How to brief a video agency: the 3 layers of a useful brief

How do you brief a video agency? Not by filling in an eight-heading template. A useful brief is written in three layers that have neither the same author nor the same shelf life: the foundation (offer, usable proof, prohibitions, rights, unit economics) that only the brand can supply; the hypothesis (angle, promise, audience) co-written with the agency and phrased so that it can be verified; and the execution (hook, script, format, specs) that must be left to the agency. A good brief locks down the business constraint and frees up the execution. Then it demands a counter-brief within 48 hours. According to the 35th Observatoire de l'e-pub by the SRI and UDECAM published in February 2026, Social grew by 15% in France in 2025, to €4.18 billion and 34% of the digital advertising market, driven by the rise of video.

Why do most video briefs fail even when they are well filled in?

A brief rarely fails for lack of information: it fails because the information is on the wrong side. The six-page document that lands on a Monday morning often contains a word-for-word script, three Pinterest references, an idea for an opening shot, the first name of the actor the client has in mind. And not a line on the product margin, the customer reviews that can be reused, what legal forbids saying, the usage period of the rushes. That brief is upside down. It dictates what we already know how to do and hides what we cannot guess.

The agency Vingt Deux, in its guide published in 2026, sums up the problem bluntly: a vague order generates “pointless back-and-forth, deliverables that miss the target and budgets that drift”. We agree with the diagnosis. We dispute the remedy proposed everywhere else, which is to make the document longer. A brief that is long on execution and empty on the business constraint is the worst of both worlds: it hobbles creativity and protects against nothing.

Potiok, in January 2026, points to the same mismatch from another angle: the question “live-action or motion design?” almost always comes too early, before the intention is clear. That is exactly it. The format is a consequence, not a starting point.

Another frequent confusion, and a costly one: briefing an agency has nothing to do with briefing an individual. A UGC creator receives a brief for one video, with their own rules on length and freedom, a subject covered separately in our UGC creator brief mistakes. An agency receives a framework for a series of decisions spread over weeks. You do not brief an executor, you brief a structure that will make calls in your absence. In the first half of 2025, the ARPP's Observatoire de l'Influence Responsable ranked non-certified creators at 51% compliant, 39% non-compliant and 10% improvable.

Hence the model we have used internally since 2025, after dropping flat templates.

The rule fits in one sentence. The brand writes layer 1, co-writes layer 2, and must not write layer 3.

Project brief or flow brief: what are you really ordering?

Before writing a line, decide: are you buying a finished deliverable, or a monthly flow of iterations? The two are briefed in opposite ways, and applying one to the other does predictable damage. A project brief applied to a paid social flow gives you twelve videos that look alike. A flow brief applied to a brand film gives you a fuzzy object nobody signs off.

The corporate film, the fundraising video, the 3-minute customer case study: these are projects. A start, an end, a final sign-off, a file you archive. The templates circulating on the web are made for that, and they work properly in that context.

Vertical video for acquisition is something else. The unit of purchase is no longer the video, it is the creative iteration: a testable variation of an angle, delivered in a batch, that lives for a few weeks and then dies. You do not brief an iteration, you brief the territory in which the agency will produce thirty of them.

A vocabulary detail that avoids misunderstandings: in industry practice, a distinction is made between the video call for tenders, meant to compare providers, and the contractual specification written after selection. This distinction between a call for tenders and a contractual specification is common industry practice; it prevents you from sharing your margin data with five agencies. This article deals with the second. We assume the agency has been chosen and the purchase order signed.

What information can only your brand provide?

Layer 1 contains everything no agency can invent, guess or buy: your offer, your proof, your prohibitions, your rights, your unit economics. It is the part that 9 briefs out of 10 leave out, and also the part that costs almost nothing to write once and for all. Two to three pages, dated, versioned, pulled out for every new partner.

The video specification: what it must contain, once and for all

• The real offer and its promotional mechanics. Displayed price, price after discount, delivery terms, offer duration. Not “we are premium”: the average basket and the repurchase frequency.
• Usable proof. Customer reviews you are allowed to quote on screen, figures validated by legal, certifications, lab tests, before/afters available with permission. Proof that is not cleared is useless proof.
• Prohibitions. Banned health claims, competitor comparisons, words the brand bans, claims already rejected by a regulator. In food supplements and cosmetics, this list often runs to half a page, and it saves two rounds of editing. The Ministry of Agriculture reminds us that the presentation and advertising of food supplements must not attribute to them properties of preventing, treating or curing a human disease, nor refer to such properties.
• Rights and usage period. Territories, media, duration, right to boost on a third-party account, extension. Treated as contract matter, so forgotten at scoping, when it determines the casting and the budget. We detailed the amounts in our article on real prices and usage rights in UGC video.
• Unit economics. Gross margin per order, acceptable CAC, average basket. An agency that knows your target CAC builds differently: more proof, less gratuitous aesthetics.
• Real targets, not personas. Who is already buying, according to your back office. Age, gender, reason for purchase, most frequent objection at customer service. The three most recurring objections are worth more than an illustrated persona.
• Resource constraints. Maximum budget, products available for the shoot and in what quantity, access to premises, availability of internal spokespeople.

