AI video23 min read9 October 2026

How to Write a Video Ad Script: A 9-Step Method With Word-Count Math and a Worked Example (2026)

To write a video ad script, start with one job and one message, write the hook first, back the message with proof the camera can show, and end with a single call to action. Then write what the viewer sees as carefully as what they hear, and cut until the words fit the length. That is the whole method in two sentences. The craft is in the details.

This guide shows you how to write a video ad script the way we do it for Meta, TikTok and YouTube campaigns. You get a one-page brief to fill in first, nine steps from blank page to final draft, the word-count math for 6, 15, 30 and 60 second ads, a full two-column example, and versions for UGC creators, AI video and studio shoots. We also cover Arabic and English scripts for the UAE and GCC, and how to plan variants so every script teaches you something.

What a video ad script actually has to do

A video ad script is the plan for every second of the ad. It sets what is said, what is shown, what appears as on-screen text and how long each beat lasts. A screenplay tells a story for people who chose to watch. An ad script has a harder job: it has to win attention from people who didn't choose to watch, deliver one idea, and get a specific action before they leave.

Most short ads that work follow the same four beats:

Hook: a first line and first frame that make the right person stop.
Problem or desire: the situation the viewer recognizes as theirs.
Proof: the product doing its job, shown rather than claimed.
Call to action: one clear next step, said and shown.

Every framework you will read about is a variation on these four beats. What changes is the order, the emphasis and how long each beat gets.

A script has two tracks, not one

The biggest gap we see in first drafts is that they are written as voiceover copy. The writer polishes the words and leaves the pictures to "whatever the editor finds." But many people watch social ads with the sound off, and even with sound on, viewers remember what they saw more than what they heard.

So a real ad script has two tracks:

The audio track: voiceover, dialogue, music cues and sound effects.
The visual track: shots, actions, product moments and on-screen text (supers and captions).

A good test: mute the ad in your head and read only the visual track. If the message still lands, the script works. If it falls apart, the visual track needs more writing. Our 10-point scorecard for a good video ad uses the same sound-off test as one of its checks.

Before you write: fill in a one-page brief

Scripts go wrong before the first line is written. If the brief is vague, the script tries to say everything and ends up saying nothing. We fill in this page before anyone writes a hook.

Audience: What to write: Who, in one sentence, with their current behavior; Example answer: Busy professionals in Dubai who order takeaway four nights a week
Pain or desire: What to write: What they feel in their own words; Example answer: "I want to eat better but I have no time to cook"
One message: What to write: The single idea the viewer should keep; Example answer: Healthy dinners that arrive ready, without the takeaway guilt
Proof: What to write: What you can show on camera; Example answer: Real meals plated in under a minute, the delivery box, a customer's fridge
Offer: What to write: Why act now; Example answer: First-week trial, free delivery in selected areas
Action: What to write: The one step you want; Example answer: Tap to see this week's menu
Platform and placement: What to write: Where it runs and the aspect ratio; Example answer: Instagram Reels and TikTok, 9:16
Length: What to write: Target seconds; Example answer: 15 and 30 second cuts
Must-haves: What to write: Legal lines, brand rules, claims to avoid; Example answer: No medical claims, logo within the first 3 seconds

Two rules make the brief useful. First, write the audience's pain in their words, not marketing words. "I'm too tired to cook" beats "consumers seeking convenient nutrition solutions." Second, the proof column must list things a camera can capture. If your proof is "high quality," you don't have proof yet.

How to write a video ad script in 9 steps

Here is the full process we use. It works for a 6-second bumper and a 60-second story ad. Only the amount of material changes.

Step 1: Give the ad one job

Decide what this specific ad is for. Is it introducing the brand to cold audiences, getting a click from people who already know you, or bringing back someone who left a cart? Each job needs a different script.

Cold audiences need a hook built around a problem they recognize and a clear explanation of what you are.
Warm audiences already know you. Lead with the offer, a new product or a strong objection-handler.
Retargeting works best with reassurance: reviews, guarantees, delivery times, how easy it is to start.

One ad, one job. If the brief asks for awareness and sales and app installs in 30 seconds, split it into three scripts.

Step 2: Write the one sentence the viewer should remember

Before the hook, write the message as one plain sentence. For our example brief: "Healthy dinners arrive ready, so you stop ordering takeaway." Every line of the script must support this sentence. If a line doesn't, it gets cut, however clever it is.

This sentence is not always spoken in the ad. It is the filter you use while writing. Teams that skip it tend to stack three or four benefits into 15 seconds, and viewers keep none of them.

Step 3: Choose a framework that fits the job

Frameworks give the four beats an order. Pick one based on what your audience needs to believe before they act.

