D - Define Multiple Perspectives

• Instead of single role, use multiple expert viewpoints collaborating
• Example: "You are three experts: behavioral psychologist, copywriter, and data analyst"
• Prevents generic single-role outputs
• Creates richer, more nuanced responses

E - Establish Success Metrics

• Define specific, measurable outcomes upfront
• Instead of "make it good" → "optimize for 40% open rate, 12% CTR, include 3 psychological triggers"
• No metrics = no optimization
• Gives AI clear targets to hit

P - Provide Context Layers

• Stack multiple context elements: business type, budget, audience, past performance
• Example: "B2B SaaS, $200/mo product, targeting overworked founders, previous emails got 20% opens"
• Context dramatically improves relevance
• More detail = more targeted output

T - Task Breakdown

• Break requests into sequential steps instead of one big ask
• Example: "Step 1: Identify pain points. Step 2: Create hook. Step 3: Build value. Step 4: Soft CTA"
• Prevents AI confusion and generic responses
• Creates logical flow in output

H - Human Feedback Loop

• Don't accept first output - build in self-critique
• Example: "Rate your response 1-10 on clarity, persuasion, and actionability. Improve anything below 8"
• Self-critique produces 10x better results
• Iterative improvement built into single prompt

TD;LR

1. Single-role prompts = generic outputs - Always use multiple perspectives
2. No metrics = no optimization - Specify exact success criteria
3. Context is king - More background = better relevance
4. Break down tasks - Step-by-step prevents confusion
5. Force self-evaluation - AI critiquing itself improves quality dramatically