00:00 — 00:30
The anatomy of a short-form video ad, second by second
Storyboard a short-form video ad beat by beat: build a hook, show the problem, demonstrate proof and make one clear ask, with timing examples.
You do not need a better idea before you can make a better short-form ad. You need a sequence that makes a stranger understand the offer, see it work, and know what to do next. This is a beat-by-beat storyboard for a founder shooting a first vertical ad, with enough room to change the timing when the product demands it.
Why structure beats inspiration#
In a feed, the opening has a job before your explanation gets a hearing. TikTok advises introducing the proposition in the first three seconds and prioritizing the hook in the first six; Meta's Instagram video guidance says to show the brand and key message within the first three seconds.[1][2] These are platform recommendations, not a stopwatch that measures the exact moment every viewer decides. The practical consequence is simpler: if the opening can be removed without changing what the viewer knows, it is probably throat-clearing.
TikTok describes a hook, selling points and a clear call to action; its Creative Codes call the same broad progression hook, body and close.[1][3] We split that body into problem and proof. A problem tells the viewer why to care; proof lets them decide whether the promised change is real. The split prevents a familiar editing failure: an ad that spends its middle describing inconvenience and reaches the actual product only when the viewer is gone.
Google's current ABCD guidance says Attention, Branding, Connection and Direction: enter the story quickly, introduce the product early and ask for an action.[4] Its underlying reference guide analyzed TrueView in-stream ads, not Shorts, and says the findings are correlational rather than a performance guarantee.[5] Treat the idea of early product recognition as useful production discipline, not as evidence that a rigid Shorts script will outperform every alternative. Google also publishes separate Shorts guidance that asks for a vertical, social-first asset.[6]
An older TikTok Creative Codes article says that much of recall impact arrived in the first six seconds; the cited analysis behind that statement dates to 2020.[3] That history is a reason to give the opening attention, not a current universal recall forecast. Build the first cut around a testable chain: what is this, why should I care, can I see it work, and what happens if I act?
The four beats, with room to breathe#
For an illustrative thirty-second cut, start with a visible result or friction, establish the context, spend the middle demonstrating the answer, then make one request. The durations below are editorial choices for a fictional example, not platform-required beat lengths. The markers show TikTok's distinct first-three-second proposition and first-six-second hook guidance, rather than implying the hook must always end at either mark.[1]
- 1Hook 0–3sShow result
- 2Problem 3–9sName friction
- 3Proof 9–24sShow the change
- 4Ask 24–30sOne next step
These are functions, not obligatory title cards. A product in the first frame can be both the hook and the start of the proof. A spoken customer's problem can set context while a demonstration is already underway. On the other hand, if each segment needs a new scene, the cut may become a sequence of labels instead of a convincing moment. Make a storyboard with the picture on one line, the claim on another, and the evidence you can actually show on a third.
Read the timeline backward from the action. If the destination is a product page, what must someone believe before they tap? Perhaps they need to see a messy task become shorter; perhaps they need to understand fit or size. That determines the proof shot. Then choose the first frame that creates appetite for that proof. This reverse pass is especially helpful when a founder has a folder of attractive footage but no argument connecting it.
Hooks that earn the next second#
A hook is an opening promise about the rest of the video. It does not have to shout, conceal the product, or manufacture a mystery. TikTok asks for the proposition early; Meta asks for a compelling visual or motion in the first frame; Google's Shorts guidance says an ad should feel native to the feed rather than like a cropped television spot.[1][2][6] The hook should therefore work as a picture even before anyone hears its first line.
The following first frames and lines are fictional creative examples. Their failure modes matter more than the wording: pick an opening you can pay off with visible evidence.
| Pattern | First frame | First line | When it fails |
|---|---|---|---|
| Viewer question | Steam over a crowded counter | “Where does the water go?” | The answer is delayed just to prolong curiosity. |
| Result first | Finished cup beside the kettle | “Here is the pour I wanted.” | The result is prettier than the product can deliver. |
| Pattern interrupt | Hand pauses an uneven stream | “That wobble was the problem.” | The interruption has nothing to do with the offer. |
| Product in hand | Spout enters frame immediately | “Watch this angle.” | Viewers cannot tell what the product is or why it matters. |
| Correct a true mistake | Water spills from a wide spout | “I was pouring too fast.” | The line blames a user for a problem the product cannot solve. |
| A number | Price tag beside the real item | “This one is $39.” | The price is inaccurate or hides a condition. |
| Before-state | Damp counter and interrupted pour | “My old setup dripped here.” | The before footage is staged to exaggerate an outcome. |
| Point of view | Hands reach for the handle | “Here is what the grip feels like.” | The camera hides the action it claims to reveal. |
You can replace the kettle with software: show the actual screen and the specific moment someone stops hunting for a control. Do not open with a logo animation that postpones that moment. Google recommends introducing product or brand early in its ABCD material, but the in-stream evidence caveat still applies; in a Shorts cut, a usable product shot can carry recognition without a separate brand bumper.[4][5]
Draft several openings from the same real proof, then change only the first frame and line. That gives you a meaningful creative comparison later. If you change the claim, audience, landing page and opening together, even a stronger result will not tell you which decision helped. The metrics guide explains what an opening-rate comparison can and cannot establish.
