Questioning the Brief with AI

Most creative leads treat a brief as a set of instructions to be answered immediately. When you feed a brief straight into an AI to generate routes, the AI fills the gaps with 'defaults'—the generic, average middle of the industry. This is where a brand loses its soul and begins to look like every competitor. To avoid this, your first move is not to answer the brief, but to audit it.

Consider a brand lead launching a repair service for worn jackets across three different markets. The brief arrives with high-level goals, but lacks specific target audience data for those regions. If the lead asks the AI for creative directions now, the AI will assume a generic 'eco-conscious consumer.' By using the AI as an auditor first, the lead can identify exactly what is missing. Instead of generating an average idea, they can go back to the stakeholder and say, "We lack the market-specific data needed to make this distinct." You move from being a producer of assets to a custodian of the idea.

To get a useful audit, you must change the AI's objective. If you ask, "What are some ideas for this brief?" the AI enters 'creation mode' and tries to please you with possibilities. To separate preference from strategy, you must force it into 'auditor mode.' This means providing a specific remit: a set of instructions that tells the AI to look for what is not there, rather than imagining what could be.

For example, take the jacket repair service brief. Instead of asking for concepts, you prompt the AI to analyze the text specifically for gaps. You instruct it to: "List all missing delivery constraints, such as specific screen dimensions for the three markets, and flag any unclear KPIs (Key Performance Indicators) that are currently described in vague terms."

The AI might then report that while the brief mentions 'in-store screens,' it doesn't specify the aspect ratio or resolution for the Japanese market. It might flag that 'increasing brand love' is listed as a KPI but lacks a measurement method. By defining this remit, you generate a punch list of strategic omissions. You now have a professional, evidence-based reason to pause production and request clarity from the stakeholder.

Once the AI identifies the contents of your brief, you must categorize every requirement as either Evidence or Hope. Evidence consists of proven insights—data, research, or established brand standards—that provide a factual foundation for a decision. Hope consists of stakeholder wishes—subjective preferences or 'gut feelings'—that describe a desired outcome without a strategic reason why.

Take the jacket repair service example. A line stating, "Data shows 60% of customers keep jackets for 5+ years," is Evidence. It is a factual anchor that justifies focusing the creative on the beauty of wear and longevity. Conversely, a line stating, "We feel the imagery should look premium," is Hope. 'Premium' is a subjective term that varies from person to person and often serves as a proxy for a stakeholder's personal taste.

By labeling these lines, you stop treating personal preference as a final creative decision. When a stakeholder pushes for a change based on Hope, you don't push back by sounding precious; you simply ask for the Evidence that supports that shift. You are now managing the project based on what is proven rather than what is wished.

Lines of 'Hope' are not just vague; they are often masks for a lack of strategic direction. When a stakeholder uses subjective adjectives, they are usually describing a feeling they want to achieve without knowing the strategic lever that creates it. If you accept these lines as instructions, you are guessing at the stakeholder's personal taste, which makes the final creative decision vulnerable to their mood on the day of the presentation.

Consider the instruction to "make it feel disruptive." In a vacuum, 'disruptive' could mean a loud, neon color palette to one person, or a stark, silent white space to another. By flagging this as a Hope line, you identify a strategic gap. You don't ignore the request, but you refuse to guess the meaning. Instead, you flag it for clarification: "To ensure we are truly disruptive in the Japanese market, do we define disruption as breaking category color norms or challenging the traditional repair narrative?"

By mapping every Hope line, you transform a list of subjective wishes into a set of specific questions. You shift the burden of definition back to the stakeholder, ensuring the creative work is guided by a shared agreement rather than a hopeful guess.

While identifying 'Hope' lines focuses on subjectivity, identifying strategic gaps focuses on logical contradictions. These are the silent killers of creative work—conflicting instructions that seem fine in a document but are impossible to execute visually. If left undetected, these contradictions lead to late-stage revisions where you are forced to 'average' the notes, resulting in work that satisfies no one and flattens the original idea.

Imagine the jacket repair brief asks for a 'minimalist, high-end design' to emphasize craft, but also requires 'prominent pricing and three discount banners' on every social asset. These are contradictory goals: minimalism relies on the removal of noise, while multiple banners and bold pricing create noise.

By prompting your AI auditor to 'find logical contradictions between the aesthetic goals and the delivery requirements,' the AI can surface these clashes before a single pixel is moved. Once the AI flags the conflict, you don't try to solve it creatively; you present the contradiction to the stakeholder. You ask them to prioritize: does the brand value the minimalist prestige of the repair service, or the immediate conversion of the discount? Solving this gap at the brief stage prevents the 'death by revisions' that happens when these two goals collide in round four.

Once the AI audit has surfaced gaps and 'Hope' lines, you have the raw data needed to push back. The secret to doing this without sounding precious is to frame your questions around the project's success rather than your creative intuition. You aren't arguing about taste; you are requesting the strategic tools required to do the job. When you use the AI's findings to build these questions, you move the conversation from 'I don't like this' to 'We need this to work.'

For example, if the AI flagged the word 'disruptive' as a Hope line with no supporting evidence, do not tell the stakeholder that 'disruptive' is too vague. Instead, frame it as a risk to the output: "To ensure the imagery hits the mark and doesn't miss the target, can we define the specific audience insight that makes this route disruptive?"

By doing this, you position yourself as a partner in the brand's success. You are essentially saying: 'I want to give you exactly what you want, but I need the strategic definition to ensure I don't waste your time with the wrong version.' This removes the ego from the room and anchors the decision in the brief.

The final step is to formalize the AI's audit into a Decision Framework. This is a living document—often a simple table—that acts as the project's 'source of truth.' By documenting what is Evidence and what is Hope before the Concept phase begins, you create a baseline that makes personal preference visible and manageable. You are no longer arguing about a visual; you are checking a visual against a pre-agreed framework.

Using the jacket repair service example, your table would have two primary columns. The first lists 'Agreed Evidence,' such as the data on jacket longevity and the specific screen dimensions for the Japanese market. The second lists 'Unresolved Wishes,' such as the request to 'feel disruptive' or 'look premium.'

You present this table to the stakeholder and agree that the Evidence column is non-negotiable. The Wishes column is flagged as the 'risk zone'—items that must be converted into Evidence (through research or a concrete definition) before the final creative release. When a stakeholder later says, "I just prefer a different blue," you simply point to the framework and ask, "Does this preference align with our agreed evidence, or is this a new wish we need to categorize?"