The Frustration of the Vague Request
We have all been there. Faced with a creative block or a new business challenge, you turn to a generative AI tool and type something like, “Give me marketing ideas for a new app.” The response is a predictable list of suggestions: “Use social media,”
“Start a blog,” “Run digital ads,” “Collaborate with influencers.” While technically correct, this advice is so generic it’s practically useless. This experience highlights a fundamental misunderstanding of how to work with these powerful tools. AI is not a magic eight-ball that can read your mind; it's a powerful engine that requires clear instructions to deliver meaningful results. Without specific direction, it defaults to the most statistically probable and, therefore, most average responses based on its vast training data.
Garbage In, Supercharged Garbage Out
The old programming adage “garbage in, garbage out” is amplified with AI. A vague prompt is the equivalent of giving a master chef a bare pantry and asking them to “make dinner.” You might get a simple omelet, but you won't get a Michelin-star meal. To get quality, you must provide quality ingredients. In the world of AI ideation, those ingredients are your decision criteria. These are the specific, measurable, and relevant constraints that define what a “good” idea looks like for you. By failing to provide this context, you force the AI to make assumptions, leading to outputs that don't align with your budget, target audience, brand voice, or technical capabilities. The result is a wasted opportunity and the false conclusion that AI isn't a good tool for creativity.
How to Define Your Decision Framework
Setting decision criteria isn't complicated. It simply means telling the AI the rules of the game before it starts playing. Think of it as building a fence around the playground; it gives the AI freedom to play within a safe and relevant space. A strong set of criteria typically includes several key elements. Start with your target audience, defining their demographics, needs, and pain points. Specify your constraints, such as budget limitations, timeline for execution, and available technical resources. Define the desired tone and brand voice—are you playful, professional, or disruptive? Lastly, clarify the desired outcome. Are you looking for names for a product, taglines for a campaign, or features for a new service? The more specific you are, the better the AI can tailor its suggestions.
Criteria in Action: A Before-and-After Example
Let’s see how this works in practice. A 'before' prompt might be: “Suggest names for a new coffee brand.” The AI would likely return generic names like “The Daily Grind” or “Morning Brew.” Now, let’s apply a decision framework. The 'after' prompt becomes: “Act as a branding expert. Generate 10 names for a new direct-to-consumer coffee brand based in India. The target audience is environmentally-conscious millennials. The names should evoke a sense of sustainability and ethical sourcing, be easy to pronounce, and have an available '.in' domain. The tone should be earthy and authentic.” The expected output would be far more specific and useful, offering names that are already vetted against your core requirements. This approach transforms the AI from a simple suggestion box into a strategic partner.
Beyond Generation: Using Criteria for Ranking
The power of a well-defined framework extends beyond just generating ideas. Once the AI has provided a list of suggestions based on your initial prompt, you can use those same criteria to have it rank and evaluate its own output. For instance, you could follow up with: “Now, rank the 10 names you provided based on the following criteria, with the most weight on brand recall and uniqueness.” The AI can then score each idea, provide a rationale for its ranking, and even identify potential risks or weaknesses associated with the top choices. This second step saves an immense amount of time, helping you move from a long list of possibilities to a short list of viable contenders without hours of manual sifting and subjective debate.