Proof is the ad#
After the hook, the viewer is owed an answer. A shot of a hand actually using a product says more about its shape, resistance and result than a row of adjectives. A screen capture can show the path through a workflow; a product demonstration can show the action without claiming a measurable outcome it cannot support. When the benefit is invisible, build a comparison the viewer can inspect: the same task under the same conditions, with the relevant step clearly visible.
Separate three things in the script: observation (“the stream stays inside this cup in this take”), claim (“the spout gives me control”), and generalization (“this will always prevent spills”). The first is visible, the second is an interpretation you may be able to substantiate, and the third needs much more evidence. The FTC says advertisers should have a reasonable basis for objective claims before disseminating them; customer enthusiasm by itself is not scientific substantiation for a performance claim.[7][8] Write down what your shot actually proves before writing the super on top of it.
Use a customer's words only when they are genuinely that customer's experience and you have permission to use them. Under the FTC Endorsement Guides, performers presented as actual consumers should be actual consumers or clearly identified otherwise; the FTC's consumer-review rule also addresses testimonials that misrepresent whether the person exists or used the product.[8][9] If an actor reads an approved script, call the performance what it is. The guide to AI avatars and UGC-style ads handles the more complicated disclosure and likeness decisions.
Proof also has a visual hierarchy. Frame the hand and the outcome before adding kinetic captions. Show the full motion rather than cutting away at the point of difficulty. If the selling point is a quieter operation, record intelligible sound but let the silent picture still communicate what changes. If the value is software speed, do not fake a timer; show the actual sequence and trim only dead time, with an honest label if the capture has been accelerated.
You do not need a customer quote when you can film the product plainly. Often the most defensible middle beat is a single coherent demonstration with an unobstructed beginning and end. Put supporting claims next to the action they describe, not on a closing slate the audience has to remember separately. That placement makes the proof easier to judge and gives you a concrete revision target if viewers leave mid-cut.
One ask, then adapt the runtime#
An ending should request one next step that follows the proof: see the product details, compare a configuration, or start a trial. A spoken ask, visible text and the platform's CTA affordance can reinforce the same destination without naming three different actions. TikTok's creative guidance explicitly calls for a clear CTA, and Google's ABCD guidance recommends reinforcing an on-screen ask with voice-over.[1][4] Put the answer to “why tap?” ahead of the tap instruction.
On Shorts, the clickable CTA button is not necessarily there when the video starts. Google's asset guidance says it appears after three seconds for Performance Max, App and Demand Gen, and after ten seconds for Video View and Video Reach campaigns.[6] Your first frame must therefore carry its own meaning. Do not rely on the button to explain the offer, and do not put a fake on-screen button over a region already used by platform controls. TikTok's ad policy specifically objects to instructions such as “swipe up to learn more” when that gesture simply advances the feed.[10]
A short cut is not just a thirty-second cut played faster. These illustrative, non-platform beat allocations show how the relative weight can change while the underlying argument survives:
- 1Hook 0–2s
- 2Context 2–4s
- 3Proof 4–12s
- 4Ask 12–15s
| Planned cut | Editorial use | Platform length to check before export |
|---|---|---|
| 15 seconds | One problem, one clear demonstration, one ask. | TikTok reservation/TopView recommend 9–15 seconds and specify 5–60; the auction spec and policy conflict.[11][10][12] |
| 30 seconds | More room to show setup and inspect the proof. | Google recommends 10–30 seconds for action-oriented Shorts ads; Instagram Reels ad specs allow much longer.[6][13] |
| 60 seconds | A more involved setup or comparison, if every beat earns its time. | Google lists 6–60 seconds for Video Reach and 10–60 for other Shorts campaign types; TikTok policy says a maximum of 60 seconds.[6][10] |
Here is the TikTok conflict in full: its current auction in-feed specification says non-Spark video ads can run up to ten minutes, while its advertising policy says five to sixty seconds; reservation in-feed and TopView list five to sixty seconds and recommend nine to fifteen.[12][10][11] Spark duration wording also differs between the auction and Spark Ads help pages.[12][14] Do not treat one of these pages as a universal format limit. Use the stricter policy when planning a portable paid cut, then inspect the actual placement and approval path in Ads Manager before producing a longer one. Google, too, has conflicting Shorts help pages about what plays after sixty seconds; for an action-oriented ad its shorter recommendation is the useful brief, not a promise about every playback surface.[6][15]
Sound on, sound off: write for both#
“Everyone watches muted” sounds decisive and is poorly scoped. The often repeated 85% sound-off claim came from publishers describing Facebook News Feed video in 2016; it was not a current platform measurement of Reels ads.[16] Meta later reported that more than 75% of Instagram Reels views were sound on, in a 2024 article without a published methodology for that figure. Its Instagram video best-practices page nevertheless says most users watch with sound off.[17][2] Those statements differ in surface, date or method, and the pages do not resolve the disagreement. A single sound-on share cannot plan your edit.
