How To Brief A Marketing Agency

Updated September 2026 · Written and maintained by the Progression Agency strategy team

Most weak agency work traces back to the brief, and most weak briefs specify outputs instead of stating a problem. This page covers the six things a brief must contain, why withholding budget produces proposals you cannot compare, how to describe an audience by behavior rather than demographics, the internal constraints agencies cannot see, agreeing the measure before work starts, what to leave out, and why a good agency will challenge what you wrote.

A brief states the problem, not the solution

The most common defect in agency briefs is that they specify outputs. Asking for six videos and a landing page tells an agency what to produce and nothing about what would count as success, which removes the judgment you are paying for.

A brief that states the problem — inquiries have fallen, a product is not understood, a new segment is not reaching us — allows an agency to propose something better than what you had in mind. A brief that specifies deliverables guarantees you get what you asked for, which is only useful if you were right.

The practical test: could a competent agency read your brief and propose an approach you had not considered? If the document only permits one answer, it is a purchase order rather than a brief.

The six things a brief must contain

Most briefs are missing at least two of these, and the omissions are predictable.

The business objective, expressed commercially rather than as a marketing metric. The audience, described specifically enough to exclude people. What you have tried and what happened. The constraints that are genuinely fixed — budget, timing, regulatory, brand. How success will be judged, agreed before work starts. And who decides, which is the one most often omitted and most often fatal.

That last item deserves emphasis. A brief that does not name the decision-maker produces work reviewed by a committee, revised toward consensus, and approved by nobody. It is the single largest predictor of a project going badly.

Objective, in commercial terms

Increase qualified inquiries from mid-market manufacturers is a brief. Increase brand awareness is a wish.

Audience, specific enough to exclude

If your audience description does not rule anybody out, it is not describing an audience.

What you have already tried

Withholding failures guarantees you pay to repeat them, and agencies frequently propose exactly what did not work last year.

Constraints that are actually fixed

Distinguish genuine constraints from preferences. Agencies design around real constraints and waste effort on imagined ones.

How success is judged

Agreed in advance and in writing. Agreeing it afterwards is how disputes start.

Who decides

One named person. Committees produce consensus work that nobody defends.

Budget belongs in the brief, and withholding it costs you

The instinct to withhold budget in order to see what agencies quote is understandable and counterproductive.

An agency without a budget figure guesses. Guess high and it proposes something you cannot afford; guess low and it proposes something too small to work. Either way you receive proposals that cannot be compared, because each is solving a different-sized problem.

Stating a range costs nothing in negotiating position. You are not committing to spend it, and an agency inclined to charge whatever you disclose will find another way. What you gain is proposals addressing the same problem at the same scale, which is the only basis on which comparison means anything.

Describe the audience by behavior, not demographics

Demographic descriptions are easy to write and rarely useful. Knowing that your audience is aged 35 to 54 tells an agency almost nothing actionable.

What is useful is behavioral and situational: what triggers them to start looking, what they search or ask, who else is involved in the decision, what they are worried about, what would make them choose a competitor, and what the alternative to buying anything is. Those shape targeting, message and channel in ways that age brackets do not.

If you do not know, say so, and treat finding out as part of the work. That is a legitimate and common starting point. Inventing a persona to fill the section is worse than leaving it blank, because the agency will design against a fiction.

Include the constraints an agency cannot see

Agencies propose things that are impossible for reasons nobody told them, then get blamed for it.

The invisible constraints are usually internal: legal or compliance review that adds weeks, a brand system that cannot be changed, a technology platform that will not support what is being proposed, a parent company with approval rights, a sales team who will not follow up leads in a particular format, or a busy season during which nothing can launch.

Every one of those is easy to state and expensive to discover in week six. A brief section listing what will slow things down is more valuable than most of the strategy content briefs usually contain.

Say how success will be judged, before work starts

Agreeing the measure afterwards is the most reliable route to a dispute, because both sides reconstruct their expectations in light of what happened.

State the primary measure, the timeframe in which it is fair to judge, and what would count as an acceptable outcome as distinct from a good one. Where attribution is genuinely difficult — brand work, long sales cycles — say so in the brief and agree the proxy measures you will use instead, rather than pretending to a precision that does not exist.

It is also worth stating what would cause you to stop. Naming the conditions for abandoning an approach makes it much easier to change direction later without it being read as failure.

Name the timeframe with the metric

A metric without a period attached is not a target, and the two sides will assume different ones.

Distinguish acceptable from good

Agencies will optimize toward whatever number you name. Naming only the ambitious version invites work that gambles.

Agree proxies where attribution is weak

Better to name imperfect measures deliberately than to argue about a precise one that was never obtainable.

State what would make you stop

Naming the exit conditions in advance removes most of the difficulty from changing course.

What not to put in a brief

Briefs are frequently padded with material that consumes reading time and contributes nothing.

Extensive company history, unless it explains a constraint. Detailed competitor analysis the agency will redo anyway. Mood boards presented as strategy. Aspirational language about being innovative or disruptive. Specified tactics when you have not diagnosed the problem. And a list of things you personally dislike, unless they are genuine brand rules rather than preferences.

The strongest briefs are short. Two pages that state the problem, the audience, the constraints, the measure and the decision-maker are more useful than twenty containing everything except those five things.

Brief the same document to everyone, and answer questions in the open

When several agencies are pitching, inconsistent information produces proposals that cannot be compared and quietly advantages whoever asked the most questions.

Issue the same document, hold a single question window, and share every question and answer with all participants. This is standard practice in formal procurement and rare in informal agency selection, where it is equally useful.

It also improves the proposals themselves. Agencies that know the questions are shared ask better ones, and seeing what others asked frequently reveals gaps in the brief that would otherwise surface after appointment.

Expect a good agency to challenge the brief

A brief is a starting position, not a specification, and an agency that accepts every premise without question is not adding the judgment you are buying.

Useful challenge sounds like: the objective is not achievable in that timeframe with that budget; the audience described is not the one that actually buys; the measure proposed will not capture the value; the constraint you listed is costing more than it protects. Any of those, raised before work starts, is worth more than the brief itself.

The response worth watching for is defensiveness on your side. A brief you are unwilling to have questioned is a specification, and you will get exactly what you wrote down, including whatever was wrong with it.

Reference videos

Marketing and measurement fundamentals relevant to the briefing points above.

Frequently asked questions