PAS (Problem, Agitate, Solution): Beat order: Name the pain, make it vivid, show the fix; Best for: Products that solve a felt problem; Watch out for: Agitating too long; keep it under a third of the ad
AIDA (Attention, Interest, Desire, Action): Beat order: Hook, explain, show the outcome, ask; Best for: New categories that need explaining; Watch out for: Turning "interest" into a lecture
BAB (Before, After, Bridge): Beat order: Life before, life after, the product as the bridge; Best for: Transformations: fitness, beauty, home, software; Watch out for: Before-and-after claims you can't support
Demo first: Beat order: Show the product working, then explain; Best for: Visual products with an obvious "wow"; Watch out for: Forgetting to say who it's for
Testimonial: Beat order: A real customer tells the story; Best for: Trust-heavy purchases; Watch out for: Scripting it so tightly it sounds fake
Us vs the old way: Beat order: The usual method, then your method; Best for: Replacing an existing habit; Watch out for: Naming or mocking other brands

For short-form social ads, PAS and demo-first scripts are usually the easiest to get working. AIDA suits longer YouTube ads where you have time to explain. If you are unsure, write the same message in two frameworks and test both (more on that below).

Step 4: Write five hooks before you write the body

The hook decides whether anyone sees the rest, so write it first and write several. We write at least five hooks per script, each with a different angle:

Call out the viewer: "If you order takeaway more than twice a week, watch this."
Show the result first: open on the finished meal, then rewind to how it got there.
Say the pain out loud: "I was spending my evenings deciding what to order."
Make a specific claim: "Dinner on the table in 60 seconds, no cooking."
Break a pattern: an unexpected visual, like a fridge opening to reveal a full week of meals.

A hook is a line and a frame together. "Watch this" over a static logo is not a hook. The same line over a hand pulling a hot meal from a box is. For 40 more openers sorted by what they are good at, see our list of video ad hooks by the job they do.

Keep all five. You will use the best one in your main script and the others as test variants.

Step 5: Build the body from proof, not adjectives

The body is where scripts get lazy. "Delicious, healthy, convenient meals" is three adjectives the viewer has heard a thousand times. Replace each adjective with something that shows it:

Instead of "delicious," show steam rising when the lid comes off and a real reaction on the first bite.
Instead of "healthy," put the ingredient list or calorie count on screen for two seconds.
Instead of "convenient," show the timer: box opened at 7:02, eating at 7:03.

Good proof types for video include product demos, close-ups of texture or detail, side-by-side comparisons with the old way (never another brand), customer reactions, numbers on screen, and process shots that show how something is made.

Keep the body tight. In a 15-second ad, you have room for one or two proof moments. In 30 seconds, three or four. More than that and the ad starts to feel like a catalog.

Step 6: Write the visual track line by line

Now go back through the script and write what is on screen for every line. Use the two-column format: visuals on the left, audio on the right, with a time stamp for each row. We show a full example in the next section.

For each row, specify:

The shot: close-up, medium, wide, overhead, handheld or locked off.
The action: what moves. A shot where nothing happens is a shot where people scroll.
On-screen text: the exact words of each super, kept to a few words so they can be read in the time they are on screen.
Sound: music change, sound effect or silence.

Write on-screen text as its own message, not a copy of the voiceover. The supers should carry the core idea for someone watching on mute, while the voiceover adds detail and personality for those listening.

Step 7: Write the call to action and the end card

Your call to action should be one action, said and shown, and it should match the button the platform shows. If the button says "Shop now," don't say "Book a call" in the voiceover.

Make the action easy and specific:

"Tap to see this week's menu" is clearer than "Learn more."
"Order before 9 pm for tomorrow" adds a reason to act now.
"Your first week, delivered Sunday" makes the next step concrete.

Then write the end card: logo, the CTA line as text, and one supporting detail (an offer or a reassurance). Keep it on screen for at least two seconds. In longer ads, put a soft version of the CTA earlier too, because many viewers will not reach the end.

Step 8: Do the word-count math and cut

Most scripts are too long. Voiceover artists typically read conversational ad copy at about 140 to 160 words per minute, which is roughly 2.5 words per second. That is a rule of thumb, not a law: fast, energetic UGC reads run quicker, and calm luxury reads run slower. But it gives you a ceiling to write to.

