Video ad script: four templates with timings, a 30-second example and a shot list

A video ad script is a second-by-second plan of what a short ad shows, says and writes on screen. A good one opens with a hook in the first three seconds, names one problem or desire, shows the product doing its job, gives proof, and ends with one clear action. Write it in columns, picture and sound side by side, so it still works with the sound off.

What goes into a video ad script

A script for a short ad is a table, not a paragraph. Each row is a beat with a start and an end time, and four columns:

The picture column comes first because the picture reaches every viewer, sound on or off. The on-screen text column is not a transcript: it carries the hook and the main claim in as few words as possible. The voice column is timed. The National Center for Voice and Speech puts average US English speech at about 150 words per minute (NCVS), which is 2.5 words a second. That gives a budget of about 75 words for 30 seconds, 37 for 15 and 15 for a 6-second bumper. Read every draft aloud with a timer. If it runs long, cut words, not pauses.

How to write the hook: the first three seconds

The hook is the first three seconds, and it has three parts that must agree: a picture with motion, a line of text and a first spoken line. Three seconds is not an arbitrary number. Hook rate is built on 3-second video plays, and our guide to hook rate shows how to measure it. On YouTube, a skippable ad gets a skip button after 5 seconds, so the hook and the product's name both need to land before it.

Good hooks open a question the viewer wants closed: a scene in motion, a surprising detail, a demonstration already under way. The VSL hook examples guide has 24 formulas, and many of them adapt to a 3-second ad once you cut them to about seven words. To study the hooks competitors run, our Facebook Ad Library guide shows how to search their ads and what to note from each.

One rule matters most for health brands. Meta's Advertising Standards say ads must not assert or imply personal attributes, a list that includes age and physical or mental health, medical conditions included. The policy page lists "Do you have diabetes?" among its health examples and flags "you" or "your" wording that points at an attribute (Meta personal attributes policy). Describe a situation instead of diagnosing the viewer.

Here is the difference, using a fictional joint-support powder called Hillward:

Hooks for a fictional joint-support powder

Rewrite itWeak or risky

• "Hi, I'm Dana from Hillward, and today I want to..." (no hook, logo first)
• "Do you have bad knees?" (asks the viewer about their health)
• "Over 50? This changes everything." (asserts the viewer's age)
• "Doctors hate this joint trick" (clickbait, and an expert view you can't show)
• "Ease arthritis pain fast" (a disease claim)

Test itStrong and safe

• "Nobody warns you about the last staircase." (a scene)
• "This goes in my coffee before every hike." (a demo already under way)
• "Three things I check on a supplement label." (curiosity, and true)
• "The trail crew that never skips a Saturday." (identity, no attribute)
• "One scoop, ten seconds of stirring, gone." (a fact you can see)

The weak column fails for different reasons: one has no hook at all, two break Meta's rule, one implies an expert opinion the brand can't show, and one is a disease claim no supplement may make. The strong column relies on scenes, demonstrations and facts the viewer can see. Write three hooks like these for every body, film them separately, and let the test decide.

Four video ad script templates with timings

Pick the angle first (our post on ad angles shows where they come from), then pick the template by the proof you can put on screen. Each template runs about 30 seconds here; the shorter cuts come later.

Start from what you can show:

Which of the four templates fits?

Pick by the proof you can put on screen; each runs about 30 seconds

1What proof do you have to put on screen?

A problem the viewer already feelsA demo you can filmA real customerA true founder story

Best fit

Problem–agitate–solve

Use it when the viewer knows the problem, not your product.

• Beats in 30 seconds: hook 0–3 · problem 3–8 · product 8–18 · proof 18–25 · action 25–30
• Proof it needs: claims you can substantiate

Best fit

Demo-first

Use it when the product does something you can film.

• Beats in 30 seconds: demo in the first frame 0–3 · steps 3–15 · why it matters 15–24 · action 24–30
• Proof it needs: one honest, unedited take

Best fit

Testimonial-led

Use it when a real customer agreed to be filmed.

• Beats in 30 seconds: their line 0–4 · their story 4–15 · product and claim 15–24 · action 24–30
• Proof it needs: a real customer, a release, typical results

Best fit

Founder story

Use it when the origin is true and specific.

• Beats in 30 seconds: a true moment 0–3 · why they built it 3–15 · what's different 15–24 · action 24–30
• Proof it needs: facts you can document

Problem–agitate–solve

Open on the problem as a scene, make its cost felt, then present the product as the way out. The agitate beat is where ads go wrong. Make the cost real and specific, never frightening: Google's misrepresentation policy bans ads that use negative life events such as illness to induce fear or guilt (Google Ads policy). An everyday cost, like skipped Saturdays, does the job honestly.

Demo-first

Start with the product already working: motion in the first frame gives the viewer a reason to look before a word is said. For a gadget, the demo is the result. For a supplement, it's the routine: the scoop, the stir, the label. A supplement demo must never stage a result. Our post on product demo videos covers formats and the honesty rules.

