The "Before, During, After" template

Industry: Physical transformation, organization, productivityDuration: 20-45 s

Act 1: Before (0-7 s)

"Show the starting point without judging it: the initial situation, the chaos, the problem. Let the image speak. Low music, dull colors, slow pace."

No voiceover in Act 1. Let the visuals build empathy on their own.

Act 2: During / Discovery (7-30 s)

""Then I started using [PRODUCT / METHOD]. Week one: [OBSERVATION 1]. Week two: [OBSERVATION 2]. [SPED-UP MONTAGE OF THE TRANSFORMATION].""

The edit should create a sense of progress: every cut is another milestone hit.

Act 3: After + CTA (30-45 s)

""Today: [CONCRETE FINAL RESULT]. If you want the same result, link in bio / swipe up / [LIMITED OFFER].""

End on the subject's smiling face. Positive emotion is contagious.

The "Shock Hook, Intrigue, Resolution" template

Industry: SaaS, courses, innovative productsDuration: 15-25 s

Act 1: Shock hook (0-3 s)

"A counterintuitive statement, a shocking stat, or a provocative rhetorical question. E.g.: "You lose [NUMBER] minutes a day because of this invisible bug." / "Most Meta ads fail for this one reason.""

The first 3 seconds determine your completion rate. Invest 80% of your energy here.

Act 2: Intrigue and explanation (3-18 s)

""Here's why: [SIMPLE ONE-SENTENCE EXPLANATION]. Most people do [COMMON MISTAKE]. What the pros do instead: [ACTIONABLE INSIGHT].""

Deliver real value. The audience should learn something, even in an ad.

Act 3: Resolution + CTA (18-25 s)

""[PRODUCT] solves exactly this problem. [3-SECOND DEMO]. Try it free, link in bio.""

Quick resolution. The audience was hooked by the value, not the product. The product is the natural conclusion.

Five complete scripts by industry

Five finished scripts, one per industry, each between 60 and 90 spoken words so it fits its runtime at a natural pace. Replace the brackets, keep the timings, and the product should appear exactly where Act 2 starts.

Beauty (a serum), 30 seconds

Act 1 (0-6 s): "Dark circles at 7 a.m., every single morning, and no concealer that survived past lunch. Same?" Act 2 (6-22 s): "Three weeks ago I started using [SERUM] under my eyes, morning and night. Week one, nothing. Week two, the puffiness went. This is week three, no filter, no concealer." [Close-up in natural light, then the bottle in hand.] Act 3 (22-30 s): "It has a 30-day money-back guarantee, so the risk is on them, not on you. Link in bio."

Fitness (a training app), 30 seconds

Act 1 (0-6 s): "I trained five days a week for a year and looked exactly the same. Turns out the workouts were not the problem." Act 2 (6-22 s): "[APP] builds the plan around my schedule and my recovery, three sessions of 35 minutes, and it adjusts when I skip one. Twelve weeks in, here is the difference." [Before and after, same mirror, same light.] Act 3 (22-30 s): "Free for 14 days, no card. If you have been stuck like me, try one week."

App (budgeting), 20 seconds

Act 1 (0-4 s): "You are still tracking your spending in a notes app in 2026?" Act 2 (4-15 s): "[APP] connects to your bank, sorts every expense on its own and tells you on the 10th of the month if you are over budget, not on the 30th when it is too late." [Screen recording, one tap per sentence.] Act 3 (15-20 s): "It is free, and setup takes two minutes. Link below."

E-commerce (a sofa), 30 seconds

Act 1 (0-6 s): "We returned two sofas before this one. The photos looked great, the seat cushions did not survive a month." Act 2 (6-22 s): "[BRAND] sends fabric samples first, delivers in a week, and the frame has a 10-year warranty. This is ours after a year, two kids and a dog." [A hand pressing the cushion, then a wide shot of the living room.] Act 3 (22-30 s): "Free delivery and returns this month, the link is in the description."