6 seconds: Rough spoken-word ceiling: 10-15 words; What fits: One hook line and the brand; Typical use: YouTube bumpers, story teasers
15 seconds: Rough spoken-word ceiling: 30-38 words; What fits: Hook, one proof, CTA; Typical use: Reels, TikTok, Stories, Shorts
30 seconds: Rough spoken-word ceiling: 65-75 words; What fits: Hook, problem, two or three proofs, CTA; Typical use: In-feed social, skippable YouTube
60 seconds: Rough spoken-word ceiling: 130-150 words; What fits: A short story or full demo with objections handled; Typical use: YouTube, long-form feed ads

Write to about 80% of the ceiling. Pauses, reactions and product moments need air, and a script that fills every second with words feels rushed. If you are over, cut adjectives first, then any line that repeats the message, then the weakest proof point.

Step 9: Read it aloud, then do a sound-off pass

Read the full script out loud with a timer. Anything you stumble over, the viewer will stumble over too. Swap long words for short ones and break long sentences in two.

Then do a sound-off pass: read only the visual column and on-screen text. Ask three questions:

Would a muted viewer understand what the product is within five seconds?
Is the brand visible early, ideally on the product itself, rather than only on the end card?
Is the CTA readable on screen, not just spoken?

Finally, hand the script to someone who wasn't in the brief meeting. If they can't tell you the one message after reading it, rewrite before you shoot.

A worked example: a 30-second script in two-column format

Here is a full 30-second script for the fictional meal delivery brand from our brief. It uses a PAS structure, runs 9:16 for Reels and TikTok, and carries about 60 spoken words, comfortably inside the 30-second ceiling from step 8.

0-2s: Visual: Close-up: phone screen scrolling a food delivery app, thumb hovering. Super: "Night four of takeaway?"; Audio: VO (tired, friendly): "Fourth takeaway this week?"
2-6s: Visual: Medium shot: person on sofa, empty delivery bags on the table, sighs.; Audio: VO: "We've all been there. No time to cook, and you know it's not doing you any favors."
6-10s: Visual: Hard cut, music lifts. Overhead shot: branded box opened, four labeled meals inside. Super: "Ready in 60 seconds"; Audio: VO: "So we cook for you. Chef-made dinners, delivered ready."
10-16s: Visual: Close-up: meal into microwave, timer shows 1:00, steam as the lid comes off.; Audio: SFX: microwave beep. VO: "Heat for a minute and dinner's done."
16-21s: Visual: Close-up: plated meal, ingredient card on screen. Super: "Under 600 calories. Real ingredients."; Audio: VO: "Balanced, filling, and nothing you can't pronounce."
21-25s: Visual: Medium shot: same person eating, smiling, phone face down on the table.; Audio: VO: "Same evening. Better dinner. No guilt."
25-30s: Visual: End card: logo, the week's menu on screen, button highlight. Super: "See this week's menu"; Audio: VO: "Tap to see this week's menu. First delivery this Sunday."

Notice a few things. The brand box appears at six seconds and the logo sits on the box itself. Every super makes sense without audio. The proof moments (the box, the timer, the ingredient card) are all things a camera can capture. And the CTA is the same in the voiceover, the super and the button.

To turn this into a 15-second cut, keep rows one, three, four and seven, and drop the agitation and the second proof. To turn it into a 6-second bumper, keep the box opening with the super "Dinner, ready in 60 seconds" and the logo.

How to adapt one script for each platform

The core message stays the same across platforms. The build changes. Here is what we change when one script needs to run in several places.

TikTok: What changes in the script: Native, creator-style voice, faster cuts, captions in the platform style; Hook window: First 1-2 seconds; Notes: Should feel like a TikTok, not an ad placed on TikTok
Instagram Reels and Stories: What changes in the script: Similar to TikTok, slightly more polished; keep text clear of the interface at top and bottom; Hook window: First 1-2 seconds; Notes: Stories favor short, punchy scripts with a tap-forward rhythm
Facebook feed: What changes in the script: Can carry more explanation; many viewers watch muted, so supers matter most; Hook window: First 3 seconds; Notes: Square or vertical framing usually wins more screen space
YouTube skippable in-stream: What changes in the script: Viewer can skip after 5 seconds, so brand and hook must land before then; Hook window: First 5 seconds; Notes: Sound-on is more common, so voiceover and music do more work
YouTube Shorts: What changes in the script: Vertical, swipe-based; treat like TikTok with a YouTube audience; Hook window: First 1-2 seconds; Notes: Build a native vertical cut, don't crop the 16:9 edit
Snapchat: What changes in the script: Very short, full-screen, young audience; fast, direct, playful; Hook window: First second; Notes: Lead with the visual, keep words minimal

On YouTube, the length rules depend on the format. Bumper ads run up to six seconds and skippable ads can be skipped after five, as Google's overview of video ad formats explains. That five-second skip point is why we put both the hook and the brand inside the first five seconds of every skippable script. Our guide to YouTube ad creative by format goes deeper on building each one.

