The Creative Constraint Generator: Turn a Stuck Project Into Rules That Force New Ideas
Why this prompt matters
Generic creative output doesn't just waste brainstorming time — it burns real production budget shooting a campaign that blends into the category, which forces an expensive second round of concepting after the first one tests flat with audiences.
What we use it for
Your team ran two open brainstorming sessions for a cold brew coffee brand's launch campaign, generated 40 concepts, and a blind test showed people couldn't distinguish any of them from three competitor brands already on the shelf.
Prompt
Role: You are a creative director known for breaking teams out of ideation ruts through structured constraint-setting, not open-ended brainstorming. Context: - The stuck project: [DESCRIBE THE PROJECT AND WHY IT FEELS STUCK, e.g. "We need a tagline and three concept directions for a cold brew coffee brand aimed at Gen Z"] - What's already been tried: [DESCRIBE THE PRIOR ATTEMPT AND WHY IT FAILED, e.g. "Two open brainstorming sessions produced 40 ideas, but a blind test showed people couldn't distinguish any of them from three competitor brands already on the shelf"] - Non-negotiable constraints that must NOT be touched: [LIST ANY REAL BRAND/LEGAL/FORMAT RULES, e.g. "Must mention 'cold brew' somewhere, no profanity, must work as a 15-second ad"] Task: 1. Diagnose the single unconscious default assumption that the stuck ideas all converged on — the thing every "different" idea secretly had in common. 2. Generate five artificial creative constraints: invented rules specifically designed to break that default. For each, state exactly which assumption it attacks. 3. For each constraint, produce one concrete concept that could only exist because of that constraint — not a generic idea that happens to comply with it. 4. Rank the five resulting concepts on two axes: how surprising they are, and how executable they are with a normal production budget. 5. Recommend which one or two concepts are worth prototyping first, and explain why in terms of risk versus differentiation. Constraints: - Every invented constraint must be genuinely restrictive. A "constraint" that doesn't rule anything out is not a constraint. - Do not invent constraints that are just the literal opposite of the generic default (e.g. "be loud" → "be quiet"). Find non-obvious restriction angles: point of view, format, sensory channel, borrowed visual language from an unrelated category, timing, audience assumption. - Never treat the non-negotiable constraints listed in Context as things to subvert — they are fixed. - Keep each concept description under 40 words. This is a divergence tool for generating directions fast, not a full creative brief. Output Format: - Section 1: Root Assumption Diagnosis (one to two sentences) - Section 2: Five Constraints (numbered list: the constraint, then the assumption it attacks) - Section 3: Five Concepts (one per constraint, each under 40 words) - Section 4: Surprise vs. Executability Ranking (ranked list or table) - Section 5: Recommendation (one or two concepts to prototype first, with reasoning)
Result
Section 1: Root Assumption Diagnosis
Every direction generated so far assumes the brand must speak in the first-person voice of a confident, rebellious drinker — high-energy, main-character framing. That default is also the default of nearly every cold brew competitor's advertising since 2018, which is exactly why blind testers couldn't tell the 40 concepts apart from the shelf competition.
Section 2: Five Constraints
- No first-person or second-person voice — copy may never address the drinker directly or imply their identity; it can only describe the coffee itself. (Attacks: the assumption that bold branding must speak to the drinker's identity.)
- The entire campaign must work as a silent, text-free visual with no spoken or written copy except the product name. (Attacks: the assumption that "bold" requires a bold tagline.)
- The concept must be built around a specific, unglamorous moment in the day — not "morning hustle" or "late-night grind." (Attacks: the assumption that coffee marketing must live in peak-energy moments.)
- The visual language must be borrowed from a category with zero overlap with coffee or energy drinks — for example, aquarium-supply catalogs or government safety posters. (Attacks: the assumption that coffee ads should look like other coffee ads.)
- The concept must land as funnier to someone who dislikes coffee than to someone who loves it. (Attacks: the assumption that the ad must flatter the existing customer.)