Local service (a dental clinic), 25 seconds

Act 1 (0-5 s): "If you have been putting off the dentist because of the price, this is for [CITY]." Act 2 (5-18 s): "[CLINIC] posts the price of every treatment on its site before you book, sees you the same week, and sends the quote by text before anything starts. I went in for a check-up and paid exactly what the page said." Act 3 (18-25 s): "Book online in under a minute. The first check-up is [OFFER]."

3-act vs Hook-Problem-Solution-CTA vs PAS: which structure when

Choose by three criteria, in this order: the runtime you have, whether the transformation is visual, and how cold the audience is. The three structures below share the same skeleton and differ in where they spend the seconds.

The three script structures used in short-form ads, and the situation each one is built for.
StructureBeatsBest whenWatch out for
3-act (problem, solution, CTA)Hook and problem, solution with proof, one CTAYou have a visible transformation to show and 20 to 45 seconds to show itAct 2 runs long: cut inside it, never from the hook or the CTA
Hook, Problem, Solution, CTAThe 3-act with the hook split from the problem: four explicit beatsShort spots of 15 to 20 seconds, and UGC where the hook is a separate spoken lineFour beats in 15 seconds leave no room for proof: put the number inside the solution line
PAS (Problem, Agitate, Solve)Name the problem, make it hurt, then solve itPain-led offers (health, money, time) and cold audiences who do not know the product yetAgitating too long reads as manipulation: one sentence of agitation is enough in a 30-second ad

In practice: under 20 seconds, use Hook-Problem-Solution-CTA; between 20 and 45 seconds with something to show, use the 3-act; for a cold audience and an offer that solves a pain, use PAS and keep the agitation to one line. All three end the same way, with a single action.

AI Testimonial
Structure to use: Problem, Solution, Proof
UGC Beauty · Beauty of Joseon
Structure to use: Before, During, After

Two creative formats produced with SociaLover, and the template each one is written for. The same three acts can be carried by the voice, as in a testimonial, or by the visuals: the Before, During, After template deliberately leaves Act 1 without voiceover and lets the image build the empathy on its own.

Common mistakes to avoid

Leading with the brand

Putting the logo or brand name in the first 3 seconds is the number-one cause of skips. The audience has no reason to stay. Start with the problem, not with yourself.

A script too long for the format

A 30-second script for a 15-second TikTok video is a classic mistake. Write the script by reading it out loud and timing yourself. If it runs long, cut from Act 2. Never from the hook or the CTA.

Multiple CTAs

"Subscribe, like, comment, share AND click the link in bio" is the best way to get zero action. One CTA, one action, one message.

A benefit that's too vague

"Our product is great" doesn't convert. "Lose 3 kg in 4 weeks or your money back" converts. Be specific, quantified, and credible.

Ignoring silent viewing

A large share of feed viewing happens with the sound off, on TikTok as elsewhere. Your script has to work with or without audio: captions, expressive body language, visual elements that tell the story on their own.

Where the seconds go in a 30-second script
Act 1: hook and problem
6 s
Act 2: solution and proof
16 s
Act 3: CTA
8 s
The 20 / 55 / 25 split applied to a 30-second format: 6 seconds, 16 seconds, 8 seconds. Read it as a cutting guide rather than as a plan: the 16 seconds in the middle are the only ones you are allowed to shorten. Taking two seconds off the hook or off the CTA costs you more than dropping an entire benefit from Act 2.

Using AI to write your scripts

Current Claude, GPT and Gemini models are strong co-pilots for video scripts, provided the prompt carries the structure: the product, the audience, the goal, the exact act timings and the hook family. Our test of which LLM writes the best UGC scripts shows where each one is strongest.

Example prompt

"Write 3 twenty-second video scripts for a TikTok Ads campaign for [PRODUCT]. Target: [AUDIENCE DESCRIPTION]. Goal: [PURCHASE / SIGN-UP / CLICK]. Use the 3-act structure (Problem 0-5 s / Solution 5-15 s / CTA 15-20 s). Each script should have a different hook: 1) shocking question, 2) surprising stat, 3) counterintuitive statement. Include editing directions."

