By skyforbes 
Person asks AI to do something → AI gives generic output → Person iterates 15 times → Frustration
The issue? They never defined what success looks like before they started.
So I built a stupid-simple framework. Three questions you ask yourself before writing any prompt:
1. What's the ONE metric that defines success?
(Not "make it good" — actual measurable outcome)
2. Who's the end user and what's their biggest barrier?
(Specific person, specific problem)
3. What's the constraint hierarchy?
(What matters MOST > Second > Third if you must compromise)
Example: I asked someone to write an article about Harry Potter audiobooks.
❌ Without framework: Generic 1000-word "here's what's new" post (forgettable)
✅ With framework: They answered the 3 questions first:
- Success = virality (social shares)
- User = new audiobook listeners (skeptical)
- Priority = authority > engagement > word count
Result: AI wrote a completely different article. Controversial takes, insider data, provocative framing. Built for sharing, not just informing.
The framework takes 2 minutes. Saves hours of iteration.
I wrote it up with examples across different use cases (writing, marketing, code, strategy): https://medium.com/aidrivenprompt/once-youve-done-that-thinking-the-ai-prompt-writes-itself-26f16a36c3db
Free. No signup. Just copy-paste and use it.
Has anyone else noticed this pattern? Curious if this resonates.