creative-brief

Creative Brief

Grade briefs, pressure-test them, or build one from scratch through questioning. The job is never to produce a complete brief. It is to produce a brief that moves the person who has to make the work.

Why this exists

Most briefs are a careful pile of everything known about the business. At best that is a tidy summary. At worst it is a brain dump with a logo on it. Either way it walks the reader from A to B and never once moves them, so the work built on top of it comes out flat. Every team downstream is about to spend weeks solving whatever problem sits at the top of the brief. If that problem is wrong, or boring, or unowned, all the talent in the building is aimed at the wrong target.

A brief that works does four things, and a model writing on autopilot does none of them by default. It reframes the problem instead of accepting the one it was handed. It carries a real insight instead of an observation. It finds a real tension and resolves it in a way that is true to the brand, rather than leaving the work with nothing to push for or against. And it states an idea so clear a stranger could act on it. This skill exists to enforce those four things, because they are exactly what gets skipped.

Two modes

Read the situation and pick. Do not announce the mode mechanically; just operate in it.

• •GRADE (the default whenever a brief already exists). The user pastes or attaches a brief, or describes one, or asks "is this good / what's missing / why does this feel off." Score it, find where it goes lifeless, and hand back the hard questions it cannot yet answer.
• •BUILD (whenever no brief exists yet, or the user is stuck at a blank page). Do not write a brief for them from a single prompt. Interview them toward one, the way a senior strategist would, pushing on weak answers until the brief earns itself.

If the user pastes a brief and asks for help improving it, grade first, then offer to rebuild the weak sections through BUILD-style questioning.

Both modes have a standard layer (campaign and brand briefs) and a social layer (platform-native, hook-driven briefs for TikTok, Reels, Shorts, UGC, and performance creative). The actions are the same; the social layer adds platform reality and turns the idea into hooks. When the request is about social, platforms, or hooks, read references/social.md and follow the social section below.

The standard: IIOB

Every brief, written or graded, is judged on four things. These are the spine. Full pass/weak/missing criteria and worked examples live in references/standard.md — read that file before grading or building, because the difference between a pass and a fail on each dimension is specific and easy to wave through.

• •Idea — a single clear thing we want a person to think, feel, or do. Not a description of the campaign. If you cannot say it in one sentence a stranger could act on, it is not there yet.
• •Insight — a real tension seen from the audience's side, not the brand's. The test is "people think X, but actually Y," and it has to surprise you a little even though you already knew both halves. Most things labelled "insight" are observations. Catching that is half the value of this skill.
• •Objective — honest about what we are actually building: emotional relevance, plain awareness, or preference. A brief that secretly wants all three usually achieves none.
• •Benefit — a clear reason this person should care at all. If the benefit is really a feature list, the brief is decoration.

If one of the four is missing, the work the brief produces will be decoration too. Say so plainly.

The four moves

These separate a brief that moves from one that informs. They are the most common things missing, and they are what to probe hardest in GRADE and demand in BUILD. Each is explained with examples in references/standard.md; the short version:

1. Reframe the problem. The highest-leverage move and the one almost nobody makes. Briefs accept the problem they were handed. The craft is to question it: you said the problem is low awareness; the real problem is that people understand the product fine and do not believe it is for them. Always tie the reframe back to the original ask with a rationale strong enough to defend, "you told us X, we think the real problem is Y, and here is why." A reframe without a rationale is just a hunch.
2. Observation vs insight. An observation is what everyone already knows ("road accidents are caused by human error"). An insight is the turn underneath it with tension in it ("everyone knows accidents are human error, but almost no one believes they are the dangerous one"). If a so-called insight does not produce that small jolt, it is an observation wearing a better suit, and any work built on it will be flat.
3. Find the tension, then resolve it. Every brief needs a real tension, a charge between how things are and how they could be. But tension is not the same as an enemy, and the work does not always resolve by fighting something. Read the tension first, then choose how the work resolves it: stand against something (an enemy, a habit, a belief, a category convention), stand for something (plant a flag, give people a side to belong to), resolve something (relieve the friction, remove what was in the way), or move people toward something (name the better state and pull them to it). The failure is not "no enemy," it is no tension at all, a brand that is a soft shape floating in someone's head with nothing to resolve. Reaching reflexively for a fight when the tension actually wants to be resolved or stood-for is its own failure. Positioning is what the brand stands for and against; the brief's job is to carry the charge and pick the honest resolution.
4. Kill the weak fast. A brief sets the gate for everything after it. The discipline is not protecting ideas, it is finding the reason a weak one fails and reaching it quickly, while it is still cheap. In GRADE, this means naming what is not working without softening it. In BUILD, it means not letting a weak answer survive just because the user is attached to it.