In SociaLover the same prompt lives inside the product: the Script step in Studio Video (Script Ideas) drafts the acts from your brief, the hook formulas for TikTok and Reels give you the three openings to test, and once a script is approved you can turn the script into an AI UGC video, cut it in the Editor, and export every placement with Creative Resizer. Start to finish, the guide to produce it in 10 minutes walks through the whole chain.

Frequently asked questions

What is the 3-act structure for a video ad?
It splits a short ad into three jobs: Act 1 names a problem the viewer recognizes, Act 2 shows the solution in action with proof, Act 3 asks for a single action. Act 1 takes the first 20% of the runtime, Act 2 runs from 20% to 75%, and Act 3 closes the last 25%. It converts because each act has to earn the next few seconds of attention instead of assuming them.
How long should each act be in a 30-second ad?
Roughly 6 seconds for Act 1, 16 seconds for Act 2, and the final 8 seconds for Act 3: that is the 20 / 55 / 25 split applied to a 30-second runtime. If the script comes out too long when you read it aloud, cut inside Act 2: shortening the hook or the CTA costs you more than losing a benefit.
How long should a TikTok ad script be?
60 to 90 words for a 30-second ad, spoken at a natural pace of two to three words per second, and 30 to 45 words for a 15-second ad. Read it aloud with a timer before you record or generate: if it runs long, cut inside Act 2, never from the hook or the CTA. On-screen captions can be shorter than the spoken line, never longer.
Does the 3-act structure work for a 15-second ad?
Yes, with the same proportions compressed: about 3 seconds for the hook and problem, 8 seconds for the solution and proof, 4 seconds for the CTA. Merge the problem into the hook line, keep one proof number instead of three benefits, and show the product from the fourth second. The Shock Hook template is the one built for that runtime.
Where should the product first appear?
At the start of Act 2, right after the problem has been named, which is around the 20% mark: the 5th or 6th second of a 30-second ad, the 3rd or 4th of a 15-second one. Not in the first frame, because that is a logo intro and the number-one cause of skips, and not only at the end, because a viewer who leaves at 10 seconds should already know what the product is.
Should the CTA be spoken or written?
Both. Speak it once, in the last act, as one simple action ("the link is in the bio"), and write the same words on screen at the same time, because a large share of feed viewing happens with the sound off. Keep the spoken and written versions identical: two different CTAs in the same three seconds read as two asks, and two asks get none.
Which template should I use for my product?
Problem, Solution, Proof (15-30 s) for e-commerce, beauty and fitness, where the offer closes the sale. Before, During, After (20-45 s) for physical transformation, organization and productivity products, where the result is visual. Shock Hook, Intrigue, Resolution (15-25 s) for SaaS, courses and innovative products, where the goal is a free trial or a sign-up.
Should I put my logo at the beginning of the video?
No. Leading with the brand (logo or brand name in the first 3 seconds) is the number-one cause of skips, because the audience has no reason yet to stay. Start with the problem, not with yourself. The brand can appear once Act 2 has earned the attention.
How many calls to action should a video ad have?
One. "Subscribe, like, comment, share and click the link in bio" is the surest way to get zero action. One CTA, one action, one message, and it should be a simple action, with a touch of urgency or a guarantee to reduce friction.
Can AI write my video ad script?
Yes: current Claude, GPT and Gemini models are strong co-pilots for this, provided the prompt is precise. Give them the product, the audience, the goal, the exact act timings, and ask for several scripts that differ only by the hook: a shocking question, a surprising stat, a counterintuitive statement. In SociaLover, the Script step in Studio Video does this from your brief.

Conclusion

The three-act structure isn't a rigid formula: it's a framework that frees your creativity by giving you a narrative backbone. Master these three templates, test the variations, and analyze your completion rates. After a few weeks of systematic testing, you will know which structure resonates best with your specific audience.