How long should a Meta video ad be?

These templates target 15 seconds, which suits prospecting and fits the sound-off, fast-scroll context most Meta video is viewed in. Longer cuts up to 30 seconds work for explanation-heavy or considered purchases, and the right way to extend is to stretch the evidence and result blocks rather than the hook.

Do I need a different script for Reels and Feed?

The script can be shared. The crop cannot. Produce 9:16, 4:5 and 1:1 of every concept, because placement selection is being removed from ad sets and you can no longer prevent an asset from appearing in a surface it was not cropped for.

How many script variants should I test at once?

Five variants of one template, changing only that template's open variable. Testing five different templates simultaneously changes several things at once and produces a winner you cannot attribute to anything reusable.

Does a talking-head ad need a written script?

It needs a structure and a first line. Word-for-word delivery usually reads as performance rather than as testimony, which defeats the point of the format. Write the opening and the ask, and leave the middle as bullet points the speaker can hit in their own words.

Which of these twelve templates performs best?

This article makes no such claim. Hit rates vary by offer, audience and category, and no controlled comparison of these twelve exists. The templates are there to generate genuinely distinct variants; your own account decides which earns budget.

What metric should decide the winner?

Hook rate as the gate, then link click-through rate, then conversion rate if your budget can reach a readable sample. Watch hold rate alongside hook rate: a high hook with a low hold means the opening over-promises relative to what follows, which is bait rather than a hook.

12 Meta Video Ad Script Templates, With Shot-by-Shot Timings

A Meta video ad script is four decisions, not a paragraph of copy: what is on screen at second zero, what claim carries seconds one to three, what evidence occupies the middle, and what the last two seconds ask for.

The twelve templates below fix three of those four and leave one open. That is what makes them testable. Swapping a template's open variable produces a comparison you can interpret. Swapping the background colour does not.

Every template targets a 15-second runtime, sound-off legibility, and a 9:16 or 4:5 frame. Production requirements are in the 9:16 production specs and the ad specs cheat sheet.

The structure every template follows

The 15-second Meta video ad timeline split into hook, evidence, result and ask blocks — The four-block structure underneath all twelve templates. Hook rate measures only the first block.

Timings are targets rather than rules. The ratio between the blocks matters more than the absolute seconds, and a 30-second cut simply stretches the middle two.

Each template below lists timing, shot, on-screen text and audio. The open variable line names the one element to change when you build variants. Hold everything else. Five variants of one template with five values of its open variable is a readable test. Five different templates at once is a shotgun.

The twelve templates

1. Problem-first

Time	Shot	On-screen text	Audio
0.0-0.5s	The failure state, mid-action	The problem in four words	Ambient, no music intro
0.5-3.0s	Same shot, closer	The consequence line	Direct address
3.0-9.0s	Product resolving the failure	The mechanism, one line	Explanation
9.0-13.0s	The result state	Outcome claim	Calm
13.0-15.0s	Product plus offer	CTA	CTA spoken

Open variable: which problem. Test three genuinely different pains, not three phrasings of one.

2. Direct offer

Time	Shot	On-screen text	Audio
0.0-1.0s	Product, full frame, moving	The offer, stated	Offer spoken immediately
1.0-5.0s	Product in use	What you get	Specifics
5.0-11.0s	Two or three feature beats	One line each	Fast
11.0-15.0s	Offer card	Deadline or condition	CTA

Open variable: the offer framing. Percentage off, currency off, bundle, or free add-on.

3. Confession

Time	Shot	On-screen text	Audio
0.0-2.0s	Face, handheld, close	I was wrong about...	First person
2.0-6.0s	Same	The prior belief	Unscripted cadence
6.0-12.0s	Cut to product	What changed	One specific detail
12.0-15.0s	Face again	Recommendation	CTA

Open variable: the admission. It has to be an objection your buyers actually hold, not a manufactured one.

4. Demonstration, no talking

Time	Shot	On-screen text	Audio
0.0-1.0s	Hands, product entering frame	None	Product sound only
1.0-10.0s	One unbroken use sequence	Two captions maximum	Natural sound
10.0-14.0s	Result, held	Result stated	Natural sound
14.0-15.0s	Product, CTA	CTA	Brief sting

Open variable: which single use case. One per ad, never two.

5. Objection handler

Time	Shot	On-screen text	Audio
0.0-2.5s	Face	The objection, verbatim	Quoted
2.5-5.0s	Face	Fair.	Acknowledgement
5.0-12.0s	Product evidence	The specific counter	Evidence
12.0-15.0s	Product	CTA	CTA

Open variable: which objection. Source them from your own ad comments and support tickets - see comments to creative.

6. Comparison

Time	Shot	On-screen text	Audio
0.0-1.5s	Split frame, both options visible	Before / After	Hook line
1.5-8.0s	Alternating, three beats	One differentiator per beat	Fast cuts
8.0-13.0s	Your option alone	The summary difference	Calm
13.0-15.0s	Product	CTA	CTA

Open variable: the axis of comparison. Time, cost, effort, or result.

