What to brief a video agency for performance creative

7 min readArtem Makarenko — Founder & Art Director, 2a4 Studio

A performance brief needs five things: the one claim the ad has to land, who specifically has to believe it, what the viewer is doing thirty seconds before they see it, the formats and markets it ships to, and what result would make you kill it. Everything else — tone words, references, mood boards — is optional. If you can only write this once, you have a one-concept month, and that is the real constraint on your account.

A performance brief is not a brand brief

A brand brief optimises for one asset that has to carry a lot: the positioning, the look, the feeling, all in a single film that will run for a year. It is written to be interpreted, and its success criteria are largely qualitative.

A performance brief optimises for something else entirely — a testable proposition that will run for maybe three weeks and then be replaced. It is written to be executed and measured, not interpreted. The difference matters because the two documents ask the agency for opposite things. One asks for a definitive answer. The other asks for a falsifiable guess, and then eleven more.

Teams that brief performance work like brand work get long lead times, high polish and too few assets. That combination cannot keep up with how fast paid social burns creative — about half of all creatives are retired before day 28.

What the brief has to contain

Five fields. If any of them is missing the agency will fill it in for you, usually wrongly, and you will find out three weeks later.

  • The single claim. One sentence the ad exists to make land. Not the product description — the argument. "Approvals take a day, not a fortnight" is a claim. "A modern spend-management platform" is a category.
  • The specific person. Not the persona document. The finance lead at a 40-person Dubai trading company who has been burned by a bank onboarding process. Specificity is what makes a script writable.
  • The context of the impression. What is the viewer doing in the feed thirty seconds before your ad appears? This decides pacing and opening far more than brand guidelines do.
  • Formats and markets, stated as separate lines. Vertical 9:16, square 1:1, 16:9 — and UAE and KSA as two productions, not one with a swapped voiceover. Local-dialect ads have been measured at 3.7 times the engagement of formal Arabic; a shared concept usually underperforms in both markets.
  • The kill condition. What number, at what point, means this concept is wrong? Writing it before launch is what stops a fatiguing ad from being defended for three extra weeks.

What to leave out

The most expensive thing in most briefs is the part that constrains execution without informing it. Shot-by-shot direction, locked-down runtime, a reference film to match — these convert a testable brief into a production order and remove the one thing you are paying an outside team for, which is a different guess than yours.

Leave out the full feature list. An ad that makes four claims makes none. If three features matter, that is three briefs, and testing them separately is the only way to learn which one moves the number.

Leave out the internal politics. If a claim is in the brief because a stakeholder wants it there rather than because a customer responds to it, it will still be there when the concept underperforms, and it will be the hardest part to remove.

The part everyone skips

Almost no brief we receive contains a kill condition, and almost every account we audit has an ad that should have been turned off a fortnight earlier. Those two facts are the same fact.

A kill condition is not pessimism, it is what makes the test a test. Frequency past 2.5 on prospecting, hook rate down 20-25% against its own first week, cost per lead past a stated ceiling — pick the one that matters for the concept and write the number down before the money is spent. Meta's own analysis puts conversion likelihood down roughly 45% by the fourth exposure, so the moment to decide is well before it feels urgent.

The corollary is a brief for what replaces it. If the answer to "what goes live when this dies" is a plan rather than a file, the account is already shrinking.

Why the brief is the real bottleneck

Production is no longer the slow step. Scripting, asset generation and voiceover run in parallel now, which is what makes a five-day cycle possible at all. What has not collapsed is the upstream work — deciding what twelve distinct things you believe about your own product are worth testing this month.

That is the constraint we hit with almost every fintech we work with. A team can name three arguments for their product. A volume plan at a roughly 5% win rate needs closer to twelve concepts a month to reliably surface a winner, and twelve concepts means twelve arguments. The gap between three and twelve is not a production problem an agency can solve by working faster.

It is also why every engagement here opens with an ICP and competitor pass rather than a camera. The brief is the deliverable that decides whether the other eleven are possible.

Assemble a performance video brief

  1. Write the claim as a sentence someone could disagree with

    If nobody could argue with it, it is a category description rather than a claim, and it will produce an ad that looks like everyone else's. Test the sentence by asking what the opposite position would be. If there isn't one, rewrite it.

  2. Name one person, not a segment

    Job title, company size, market, and the specific frustration this claim answers. Personas are built for alignment; scripts are built for one person. Hand over the person.

  3. Describe the scroll, not the storyboard

    What comes before your ad in that feed, and what is the viewer already tired of seeing from your category? This is the input that shapes the opening seconds, and it is the one an outside team cannot guess.

  4. Split the brief by market before you send it

    UAE, Saudi Arabia and Egypt are separate lines with separate registers, not one brief with a translation note. A concept written for both usually lands in neither, and you will read the result as a creative failure rather than a briefing one.

  5. Write the kill condition and the replacement

    State the number that ends this concept, and name what is briefed to replace it. A brief without both produces an ad that gets defended past its useful life, because nothing was ever agreed about when to stop.

Questions we get asked

How long should a performance creative brief be?

One page per concept. If it runs longer, it is usually carrying feature lists or internal positioning debates that belong in a different document. The five fields — claim, person, context, formats and markets, kill condition — fit comfortably in a page, and a page is short enough that writing twelve of them in a month is realistic.

Should the brief include a script or references?

References help, a script does not. A reference communicates register faster than adjectives can. A script hands the agency your guess and removes theirs, which defeats the point of briefing an outside team — you are buying a different premise, not just execution of your own.

Do we need a separate brief for each market?

Yes, if you are running the market seriously. Dialect, casting and pacing norms differ across the Gulf, Egypt and the Levant, and ads in local dialect have been measured at 3.7 times the engagement of formal Arabic. Treating KSA as a translation of UAE is the most common and most expensive shortcut in MENA paid social.

What if we cannot write twelve briefs a month?

Then twelve is not your volume, and it is better to know that before you buy production capacity. The honest sequence is to fix the upstream first: an ICP and competitor pass usually surfaces more distinct arguments than an internal workshop does, because it starts from what the market already responds to rather than from what the team already believes.

Sources

Keep reading