Make the picture carry the causal change: show what is wrong, what moves, and what is different afterward. Give the voice-over the nuance the picture cannot carry, such as a qualification or a reason to choose the product. A silent pass should still let someone name the offer and understand the next step; a sound-on pass should reward attention with a clearer explanation. TikTok's policy requires audio in the ad, while Meta's Instagram Reels spec lists sound as optional but strongly recommended.[10][13] Designing a silent-readable cut does not mean uploading a silent TikTok ad.
Captions should support the actual speech, not cover the demonstration. Put them where interface chrome cannot hide them and check their timing against a phone-sized playback. The caption guide covers readable line lengths, sound cues and contrast; the safe-zone guide gives the platform overlays and a cross-platform working box. The opening image and on-screen price need the same placement check as captions.
Worked example: a fictional kettle cut#
Here is a fictional $39 pour-over kettle, a made-up offer for demonstrating the storyboard. The price, shots and product behavior below are invented; they are not claims about a real kettle. If you adapt the script, verify your own price, availability and performance before publishing. The sequence deliberately shows the pour instead of claiming it improves taste or guarantees a result.
| Time | Picture | On-screen text | Voice |
|---|---|---|---|
| 0–3 s | A hand pauses a splash over a cup; the kettle enters frame. | “A steadier pour?” | “I wanted to control where the water went.” |
| 3–9 s | The old wide-spout pour reaches the counter; cut to the kettle's handle and spout. | “Wide spout / narrow spout” | “My old setup made the stream hard for me to place.” |
| 9–15 s | One continuous overhead pour begins. Keep the cup edge visible. | “Watch the stream” | “Here is the pour with this kettle.” |
| 15–24 s | Continue the same action from a side angle, then show the counter and cup. | “See it in use” | “I can aim it at the part of the cup I want.” |
| 24–30 s | Product and price on a clean surface; show the actual destination on a phone mock. | “$39 · See the details” | “Take a closer look at the kettle and its dimensions.” |
The hook does not conceal the object for a reveal. The problem is the speaker's own difficulty rather than a universal allegation about wide spouts. The proof occupies the middle, with a continuous action so the editor cannot create the apparent result by omission. If the actual product does not produce this shot under ordinary use, change the script; do not fix the discrepancy with a tighter crop.
Now cut it to the shorter timeline. Keep the opening image, one problem line, the uninterrupted proof, and the ask. Remove the second angle and the explanatory adjective, not the evidence. For a longer cut, add a useful comparison or a handling detail that helps a buyer decide, rather than slowing the existing pour. This is how duration follows the amount of proof you can show.
From brief to cut#
Write a one-sentence brief before opening the editor: “Show [person] how [product] handles [specific problem] by filming [observable proof], then invite [one action].” List what must be true for every clause. A claim without a shot or another supportable source should be rewritten. A shot without a role in the argument should earn its place or leave the cut.
Shoot the proof first, because the opening should promise only what you have footage to deliver. Film a clean wide version of the action and a closer version that preserves context. Capture room tone and a spoken explanation separately if needed. Then make opening alternatives from footage you own. Azelify can help draft the storyboard and cut; you still decide whether the action, claim and destination are accurate.
At the phone check, watch the first frame without waiting for the voice, pause on the price, and let the CTA play with actual interface chrome in view. Verify captions against the spoken line and the landing page against the price and promise. A storyboard is finished when the viewer can track the argument at playback speed, not when every frame looks busy.
Common mistakes#
- Starting with a logo bumper. It delays the proposition. Put the product or outcome in view and let the brand be legible inside the action.
- Replacing proof with praise. “Better” is an assertion; a continuous demonstration gives the viewer something to inspect.
- Treating every platform's duration as identical. Auction, reservation, policy and Shorts help pages differ. Choose the actual placement before final export.[12][11][6]
- Making three asks. A discount, a follow request and a shop instruction compete for the same closing moment. Pick the action your destination supports.
- Using a testimonial as a shortcut to a performance claim. The FTC does not treat customer endorsement alone as competent scientific evidence for claims that need it.[8]
- Judging only on a desktop preview. Captions and UI rails can hide the product or the ask; test the real crop on a phone.[18]
Checklist#
- State the audience, offer, supportable claim and single next action in one brief.
- Show the product or meaningful result immediately; make the opening pay off with proof.
- Film the actual action in a shot a viewer can inspect, without a misleading cut.
- Give the problem only as much runtime as the viewer needs to understand the demonstration.
- Match every on-screen price, comparison and customer quote to evidence and permission.
- Select the paid placement and confirm its current length and audio requirements.
- Watch the exported cut muted and with sound, in the platform preview and on a phone.
- Make the closing text, voice and destination ask for the same action.
Keep reading
Safe zones for TikTok, Reels and Shorts: where your text actually survives
Compare official TikTok, Reels and Shorts safe-area templates in pixels, find the shared box for text, and understand why published measurements differ.
Hook rate, hold rate and ThruPlay: the short-form ad metrics that matter
Separate platform-defined video metrics from practitioner ratios, calculate a fictional ad funnel, and diagnose whether viewers stop, stay or act.