A creative brief for video ads needs five fields and should fit on one screen: the offer, the audience and their objection, the angle, the constraints, and the format. Every field beyond those five is either restating strategy that belongs in a deck or specifying craft decisions the editor or the generator should make. Long briefs get skimmed, and a skimmed brief is worse than a short one because everyone assumes it was read.
The test of a brief is not whether it is thorough. It is whether an output can fail review on a ground the brief did not cover. If it can, the brief was too short in the places that mattered and too long everywhere else.
The five fields, and what breaks when each is blank
| Field | What it must contain | Failure mode when left blank |
|---|---|---|
| Offer | The specific thing the viewer gets and any condition on it | Ads with no reason to act, killed in review for being "flat" with no fixable note |
| Audience and objection | Who it is for and the one sentence they would use to say no | Generic hooks that address nobody, so hook rate collapses and nobody can say why |
| Angle | The single claim this batch argues | Every output argues something different, so the test compares nothing |
| Constraints | Banned claims, required disclaimers, brand rules, ratio list | Outputs fail on rules nobody wrote down, and your keep rate drops with no lesson learned |
| Format | Length, aspect ratios, placement, language, caption policy | Rework at export, ads that do not fit the placement they were bought for |
The pattern to notice: four of the five failure modes are review failures, not production failures. That is the point. The brief is not documentation. It is the thing that decides how much of what you produce you can actually publish.
Offer
Write the offer as a sentence a customer would repeat. "30 percent off the starter bundle until Sunday" is an offer. "20 minutes back per shift" is an offer. Free trial length, shipping terms, guarantee window: all offers.
"Learn more" is not an offer, and it is the single most common reason a batch comes back weak. It gives the script nothing to be specific about, so the writer or the model reaches for adjectives, and adjectives are what a reviewer rejects when they say an ad feels generic. If the honest answer is that there is no offer yet, that is a strategy problem being laundered through a creative brief, and it will not be fixed by producing more ads.
Audience and their objection
One line on who, one line on the objection. Not a persona document. "Ops managers at 50 to 200 person manufacturers" plus "we already have a system and switching sounds painful" is enough to produce a dozen distinct hooks.
The objection is the field people skip and the one that changes hooks most. A hook that names the objection out loud qualifies the viewer in the first two seconds, which is the only job the first two seconds has.
Angle
The angle is the one claim this batch argues: speed, price, ease, risk reversal, social proof, a specific use case. One angle per brief. If you want to test three angles, write three briefs.
This is what makes a batch readable as a test. Vary the hook within one angle and you learn which hook works. Vary the angle and the hook together and you learn nothing you can act on next week.
Constraints
Everything that would cause a rejection, written down before production. Banned claims and comparatives, required disclaimers, whether the logo must appear and when, music licensing, whether a competitor can be named, the aspect ratios you need, the language.
This field is where the money is. Constraints discovered during review are rework; constraints stated in the brief are just instructions. The section below is about why that distinction is worth arguing for.
Format
Length in seconds, aspect ratios, placement, language, caption policy. Be specific about ratios, because "make it vertical" gets you one file when you needed four. The usual set is 9:16 for Reels, Stories and TikTok, 4:5 for Meta feed, 1:1 where it still appears, 16:9 for YouTube in-stream.
Why compliance belongs in the brief, not in review
Most teams treat brand and legal as a gate after production. It is the most expensive place to put it. An output that fails a compliance check is a fully produced, fully reviewed asset that gets binned, and every hour in it is sunk.
Move the same rules to the brief and they cost nothing. This is the mechanism behind keep rate, meaning published divided by produced, which is the number that governs your true cost per ad. If your keep rate is 60 percent, shipping 40 ads means producing 67, and every hour of brief, production and review time scales off 67 rather than 40. Constraint-first briefing is the cheapest way to move that ratio, and keep rate is worth measuring before you change anything else.
A practical way to do it: keep a standing constraints block that never changes, and paste it into every brief. Legal signs the block once a quarter instead of signing ads one at a time.
A filled brief
For a fictional B2B scheduling product, run as a weekly test.
Offer. 14-day trial, no card, imports your existing rota from a spreadsheet on day one.
Audience and objection. Operations managers at 50 to 200 person manufacturers. Their objection: "we already run this on spreadsheets and moving would break the week."
Angle. Migration risk reversal. This batch argues that switching costs you one afternoon, not one quarter.
Constraints. No claims about hours saved or cost reduction without the customer-study footnote. Do not name competitors. Logo in the final two seconds only. Captions burned in, sentence case. Deliver 9:16 and 4:5. English (UK) only this month.
Format. 15 to 20 seconds. Hook must name the spreadsheet objection in the first three seconds. One product screen recording minimum. Voiceover plus captions.
That is around 130 words and it is enough to produce ten to fourteen distinct outputs that all belong to the same test. Two of the fields (constraints, format) are effectively boilerplate after the first week, so the real weekly writing job is offer, objection and angle, which takes about ten minutes once the pattern is set.
If you are briefing a tool rather than a person, the same five fields map onto the inputs directly. Our guide to setting up your first campaign walks through where each field goes, and our AI ad script generator takes offer, objection and angle as the inputs that actually change the script. Genyad writes scripts natively in English, German, French, Spanish, Italian and Hindi rather than translating them, which is why the language line belongs in the brief rather than in a post-production pass.
What to leave out
Things that look like brief fields and are not:
- Reference ads. A competitor's ad in the brief produces a copy of that ad. If you want a structural pattern, name the structure ("problem, demo, offer"), not the ad.
- Music direction, unless it is a licensing constraint. It is a craft decision, and it does not survive being briefed.
- Shot-by-shot direction. If you are specifying shots, you are editing in a document. Name the shot types you need and let production choose the takes. How our script writing works assumes shot selection follows the script rather than preceding it.
- Tone words. "Bold, warm, confident" is unfalsifiable, so it becomes a licence for taste-based rejection in review, which is exactly what constraints exist to prevent.
Frequently asked questions
How long should a video ad creative brief be?
One screen, roughly 100 to 200 words across five fields: offer, audience and objection, angle, constraints, format. Anything longer gets skimmed, and a brief that is assumed to have been read is more dangerous than a short one. The constraints and format fields become boilerplate after the first week, so the weekly writing is about ten minutes.
What makes a creative brief produce weak ads?
An offer of "learn more" and a missing objection, in that order. Without a concrete offer the script has nothing specific to say, so it fills the gap with adjectives that reviewers reject as generic. Without the objection, hooks address nobody and hook rate suffers.
Should one brief cover multiple angles?
No. One angle per brief, because a batch that varies both angle and hook produces results you cannot attribute to either. If you have three angles worth testing, write three briefs and run them as three sets, then compare at the set level.
Where do brand and legal rules belong?
In the constraints field of the brief, not in a review gate afterwards. A rule stated up front is an instruction that costs nothing, while the same rule applied in review bins a finished asset and wastes every hour in it. Keep a standing constraints block that legal approves once a quarter and paste it into every brief.