Testimonial-led

Open on the customer's own sentence, in their words, then let the brand state the claim. The customer describes their experience; the claim about what the product does belongs to you and must be substantiated. If the person on screen is presented as a customer, the FTC's Endorsement Guides call for actual customers, or a clear disclosure that they aren't (16 CFR Part 255, section 255.2(c)). Paid creators bring their own rules, covered in our post on UGC ads.

Founder story

Open on a true moment from the founder's life, explain why they made the product, and show what's different about it. It works when the story is specific and documented. It fails when it's invented, and for a supplement it can never be "I cured my own condition with it".

A 30-second video ad script, worked example

Here is a problem–agitate–solve script for Hillward, the fictional powder. First, the shape of it:

A 30-second problem–agitate–solve ad

1. 10–3 s · HookA scene in motion, about seven words on screen
2. 23–8 s · ProblemThe cost of the problem, real and specific
3. 38–18 s · ProductWhat it is, how it fits the day, one approved claim
4. 418–25 s · ProofSomething the viewer can check
5. 525–30 s · ActionOne step to take, and the guarantee if there is one

And the full script. The voice column holds 68 words, about 27 seconds at 150 words a minute (68 ÷ 2.5), which leaves room to breathe:

Notice what it doesn't say. There's no "pain", no named condition, no promise of how much better anyone will feel, and no "you" attached to a health problem. The one product claim is a structure/function claim, and the brand needs evidence for it. A disease claim, such as easing arthritis, is off limits for a supplement: under 21 CFR 101.93(g), claiming an effect on a disease or its characteristic symptoms makes it a disease claim (eCFR). The guarantee line goes in only if the offer really has one and honors it. This is a summary of the rules, not legal advice; the linked primary sources have the full text.

Writing for sound-off: captions and on-screen text

You don't control whether the viewer has the sound on, so write every ad to work both ways:

• Tell the story in the picture first. Mute your rough cut. If the beats don't make sense, fix the picture, not the voice.
• Burn in captions for every spoken line. Short lines, high contrast, no more than two lines on screen at once.
• Use on-screen text for the hook and the claim, worded like the voice so the two don't compete.
• Keep text away from the edges. App buttons and ad captions cover parts of the frame. TikTok publishes safe-zone templates for in-feed ads, and their size depends on the video's shape and the length of the ad caption (TikTok in-feed ad specs).
• Let sound add, not carry. The voice can add warmth and detail, but no fact should exist only in the audio.

Meta's own explainer on Reels creative covers designing for the full screen, audio and safe zones:

How long should a video ad be? Length by placement

Length follows the placement, and placement rules change, so check the current spec page on the day you edit. Here is what Google and TikTok publish today:

The YouTube rows come from Google's video ad formats page and the TikTok row from its in-feed ad specs, linked above. For Meta's placements, check the current video specs in Ads Manager before you edit.

The decision rule: write the 30-second version first, because it forces the whole argument into the smallest space that still holds one. Cut a 15-second version by dropping the problem beat and shortening the proof, and a 6-second bumper that keeps only the hook, the product and the name. Go longer when the product needs a mechanism explained, for example on YouTube with cold traffic; our post on long-form video ads covers when that wins.

From script to shot list

A script tells the editor what to say. A shot list tells the camera what to capture. Turn every picture cell into numbered shots with framing, action, props and duration, and film more than the script needs:

• Three versions of the hook, filmed separately, so you can test openings without a reshoot.
• The product in hand, in use and in close-up, plus a clean end frame with the name and the offer.
• Cutaways of the setting, so the editor can cover jump cuts.
• Clean voice audio recorded close to the speaker, even if you plan to add music.
• Vertical framing with room at the top and bottom for captions.

Before the shoot, run the script through this list:

Before the shoot: script check

• Every beat has a time, a picture, on-screen text and a spoken line
• The hook works muted: picture and text tell the story without audio
• The voice fits the time: about 75 words for 30 seconds
• Every claim is on your approved list, with the evidence on file
• No line asserts or implies the viewer's health condition or age
• Three hooks are written for the same body
• On-screen text stays clear of the edges the app covers
• A 15-second and a 6-second cut are planned from the same footage
• The shot list covers product in hand, in use, close-up and an end frame
• Captions are planned for every spoken line

0 of 10 done

Hand the promise to the VSL

If the ad sends people to a video sales letter, the ad's promise should be the VSL's first line. A viewer who clicked for the staircase story and lands on a lecture about cartilage thinks they're in the wrong place. Script the ad and the VSL's opening together, and reuse the ad's on-screen line as the VSL's first caption.

The VSL page has the same sound problem as the feed. On a page that autoplays muted, the opening plays as a silent teaser first. RevvyScales restarts the video at 0:00 when a first-time viewer taps for sound, so they hear the promise from the first word instead of joining mid-sentence (player conversion features). If the ad points to a short VSL rather than a long one, the next guide covers how to build it.

Write your next script this week

1. Pick one angle and one template, based on the proof you can show.
2. Draft the 30-second version in the four-column table, at about 75 spoken words.
3. Write three hooks for the same body, each with a picture, a line of text and a spoken line.
4. Check every line against your approved claims and against the personal-attribute rule.
5. Mute the rough cut and fix any beat that doesn't make sense without sound.
6. Build the shot list, film the hooks separately, and cut the 15- and 6-second versions from the same footage.