For vertical feeds, the main rule is speed. The script has to make sense from the first frame, and the pacing has to change every two to three seconds. Our short-form video ads guide covers lengths, formats and pacing for TikTok, Reels and Shorts in detail.

How the script changes for UGC creators, AI video and studio shoots

Who or what reads your script changes how you write it. A creator, an AI video tool and a film crew each need different levels of detail.

UGC creator: Who reads the script: A creator filming on their phone; How detailed: Beat sheet, not word for word; What to include: The hook line, key points, must-say claims, must-avoid claims, the CTA, shot list
AI video: Who reads the script: A generation tool, shot by shot; How detailed: Very detailed per shot; What to include: One visual prompt per shot (subject, action, setting, lighting, camera, framing), the exact line, the duration
Studio or location shoot: Who reads the script: Director, crew, talent and editor; How detailed: Full two-column script plus storyboard; What to include: Every shot, every line, supers, music cues, props, wardrobe, locations

Writing for UGC creators

Creators sound real because they speak in their own words. If you hand them a word-for-word script, they read it like an actor, and the ad loses what made UGC work. Give them a beat sheet instead: the hook line to say exactly, then three or four points to cover in their own words, then the exact CTA.

Be strict about two things: claims they must not make (health claims, guarantees, results you can't support) and the shots you need (product in hand, a close-up of the key feature, a reaction). For scripts written by industry and what to borrow from each, see our UGC ad examples with scripts.

Writing for AI video

AI video tools render what the words describe and nothing more. So an AI script is a numbered list of shots, and each shot gets its own complete prompt: who or what is in frame, the action, the setting, the lighting, the camera movement and the framing, plus the line spoken over it and how long it lasts.

Write one action per shot. "She opens the box, smiles, takes out a meal and heats it" is four shots, not one. Keep product shots honest: if the tool can't render your real packaging accurately, use real footage of the product and generate only the scenes around it. This is how we work at XMA. AI handles speed and variant volume, and our human creative team handles the script, the edit and the brand details the tools get wrong.

Writing for a studio or location shoot

A crew needs the full two-column script, a storyboard or shot list, and production notes. Write the locations, props and wardrobe into the script, and flag any shot that needs special equipment. Plan alternate takes of the hook and CTA while you have the set, because they are the cheapest variants you will ever get.

Writing Arabic and English video ad scripts for the UAE and GCC

If you advertise in Dubai or across the Gulf, you will often need an Arabic and an English version of the same ad. This is where many scripts fail, because the Arabic version is treated as a translation job at the end.

Transcreate, don't translate

A line that works in English can feel flat or odd when translated word for word. Write the Arabic version from the brief, not from the English script. Keep the same message, structure and CTA, but let the hook and phrasing be native. Wordplay almost never survives translation, so if your English hook relies on a pun, write a different hook for Arabic.

Choose the right register

Decide early between Modern Standard Arabic and a Gulf dialect. Modern Standard Arabic reads as formal and works for official, premium or pan-Arab ads. Gulf dialect sounds closer to how people actually talk and often suits UGC-style and social-first ads aimed at UAE and Saudi audiences. Mixing registers in one ad sounds off to native speakers, so pick one per version.

Plan for different lengths and right-to-left text

The same message can take noticeably more or fewer words, and more or less screen space, in Arabic than in English. Time the Arabic voiceover separately rather than assuming it fits the English edit. On screen, Arabic supers run right to left, so check that text placement, animations and arrows still make sense when mirrored, and keep text clear of the platform interface.

Respect timing and culture

Scripts for Ramadan, Eid and UAE National Day need their own tone. Ramadan ads often shift toward family, generosity and evening routines, and viewing patterns move later into the night. Keep humor, wardrobe and scenarios appropriate for a broad Gulf audience. If creators or influencers deliver your script in the UAE, check the current advertising rules from the UAE Media Council before you brief them.

How to write variants so every script teaches you something

One script is a guess. Several planned variants are a test. The difference between teams whose ads keep improving and teams who keep starting from zero is usually how they plan variants at the script stage.

