How to write a creative brief that produces a 78, not a 42

A creator handed a product link and a deadline will make creative decisions — hook formula, proof element, CTA placement — based on whatever felt right in the moment. Sometimes that produces a strong ad. More often it produces an ad that scores well on style and poorly on clarity, because no one told the creator what problem the product actually solves or who it solves it for.

The brief is where strategy gets translated into production direction. Without it, every decision made on set or in the edit is a guess. A 42-scoring ad and a 78-scoring ad often have similar production quality — the gap is that the 78 had a clear hook direction, a specific proof element, and an exact CTA, and the 42 had none of those things written down before filming started.

The Ad Bench scores reflect what the brief produced. If the rubric flags weak hook, low clarity, or missing proof, the brief is where you fix it — not in a re-edit, not in a retake, but before the brief leaves your desk.

A complete brief has exactly six elements. Each one maps to a dimension the scoring rubric evaluates. (1) A one-sentence product promise: what does it do, for whom, and in how long? Not a tagline — a plain-language description that the creator can turn into a hook. (2) The primary objection to overcome: why wouldn't someone buy? Name it explicitly so the creator can address it without being asked.

(3) Hook direction: which of the six hook formulas, and a sample first line. Don't leave hook selection to the creator — specify “curiosity gap, first line: ‘I didn't think this would actually work until…’” or equivalent. (4) The proof element to include: a specific number, a before/after, or an attributed testimonial. Vague proof (“results may vary”) scores worse than specific proof (“12,000 sold in 30 days”). (5) The CTA in exact words — not “include a call to action” but “say ‘link in bio, limited to 200’ at second 12.” (6) Platform and format: 9:16, 15–30s, TikTok-native, sound-off-ready. Each platform has different constraints; specify them so you don't get a Reels-aesthetic cut submitted for a TikTok placement.

Six elements. If any one is missing, the brief is incomplete and the creator will guess. They will guess wrong at least half the time, and “half the time” on a paid creative brief is an expensive error rate.

Brand history and company mission do not belong in a creative brief. The creator is making a 15-second ad, not a brand film. Company founding stories and mission statements consume brief real estate that should be occupied by hook direction and proof elements. If the creator needs brand context, give them a separate brand voice document — not a brief that buries the actionable instructions under three paragraphs of “we were founded in 2019 because we believed…”

Long lists of approved and forbidden words belong in a brand voice document, not a brief. A brief that runs to four pages because it includes a 40-word approved vocabulary and a 30-word forbidden vocabulary will not be read carefully. The six required elements will be skimmed. Move word lists to a separate reference doc and link to it.

Multiple CTAs kill scores. One brief, one CTA. If you want a soft CTA mid-roll and a hard CTA at the close, you have written two briefs and tried to combine them — split them. Vague adjectives like “authentic” and “engaging” tell a creator nothing they can act on. Describe the behavior you want: “hand-held camera, no ring light, natural background” says the same thing as “authentic” and actually directs the shoot.

If you want three hook variants — which you should, per the A/B testing framework — write three explicit hook directions in the brief. Do not leave variant generation to the creator. A creator asked to “try a few different hooks” will produce three versions of the same formula with different first words. That is not variant testing; that is one idea with three intros.

A properly briefed variant set specifies each formula by name: hook A = curiosity gap (“I didn't expect this to actually fix my…”), hook B = social proof with a number (“47,000 people switched in the last month — here's why”), hook C = pattern interrupt visual (product enters frame at an unexpected angle with no spoken intro for the first 2 seconds). The body of the ad is identical for all three variants. Only the hook changes.

This structure produces clean test data. If hook B outperforms hook A by 30 points on the rubric, you know the formula drove the delta — not a different product demo or a different CTA, because those were held constant. Variant briefs that don't hold the body constant produce noise, not signal.

Before handing off a brief, read it out loud as if you were the creator seeing it for the first time. If you cannot picture the first three seconds of the video after reading the hook direction, the hook direction is too vague — rewrite it with a sample first line. If you don't know exactly what text appears on screen, the brief is incomplete — specify it.

Check that every one of the six elements is present, specific, and in plain language. If the proof element says “highlight positive results” instead of naming a specific number or testimonial, it will not score well on the proof dimension. If the CTA says “encourage viewers to shop” instead of the exact words to say and the timestamp, a creator cannot execute it consistently across variants.

The final check: if The Ad Bench would score your brief description as a script and it comes back below 60, rewrite the brief before briefing anyone. A brief that cannot pass its own rubric will not produce creative that passes it either. The brief is the strategy; the creative is the execution. Fix strategy problems at the strategy layer.