Section 3: Five Concepts
- A single slow-motion shot of cold brew pouring over ice, no voiceover, no copy except the product name appearing at the very end — texture and sound carry the entire ad.
- A static 15-second shot: a can sweating condensation onto a kitchen counter in real time, nothing else, tension built entirely from anticipation.
- A single static frame: an office desk at the exact 3:47pm dead-zone slump, half-eaten snack in frame, a can entering from off-screen — the only text on screen is the timestamp itself.
- Shot like a pet-store fish-tank equipment catalog: clinical, deadpan closeup product photography against a blue-lit background, deliberately unglamorous by contrast with typical coffee advertising.
- A mock public-safety announcement about caffeine tolerance thresholds, delivered completely deadpan like a government seatbelt ad — plays as comedy to people who find coffee culture absurd.
Section 4: Surprise vs. Executability Ranking
High surprise, high executability: Concept 3 (3:47pm desk shot) — cheap to shoot, genuinely unexpected timing angle.
High surprise, medium executability: Concept 4 (aquarium-supply aesthetic) — needs a photographer who can nail deadpan tone precisely.
Medium surprise, high executability: Concept 2 (condensation shot) — easy to shoot, moderate novelty.
Medium surprise, low differentiation: Concept 1 (slow-motion pour) — closest to existing category norms.
High surprise, high risk: Concept 5 (PSA parody) — funniest on paper, but risks reading as mocking the customer if the tone misses.
Section 5: Recommendation
Prototype Concepts 3 and 4 first. Concept 3 is the safer bet: cheap to shoot, directly attacks the single most-repeated assumption (peak-energy timing) from the original brainstorm, and the "boring specific timestamp" device is memorable without depending on a joke landing. Concept 4 is the higher-risk, higher-reward option — if the deadpan product-photography tone lands, nothing else in the category looks like it, but the entire concept depends on getting that photographic tone exactly right, so test it as a single still frame before committing video production budget.
Open brainstorming has a well-known failure mode: a room full of smart people independently reach for the same unconscious default, and twenty “different” ideas turn out to be variations on one idea. This prompt doesn't try to generate more ideas faster. It tries to name the default everyone is stuck inside, then forces a way out of it through constraints specific enough to actually restrict something.
Why diagnosis comes before constraint generation
Most creative-prompting approaches skip straight to “generate options.” This one insists on Section 1 first: naming the single assumption that every prior idea secretly shared. Without that diagnosis, constraint generation is just random variation — you might stumble onto something different by accident, but you have no way to know whether a new idea actually escapes the stuck pattern or just looks different on the surface while making the same underlying choice.
Why the constraints have to be genuinely restrictive
The Constraints section explicitly bans constraints that don't rule anything out, and bans the lazy move of just inverting the default (loud becomes quiet, fast becomes slow). A constraint like “borrow visual language from aquarium-supply catalogs” is restrictive in a way that “be more unexpected” is not — it forces specific, checkable decisions about framing, lighting, and tone that a vague instruction to “think outside the box” never produces. The five suggested attack angles in the prompt — point of view, format, sensory channel, borrowed category, timing — are deliberately concrete because abstract constraints tend to produce abstract (read: still generic) ideas.
Why the output ranks surprise against executability
A constraint-generation exercise that only produces wild ideas is just as useless as one that only produces safe ideas — wild ideas that cost a production budget you don't have are not a deliverable. Section 4 forces an honest tradeoff assessment so a creative lead can choose deliberately between a safe differentiator and a risky one, rather than discovering the executability problem after the client has already fallen in love with a concept that needs a crew of twelve.
How to use this well
Be specific in the “what's already been tried” field — vague descriptions of the stuck brainstorm produce a vague diagnosis in Section 1, which weakens every constraint that follows. This prompt works on more than advertising: it applies equally to a stuck product-naming session, a game-design pitch that keeps reinventing the same mechanic, or a research proposal that reads like every other grant application in the pile. Run it when a team has already tried open brainstorming and gotten a pile of samey options — not as a replacement for the first brainstorm, but as the second move once that one has visibly plateaued.