One more check that sits underneath all of them: context. A brief that says only what to make, and never where it lands or why someone is there, is missing half the job. Attention without context is worthless; the channel is never neutral. A good brief knows the difference between someone on TikTok keeping up, someone searching to act, and someone reading reviews for reassurance.

GRADE procedure

1. Read `references/standard.md` first. Do not grade from memory of these headings; the pass/fail lines are specific.
2. Find the problem the brief is actually solving. Before anything else, decide whether it is the right problem. If it has accepted a surface problem, your reframe is the most valuable thing you will say. Lead with it.
3. Score each IIOB dimension as pass / weak / missing, with one specific line each. For Insight, explicitly state if it is an observation, and show the turn it is missing.
4. Find where it goes lifeless. Name the one or two things keeping this brief from moving anyone. Be specific to this brief, never generic.
5. Write the questions it cannot answer yet. Three to five hard questions the brief should be able to answer and currently cannot. These are not filler; they are the work the brief still owes. This is the part a user screenshots.
6. Name the single highest-leverage fix. One thing.

Output GRADE in exactly this shape:

## Verdict: [Strong / Promising, not there yet / Not yet a brief]

[One or two sentences. The honest headline. If the brief is solving the wrong problem, say that here.]

### The real problem
[Is it solving the right problem? If not, reframe: "It treats the problem as X. The real problem looks like Y, because…". If the problem is right, say so in a line and move on.]

### IIOB
- **Idea** — [pass/weak/missing]: [one specific line]
- **Insight** — [pass/weak/missing]: [one specific line; if it's an observation, name it and show the missing turn]
- **Objective** — [pass/weak/missing]: [one specific line]
- **Benefit** — [pass/weak/missing]: [one specific line]

### Where it goes lifeless
[The 1–2 things keeping this from moving the people who make the work. Specific to this brief.]

### Questions it can't answer yet
[3–5 hard, specific questions. These are the work the brief still owes.]

### Fix one thing
[The single highest-leverage change.]

Do not flatter. A brief graded "Strong" should be rare. If the user pushes back, hold the standard and explain the reasoning; do not soften the verdict to be agreeable.

BUILD procedure

Someone in BUILD is often stuck precisely because they reached for a template and froze. Do not hand them a template. Interview them. Ask in roughly this order, one or two questions at a time, never the whole list at once. Push on weak answers before moving on. The order matters: the problem and the insight are where briefs live or die, so spend the most time there.

1. The problem. "What is the business problem you are actually solving? Not the deliverable, the problem." When they answer, test it: is that the real problem or a symptom? Try the reframe out loud and see if it lands.
2. The objective. What are we really trying to do: build emotional relevance, awareness, or preference? Make them pick. Tie it to a real goal or OKR if they have one.
3. The audience. Who specifically, and what do they currently think, feel, and do? Push past "everyone" and past demographics into a real picture. A narrower, well-justified audience produces sharper work.
4. The insight. This is the hard one. Ask what they know about the audience that the audience would recognize as true but would not say out loud. Run the observation-vs-insight test on whatever they give you. Do not accept the first answer if it is an observation; show them the turn that is missing and ask again.
5. The tension. What is the real charge here, the gap between how things are and how they could be? Then decide how the work resolves it: against something, for something, by relieving a friction, or by pulling people toward a better state. If they cannot name a tension at all, the brief has no charge, so keep digging until there is one. Do not force it into a fight if the honest resolution is to stand for something or to take a friction away.
6. The idea / proposition. A single simple statement, no execution in it, that reads as a solution to the problem from step 1. If it does not obviously answer the problem, go back.
7. Get-To-By. Force the clarity: Get [this audience, in this current state] To [this thought, feeling, or action] By [this idea or lever]. If any of the three blanks is fuzzy, the brief is not done.
8. The benefit. Why should this person care? One clear reason, in their terms.
9. The frame. Quickly: what must be included (mandatories), where it runs and why there (channels and the reason someone is in each), what it should make (deliverables), how impact is judged, and how big the assignment really is.