The simplest system is a hook by body matrix. Write three hooks and two bodies, and you have six ads from one script session. Change one thing at a time so you can read the result:

A: Hook: Call out the viewer; Body: PAS with timer proof; CTA: Menu; What it tests: Baseline
B: Hook: Result first; Body: PAS with timer proof; CTA: Menu; What it tests: Which hook stops more people
C: Hook: Say the pain; Body: PAS with timer proof; CTA: Menu; What it tests: Which hook stops more people
D: Hook: Call out the viewer; Body: Testimonial; CTA: Menu; What it tests: Which body holds attention and converts
E: Hook: Best hook from A-C; Body: Best body from A, D; CTA: Trial offer; What it tests: Whether the offer changes conversion

Then read the numbers back into the next script. If people don't stop, the hook is the problem. If they stop but drop off early, the body loses them. If they watch but don't act, the CTA or offer needs work. Treating creative as a measurable, tested system like this is the core idea behind performance creative, and our ad creative testing framework shows how to structure the tests themselves.

Google's research on its ABCD approach to YouTube creative, summarized in its creative best practices for YouTube ads, points in the same direction: attention, early branding, connection and clear direction all start with what you write before you shoot.

Mistakes that weaken a video ad script

These are the problems we fix most often when clients send us scripts to review.

Opening with the logo. A slow logo reveal spends your best seconds on something the viewer doesn't care about yet. Brand early, but on the product or in the action.
Writing for sound only. If the visual column just says "B-roll," the ad fails for muted viewers.
Too many messages. Three benefits in 15 seconds means none of them land.
Adjectives instead of proof. "Premium quality" proves nothing. A close-up of the stitching does.
Mismatched CTA. The voiceover says one thing, the button says another.
Overlong scripts. Ninety words in a 30-second ad forces a rushed read and no room for product moments.
Over-scripted UGC. Word-for-word lines make creators sound like actors.
Translated, not transcreated, Arabic. It shows, and native speakers notice instantly.
Claims you can't support. Health results, guaranteed outcomes and invented statistics can get ads rejected and damage trust.
One version only. Without variants, you learn almost nothing from the spend.

A pre-production script checklist

Run every script through this list before it goes to production:

The ad has one job and one message, written as one sentence.
The hook works as both a line and a frame, and there are at least four alternates.
The brand appears in the first few seconds, ideally on the product.
Every claim has a proof moment the camera can capture.
The visual column and supers carry the message with the sound off.
The spoken words fit about 80% of the length's word ceiling.
The CTA is one action, matches the platform button and appears as text.
There is a native version for each platform and aspect ratio you plan to run.
Arabic and English versions are written separately and timed separately.
Variants are planned so each test changes one thing.

Frequently Asked Questions

How do you write a video ad script step by step?

Start with a one-page brief: audience, pain, one message, proof, offer and action. Give the ad one job, write the core message as one sentence, pick a framework such as PAS or AIDA, and write five hooks before the body. Build the body from proof you can show, write the visual track line by line, add one clear call to action, then cut to length and read it aloud.

How many words should a video ad script be?

Use about 2.5 spoken words per second as a rule of thumb. That gives roughly 10-15 words for a 6-second ad, 30-38 for 15 seconds, 65-75 for 30 seconds and 130-150 for 60 seconds. Write to about 80% of that ceiling so pauses, reactions and product moments have room. Fast UGC reads can carry more words, while calm premium reads need fewer.

What is the best framework for a video ad script?

It depends on what the viewer needs to believe. PAS (problem, agitate, solution) works well for products that fix a felt problem. AIDA suits new categories that need explaining. Before-after-bridge fits transformations. Demo-first works for visual products. If you are unsure, write the same message in two frameworks and test both versions against each other.

How do you format a video ad script?

The standard format is two columns with time stamps. The left column describes the visuals: shot type, action, product moments and on-screen text. The right column holds the audio: voiceover, dialogue, music and sound effects. Each row covers a few seconds. This format makes it easy to check the ad works with the sound off and to time each beat.

Should I give a UGC creator a full script?

Usually not. A word-for-word script makes creators sound like actors, which removes the authenticity that makes UGC work. Give them a beat sheet instead: the exact hook line, three or four points to cover in their own words, the exact call to action, a shot list, and a clear list of claims they must not make. Ask for alternate takes of the hook.

Can AI write a video ad script?

AI can draft scripts quickly, especially hook variations and first drafts from a good brief. It still needs a human to check the message, cut generic phrases, confirm every claim is true and make sure the script fits the brand and culture. For AI-generated video, the script also needs a detailed visual prompt for every shot, not just dialogue.

Get your next video ad scripted and produced

A strong script is the cheapest place to improve an ad. Every fix you make on the page saves a reshoot, a re-edit or a test that teaches you nothing.

At XMA, our team writes and produces video ads with AI plus a human creative team, for brands in Dubai, across the GCC and globally. We script the hooks, the visual track, the Arabic and English versions and the variants, then deliver ready-to-run ads for Meta, TikTok and YouTube. If you want scripts and ads built this way, book a strategy call and get a quote for your next campaign.