A point you never read elsewhere: layer 1 does not need to be pretty. A shared spreadsheet does the job. What matters is that it is dated and that one person on your side owns it. When it changes, new price, new range, new legal framework, it is updated as version 2, and all agencies receive it.

How do you phrase a creative hypothesis rather than an objective?

An objective cannot be tested, a hypothesis can: replace “we want awareness” with “we bet that this audience buys if we remove this specific barrier”. Layer 2 is the only one that is co-written. The brand brings knowledge of the customer, the agency brings its reading of the platforms and of the angles that are running for other advertisers.

The template we use fits on four lines:

We bet that [specific audience] will buy [product] if we remove [identified barrier no. 1] through [angle]. We will test it on [media budget / number of iterations] over [duration]. The hypothesis is refuted if [numerical signal] is not reached.

The last line is the one nobody writes, and it is the most useful. It forces the question that no template on the market asks: what, in three weeks, will prove that this brief was wrong? Without an answer, you are not doing creative testing, you are doing decoration with a dashboard.

A hypothesis is not an angle, and confusing the two makes batches spin their wheels. The angle is the narrative mechanism that carries the promise, we listed the main families in our article on what a creative advertising angle really is. The hypothesis is the business bet that justifies testing this angle rather than another. An angle with no hypothesis behind it is an idea. Sometimes a pretty one. Not a learning one.

How many hypotheses per cycle? Our working rule: a single dominant hypothesis, two at most, per four-to-six-week period. Beyond that, media volume fragments and no signal emerges from the noise. What we cannot predict, honestly, is the exact length of the cycle: it depends on your daily media budget and your conversion volume. At €50 a day, forget about conclusions in ten days.

What must you absolutely not write in the brief?

The hook, the word-for-word script, the casting, the shot-by-shot breakdown, the music: all of that belongs to the agency, and writing it yourself transfers the risk onto you. It is the most uncomfortable position in this article, because it asks you to let go of the fun part of the job.

The case of the hook is the clearest. A hook is not decreed, it is measured: 3-second retention rate, hook rate, then comparison between variants on the same angle; the mechanics are detailed in our sheet on the definition and measurement of an advertising video hook. When a client imposes the opening line on us, they deprive us of the very object of the test. On our 2025-2026 projects, going by production memory, roughly one file in three arrives with an imposed hook, and in the majority of those cases the imposed hook is not the one that survives the first media trade-off. An important caveat: these are our internal observations, not a published statistic, and the sample is that of one studio, not of a market.

Same logic for visual references. Three examples to set a territory, fine. Fifteen links with shot-by-shot comments is an execution order in disguise. You pay the price of a creative agency to get an execution provider, and you keep responsibility for the result.

The formula circulating in the trade is the right one: tell the pilot where to go, not how to fly. We add a nuance it does not state. There are cases where imposing the execution is legitimate: the strict guidelines of a listed group, regulated video (health, finance, gambling), reuse of an already validated format that has been performing for six months. In those cases, say so in layer 1 as a constraint, not in layer 3 as an idea.

The only format question worth the brand deciding

Placements. Not the ratio, not the length: the placements where the video will run, because they depend on your media plan and your accounts, not on the agency's creativity. Say “Reels + Stories + TikTok In-Feed, and we want to be able to push to Pinterest in Q3”. The agency will deduce the framing and the safe zones.

How do you organise validation to avoid back-and-forth?

Schedule overruns almost always come from validation governance, never from the shoot: name a single decision-maker, contract the number of rounds and the feedback window. This is the complete blind spot of the briefs we receive. Ten pages on the message, zero lines on who says yes.

Four elements to put in black and white, in the brief and not in an email:

• A named decision-maker, with first name and job title. The others give advisory opinions. If three people have a veto, you have three briefs.
• A number of included feedback rounds: two for a batch of social creatives, sometimes three for a corporate film. Beyond that, it is billed. Say so beforehand, not when the invoice arrives.
• A feedback window: 48 or 72 working hours. After that, the deliverable is deemed approved or the delivery date slips by the same amount. Both options can be discussed; the absence of a rule cannot.
• A mandatory feedback format: timecode + problem + intention, never a bare judgement.