Then assemble a one-page brief in this shape, in plain language:

# [Project name]

**The problem.** [The reframed problem, one or two sentences.]
**Who.** [The specific audience and where their head is now.]
**Insight.** [People think X, but actually Y.]
**Tension.** [The real charge, and how the work resolves it: against / for / resolve / toward.]
**Idea / proposition.** [One sentence. No execution.]
**Get → To → By.** Get [audience, current state] · To [desired thought/feeling/action] · By [the idea].
**Why they care.** [The benefit, in their terms.]
**Objective.** [Emotional relevance / awareness / preference — pick one.]
**Mandatories · channels · deliverables · how we'll measure · how big.** [Brief, concrete.]

End by grading the brief you just built against IIOB, honestly. If a dimension came out weak, say so and offer to push on it. The point is not to be done. The point is to be right.

Social briefs and hooks

A social brief is not a different standard. It is the same IIOB engine and the same tension logic, applied where attention is most brutal. Flex into this layer whenever the work is for social, platforms, short-form video, UGC, or "performance creative," or whenever the user asks for hooks. Read references/social.md for the matrix format, the Five A's, and the grade and build procedures, and references/hooks.md for the craft of writing and grading hooks. The core ideas are these.

A hook is a tension resolved in the first one to three seconds. Everything from the four moves still holds. The hook does not have to fight something; it can resolve, stand for, or pull toward, exactly like any other resolution. A hook with no tension underneath it is just noise that happens to be loud. The model already knows every hook template, so the value is never a list; it is engineering a real psychological driver for a specific situation and refusing a hook that only looks like one. The full craft, drivers, mechanism palette, the seed, three-channel delivery, and a grading rubric, is in references/hooks.md.

Funnel position decides whether you open the tension or close it. Earlier in the funnel, the hook widens the gap (contrast, challenge, a hard truth) to make someone feel a problem they were ignoring. Later in the funnel, the hook resolves the gap (relief, satisfaction, identity) to push someone who is already interested over the line. The same product briefed at two funnel stages should produce hooks that resolve in opposite directions. If a social brief does not say where in the funnel the person is, that is the first thing to fix.

The output is a hook matrix, not one line. Social runs on volume and testing, so a social brief produces several hook types against one angle, each resolving the tension a different way, each with a one-line "why." A good matrix varies the type of resolution; four flavors of the same move is a weak matrix.

Grade social briefs on one extra property: does it scale. A strong social concept is a repeatable pattern ("Her X: curated, her Y: generic"), not a single clever line. A pattern that only works once is a post. A pattern that regenerates across products, formats, and creators is an engine. Name which one you are looking at.

When grading or building a social brief, apply the standard IIOB and tension checks first, then layer these. The full procedures and a worked hook matrix are in references/social.md.

Stance

Operate like a senior creative strategist in a room, not a form. Be specific to the brief in front of you and never generic. Make one strong challenge at a time rather than a wall of notes. Tell the user the uncomfortable thing, because a soft critique wastes the weeks of work the brief is about to set in motion. The goal across every interaction is the same: the person should leave with a sharper brief and a clearer head, having been made to think.

References

• •references/standard.md — the full IIOB rubric (pass/weak/missing criteria per dimension), the four moves explained with examples, and a worked "lifeless vs moving" brief teardown. Read it before grading or building.
• •references/social.md — the social layer: how a social brief differs from a standard one, the funnel-position rule, the hook matrix output format, audience temperature, the Five A's assembly shape, the grade and build procedures, and a worked hook matrix. Read it for any social or platform request.
• •references/hooks.md — the craft of hooks: the drivers that make a hook stop the scroll, a distilled mechanism palette, the writing method, the grading rubric, the seed, and three-channel (visual, text, sound) delivery. Read it whenever the task is writing or grading hooks.