7. Unboxing

Time	Shot	On-screen text	Audio
0.0-1.0s	Package already opening	Finally arrived	Natural sound
1.0-7.0s	Contents revealed one at a time	Name each item	Reaction
7.0-12.0s	First use	First impression	Unscripted
12.0-15.0s	Product assembled	CTA	CTA

Open variable: the reveal order. Lead with the item that best differentiates, not the largest one.

8. Founder explanation

Time	Shot	On-screen text	Audio
0.0-2.0s	Founder, workspace visible	We built this because...	Direct
2.0-7.0s	Founder	The gap in the market	Specific
7.0-12.0s	Product, handled by founder	The design decision	Detail
12.0-15.0s	Founder	CTA	Personal CTA

Open variable: which design decision. More in founder-led selfie video.

9. Listicle

Time	Shot	On-screen text	Audio
0.0-1.5s	Product, count on screen	3 things nobody tells you about X	Hook
1.5-5.0s	Beat one	1. plus the point	Fast
5.0-9.0s	Beat two	2. plus the point	Fast
9.0-13.0s	Beat three	3. plus the point	Fast
13.0-15.0s	Product	CTA	CTA

Open variable: the count and the category. Three is completable in 15 seconds; five is not.

10. Testimonial stitch

Time	Shot	On-screen text	Audio
0.0-2.0s	Customer face, mid-sentence	The strongest quote fragment	Customer voice
2.0-8.0s	Two more customers, cut together	Each quote as caption	Customer voices
8.0-12.0s	Product	The common thread	Narrator
12.0-15.0s	Product	CTA	CTA

Open variable: which benefit the quotes share. All three have to support one claim, or the ad makes none.

11. Process reveal

Time	Shot	On-screen text	Audio
0.0-1.5s	Raw material or first step	How it is actually made	Ambient
1.5-10.0s	Four process beats	Name each step	Ambient, light narration
10.0-13.0s	Finished product	The quality claim the process earns	Narration
13.0-15.0s	Product	CTA	CTA

Open variable: which step to dwell on. Choose the one a competitor cannot replicate.

12. Category education

Time	Shot	On-screen text	Audio
0.0-2.0s	Text card over product motion	The question the buyer asks	Question spoken
2.0-9.0s	Explanatory visuals	The answer in three parts	Teaching tone
9.0-13.0s	Product as the applied answer	Where the product fits	Calm
13.0-15.0s	Product	CTA	Soft CTA

Open variable: the question. Pull it from search queries and support volume, not from imagination.

Production rules that apply to all twelve

These are not style preferences. Each one is a constraint imposed by how the ad is actually viewed:

1.Legible with sound off. Assume no audio. Captions are load-bearing, not decorative.
2.Respect the safe margins. Keep text clear of the top and bottom of a 9:16 frame where interface elements sit.
3.Produce every ratio. Ship 9:16, 4:5 and 1:1 per concept. Placement selection is being removed from ad sets, so the crop is now the only reliable control over how an asset renders - see [the placement controls change](/blog/meta-placement-controls-removed-value-rules-2026).
4.No ramp. Second zero must already contain the subject. A logo card is a second of paid silence.
5.One idea per ad. Two claims in 15 seconds means neither lands.
6.Rights before delivery. Music, voice and likeness need licences that cover paid social specifically, which is a narrower grant than general commercial use.

On that last point, voiceover and music licensing covers the clearance question, creator whitelisting covers running a creator's footage from their handle, and Meta disclosure rules covers what has to be declared.

How to test these without wasting budget

Pick one template. Build five variants changing only its open variable. Hold the offer, the audience, the landing page and the runtime constant.

Judge on hook rate first. Attention metrics reach a readable sample far sooner than conversion metrics, because the denominator is impressions rather than clicks - at a $15 CPM, tens of thousands of impressions are inexpensive, while tens of thousands of clicks are not. Only promote to a conversion judgment what clears the attention gate. The sample-size arithmetic is in the creative testing budget calculator, and what counts as a good hook rate sets the gate.

Then apply the guardrail that catches the most common self-inflicted failure: a high hook rate with a low hold rate means the opening over-promises relative to what follows. That is bait, not a hook, and scaling it buys attention you cannot convert.

What these templates do not claim

No hit rate is attached to any template here, and none should be inferred from the order they appear in. Attaching one would require a controlled comparison of all twelve against a shared offer and audience, which this article does not have.

Published creative benchmarks report hit rates by hook category, and those are useful as hypothesis generators rather than as rankings: ads are not randomly assigned to hook types, business models or offers, so the comparison measures the advertisers as much as the format. Use the twelve structures to generate genuinely distinct variants, and let your own account decide which one earns budget.