How to Write Better Prompts: 5 Core Principles
prompt-writingprinciplesbeginner# How to Write Better Prompts: 5 Core Principles
## The Problem with Most Prompts
Most people treat AI prompts like Google searches — short, vague, and hope for the best. But AI language models aren't search engines. They're reasoning engines. And they need clear reasoning instructions.
## Principle 1: Be Specific About the Role
**Bad**: "Write a marketing email."
**Good**: "You are a direct response copywriter with 10 years of SaaS experience. Write a marketing email..."
Why it works: Assigning a role gives the AI a specific voice, expertise level, and perspective to write from.
## Principle 2: Define the Audience
**Bad**: "Explain machine learning."
**Good**: "Explain machine learning to a 35-year-old marketing manager who has never coded but is curious about AI."
Why it works: When the AI knows who will read the output, it adjusts vocabulary, depth, and examples accordingly.
## Principle 3: State the Format and Structure
**Bad**: "Help me plan a project."
**Good**: "Create a project plan with: 1) Executive summary (2 paragraphs), 2) Timeline (table format, 4 phases), 3) Risk assessment (top 5 risks with mitigation), 4) Resource requirements."
Why it works: Specifying structure prevents rambling output and ensures you get exactly what you need.
## Principle 4: Provide Context and Constraints
**Bad**: "Write product descriptions."
**Good**: "Write product descriptions for handmade ceramic mugs. Target audience: design-conscious millennials. Tone: warm and artisanal. Max 100 words per description. Avoid words like 'unique' and 'handcrafted.'"
Why it works: Constraints force creativity and relevance. Without them, the AI defaults to generic output.
## Principle 5: Iterate and Refine
Your first prompt is a starting point, not the final answer. After getting a response:
1. Identify what's good and what's off
2. Adjust the prompt to fix specific issues
3. Add examples of desired output if needed
4. Repeat until satisfied
## The CARE Framework
A simple framework that combines all 5 principles:
- **C**ontext: Set the scene and role
- **A**udience: Define who the output is for
- **R**equest: State exactly what you want
- **E**xample: Provide a sample of desired output
## Try It Yourself
Use our [Prompt Generator](/generator) to practice these principles, or browse [prompts by scenario](/for/writing-blog-post) to see them in action.
## The Problem with Most Prompts
Most people treat AI prompts like Google searches — short, vague, and hope for the best. But AI language models aren't search engines. They're reasoning engines. And they need clear reasoning instructions.
## Principle 1: Be Specific About the Role
**Bad**: "Write a marketing email."
**Good**: "You are a direct response copywriter with 10 years of SaaS experience. Write a marketing email..."
Why it works: Assigning a role gives the AI a specific voice, expertise level, and perspective to write from.
## Principle 2: Define the Audience
**Bad**: "Explain machine learning."
**Good**: "Explain machine learning to a 35-year-old marketing manager who has never coded but is curious about AI."
Why it works: When the AI knows who will read the output, it adjusts vocabulary, depth, and examples accordingly.
## Principle 3: State the Format and Structure
**Bad**: "Help me plan a project."
**Good**: "Create a project plan with: 1) Executive summary (2 paragraphs), 2) Timeline (table format, 4 phases), 3) Risk assessment (top 5 risks with mitigation), 4) Resource requirements."
Why it works: Specifying structure prevents rambling output and ensures you get exactly what you need.
## Principle 4: Provide Context and Constraints
**Bad**: "Write product descriptions."
**Good**: "Write product descriptions for handmade ceramic mugs. Target audience: design-conscious millennials. Tone: warm and artisanal. Max 100 words per description. Avoid words like 'unique' and 'handcrafted.'"
Why it works: Constraints force creativity and relevance. Without them, the AI defaults to generic output.
## Principle 5: Iterate and Refine
Your first prompt is a starting point, not the final answer. After getting a response:
1. Identify what's good and what's off
2. Adjust the prompt to fix specific issues
3. Add examples of desired output if needed
4. Repeat until satisfied
## The CARE Framework
A simple framework that combines all 5 principles:
- **C**ontext: Set the scene and role
- **A**udience: Define who the output is for
- **R**equest: State exactly what you want
- **E**xample: Provide a sample of desired output
## Try It Yourself
Use our [Prompt Generator](/generator) to practice these principles, or browse [prompts by scenario](/for/writing-blog-post) to see them in action.