10 Tips for Writing Better AI Prompts

Most people type “write me a blog post about marketing” and wonder why ChatGPT sounds like a corporate robot.

The difference between useless AI output and genuinely good results usually comes down to about 30 extra seconds of prompt writing. I’ve tested hundreds of prompts across different tools, and these 10 techniques consistently make the biggest difference.

1. Be Specific, Not Vague

Vague prompt: “Write about productivity” Specific prompt: “Write a 1,200-word blog post about time-blocking for busy professionals earning $100K+. Include 5 specific strategies, realistic time blocks, and how to handle interruptions. Target a professional but conversational tone. Include one detailed case study of a real implementation.”

Why it matters: AI thrives on specificity. Specific prompts produce specific, valuable results. Vague prompts produce generic, forgettable content.

Specificity includes:

• Exact word count (not “around 1000 words”)
• Target audience details (not just “professionals”)
• Specific tone and voice
• Number of sections or items
• Concrete examples needed

Result: Instead of generic advice, you get tailored, specific content that actually serves your audience.

2. Provide Context and Background

Without context: “Analyze this document” With context: “You’re a marketing strategist for SaaS companies. Analyze this product launch timeline and identify potential marketing risks for the B2B buyer journey.”

Why it matters: Context helps AI understand not just what to do, but why and who for.

Elements to include:

• Your role or expertise area
• The audience being served
• The business context
• The ultimate goal
• Any constraints or requirements

Example: Instead of “Summarize this article,” try “You’re a content curator for a sales team. Summarize this sales methodology article, focusing on techniques our team could implement immediately.”

3. Use Examples to Shape Output

Without examples: “Generate 5 social media captions for our product” With examples: “Generate 5 social media captions for our productivity tool. Here’s the tone/style we use: [example 1, example 2]. Each caption should be under 280 characters, include a relevant emoji, and encourage clicks to our landing page.”

Why it matters: Showing examples teaches AI exactly what good looks like for your specific needs.

How to do it:

• Provide 2-3 examples of output you like
• Explain what makes them work
• Let AI use those as reference points

Result: Output matches your existing brand voice and style perfectly.

4. Use Role-Playing Prompts

Generic: “Explain machine learning” Role-based: “You’re a tech expert explaining machine learning to a CEO with no technical background. Avoid jargon. Use business metaphors. Explain why they should care about it in 500 words.”

Why it matters: Assigning AI a role dramatically improves output relevance and tone.

Role examples:

• “You’re a financial advisor for millennials…”
• “You’re a demanding editor for a top publication…”
• “You’re a customer success manager helping onboard new users…”

Result: The response is tailored to exactly what that role would produce.

5. Break Complex Tasks into Steps

All at once: “Create a complete marketing strategy for launching a new app” Broken into steps:

1. “First, identify the target customer persona for a mobile productivity app”
2. “Create a 3-month launch timeline with key milestones”
3. “Develop a go-to-market strategy for the first 30 days”
4. “Create metrics to measure success”

Why it matters: Complex tasks get better results when broken into sequential steps. This also helps you iterate and refine at each stage.

Bonus: You can often use outputs from step 1 as inputs for step 2, creating a building chain of increasingly customized content.

6. Ask for Different Formats

Generic: “Give me ideas for blog content” Formatted: “Give me blog content ideas in this format:

• Topic title
• Target search term (with monthly search volume if known)
• Unique angle
• Why our audience cares
• Potential formats (article, video, infographic)”

Why it matters: Structured formats make AI output immediately usable. You save the formatting step.

Format examples:

• Tables
• Numbered lists
• Bullet points
• JSON
• CSV format
• Markdown

Result: Output you can immediately paste into your systems without reformatting.

7. Use “Chain of Thought” Reasoning

Without: “Should we hire for this role?” With: “I’m deciding whether to hire for this open role. Walk me through your thinking step-by-step: 1) What would this role cost? 2) What revenue/value would it generate? 3) What if we waited 6 months? 4) What are the risks? Then give me a recommendation.”

Why it matters: Asking AI to show its reasoning produces better insights than just asking for conclusions.

How to use it:

• Ask “Think step-by-step”
• Request “Show your reasoning”
• Ask for “Pros and cons for each option”
• Request “Three different perspectives on this”

Result: More thorough analysis that helps you understand the decision, not just the answer.

8. Specify Constraints

Without constraints: “Create a tagline for our brand” With constraints: “Create a tagline for our brand with these constraints: under 5 words, no jargon, appeals to busy professionals, memorable, differentiates us from competitors, and avoids overused phrases.”

Why it matters: Constraints force creative solutions. Unconstrained prompts produce unconstrained, often mediocre results.

Useful constraints:

• Word/character limits
• Must include specific elements
• Cannot include certain topics/jargon
• Must appeal to specific audience
• Must differentiate from competitors

Result: Output is refined and focused, not bloated or generic.

9. Ask for Multiple Options and Rankings

Single option: “Write an email to prospects” Multiple options: “Write 3 different email approaches to prospects: 1) curiosity-based hook, 2) problem-based opener, 3) social proof approach. For each, write the first 50 words. Then rank them by likely open rate and explain your ranking.”

Why it matters: Multiple options let you see different approaches. Rankings help you understand what works.

How to use it:

• “Give me 5 options”
• “Create 3 variations”
• “Show me the best, worst, and middle-ground approach”
• “Rank these by effectiveness with explanation”

Result: You choose the best option for your specific situation instead of accepting the first output.

10. Include Feedback and Iterate

One shot: Ask one prompt, use output as-is. With iteration:

1. Ask initial prompt
2. Analyze output
3. Provide specific feedback: “This is good but too formal, add more personality, include specific numbers, remove the last paragraph”
4. Ask AI to revise
5. Keep iterating until perfect

Why it matters: AI improves dramatically with feedback. Your iterative refinement creates output that’s actually tailored to your needs.

Feedback examples:

• “Make this 40% shorter”
• “Add more personality and humor”
• “Include specific examples from [industry]”
• “The third section is confusing, simplify it”
• “This is good but missing a clear call-to-action”

Result: Output evolves from good to great to exactly what you need.

The Perfect Prompt Template

Combine these techniques:

“[ROLE]: You’re a [specific role/expertise].

[TASK]: Create [specific deliverable type]

[DETAILS]:

• Length: [specific]
• Audience: [specific details]
• Tone: [specific style]
• Format: [structure requested]

[CONSTRAINTS]:

• Must include [requirements]
• Cannot include [restrictions]
• Should differentiate by [unique angle]

[EXAMPLES]: Here are examples of the style I want: [example 1, example 2]

[PROCESS]: Think through this step-by-step:

1. [Step 1]
2. [Step 2]
3. [Step 3]

Then provide [number] options and rank them by [criteria].”

Conclusion

Great AI output doesn’t come from luck—it comes from prompt skill. These 10 techniques transform your results from “occasionally useful” to “consistently exceptional.”

Practice with one technique this week. Then add another. Over a month, you’ll see 200-300% improvements in AI output quality and usefulness.