Our internal figures on this part, given as an order of magnitude and not as a market benchmark: when layer 1 is supplied before the first shoot, a batch comes out in 1 to 2 validation rounds. When it is missing, we regularly climb to 4 rounds or more, and the brief → first delivery lead time easily doubles. The cost of that gap is easy to reconstruct on your side: take your editor's day rate and multiply it by the number of extra rounds. It is the most profitable calculation you will make this quarter.

What is a counter-brief and why demand it within 48 hours?

The counter-brief is the document in which the agency restates your brief in its own words, puts figures on it, and challenges at least one point: without it, you do not know whether it understood or just said yes. None of the pages that dominate the search “how to brief a video agency” mention it. The brief is treated there as a one-way flow, client to agency. That is precisely why it fails so often.

What a counter-brief must contain, in one to two pages, delivered within 48 working hours:

• The hypothesis restated in one sentence, in the agency's words.
• The number of iterations delivered, the composition of the batch, the delivery date.
• The missing information that blocks production, and who must supply it.
• At least one reasoned disagreement. An angle it considers saturated, a target it thinks was badly chosen, a lead time it considers unrealistic.
• The measurement signal chosen, and the volume from which it becomes readable.

A counter-brief without disagreement is a red flag. Either the agency has not read it, or it has read it and does not dare, and in both cases it will let you be politely wrong. On our files, counter-briefs that change the initial angle are more the norm than the exception, often on one specific point: the target is too broad, or the central proof does not hold on screen. If you are at the selection stage, the 12 questions to ask before signing cover the earlier step; the counter-brief is the first deliverable after signing.

Why 48 hours and not a week? Because the counter-brief is not research work, it is a comprehension test. An agency that takes eight days to restate your bet does not have a Creative Strategist on your file, it has a project manager waiting for the schedule.

How do you evolve the brief after the first batch?

The brief is updated layer by layer, at different rhythms: execution moves with every batch, the hypothesis at the end of every cycle, the foundation once or twice a year. Confusing these rhythms produces the two classic pathologies of video management: the frozen brief pulled out for eighteen months, and the brief rewritten every Monday that prevents any reading of results.

After each cycle, three decisions are enough. The hypothesis is confirmed: you keep it, multiply iterations on the same angle, vary the hooks. The hypothesis is refuted: you archive it with its figure, write the next one, and do not replay it in six months hoping for better. The signal is unreadable: check the media volume first before blaming the creative, the diagnosis most often missed by advertisers who test on too small a budget.

Keep a written record of dead hypotheses. A simple table: date, hypothesis, number of iterations, observed signal, decision. After four cycles, this document is worth more than any downloaded brief template, because it is the only one that talks about your brand.

And if you take away only one trade-off from all of the above: learning how to brief a video agency is not learning to write more. It is accepting to write less, on fewer subjects, but the right ones, and letting the other side of the table do the work you pay them for. Brands that hold to this split get faster batches and cumulative learnings. Those that write the scripts themselves get exactly what they had in mind. Rarely better.

Frequently asked questions

How long should a brief for a video agency be?

Length is not the right criterion, the layer is. Count 2 to 3 pages for the business foundation, written once and reused, then one page per creative hypothesis. The execution is not your remit. A ten-page brief in which eight describe the desired shots is riskier than a well-distributed three-page brief.

What is the difference between a video call for tenders and a video specification?

The call for tenders is used to compare several providers: it describes the need, the context and the selection criteria, without confidential detail. The video specification comes after selection and becomes contractual: deliverables, deadlines, rights, validation rounds, governance. Arimedias already drew this distinction, and it remains useful so that you do not share your margin data with five agencies.

Should you state your budget in the video brief?

Yes, as a range, in the foundation layer. An agency that knows the ceiling builds a coherent setup; without this information, it proposes either something too expensive or a degraded format to protect its margin. Also give the planned media budget: it determines the number of readable iterations and therefore the size of the batch.

How do you avoid back-and-forth on a video project?

Three levers, in order of impact: a single decision-maker named in the brief, a contracted number of feedback rounds with a 48 or 72-hour response window, and a mandatory timecode feedback format. Add the list of claims validated by legal upstream: it is the first cause of late re-edits that we see.

Is the brief the same for ads and for organic content?

No, the two require opposite information. A creative ads brief needs the offer, usable proof and the measurement signal. An organic brief needs an editorial territory, recurring series and a publishing cadence. Imposing an offer and a CTA on organic content is the most common mistake with this type of set-up.

What if the agency challenges nothing in my brief?

Treat it as a negative signal. A counter-brief without a reasoned disagreement means nobody has confronted your bet with the reality of the platforms. Ask again explicitly: which angle seems saturated, which target seems too broad, which lead time is unrealistic. An agency with no opinion on your hypothesis will have none on the third batch either.