Bad prompts get bad answers. Good prompts get useful answers. The difference is learnable in ten minutes.
This guide gives you the core rules, ready-to-use templates, and specific strategies for different tasks. No jargon, no fluff.
Three Rules That Actually Matter
Everything else is a variation of these.
1. Be Specific
Vague in, vague out. Tell the AI exactly what you want: format, length, tone, audience.
Vague: “Write about marketing.”
Specific: “Write 5 Instagram post ideas for a small bakery. Casual tone. Each post under 50 words.”
The specific version answers every question the AI might have. No guessing required.
More examples of vague vs specific:
| Vague | Specific |
|---|---|
| “Help me with my resume” | “Review my resume for a senior marketing role. Focus on quantifying achievements. Keep bullet points under 15 words each.” |
| “Explain blockchain” | “Explain blockchain to a small business owner considering crypto payments. No technical jargon. Use a real-world analogy.” |
| “Write a thank you note” | “Write a thank you note to a client who just signed a $50K contract. Professional but warm. 3-4 sentences.” |
Each specific version eliminates ambiguity. The AI knows exactly what success looks like.
2. Give Context
The AI doesn’t know your situation until you explain it.
Without context: “Help me write an email to my boss.”
With context: “I need to ask my boss for a raise. I’ve been here 2 years, got promoted once, and just finished a big project. She’s direct and prefers short emails. Help me write the request.”
Context shapes everything: tone, detail level, what to include, what to skip.
What counts as useful context:
- Your role or situation
- Who will read/use the output
- What you’ve already tried
- Constraints (time, budget, word count)
- The outcome you’re hoping for
You don’t need all of these every time. But the more relevant context you provide, the less the AI has to guess.
3. Iterate
First answer not great? Don’t start over. Refine.
Useful follow-ups:
- “Good, but make it shorter”
- “More casual tone”
- “Focus on the second point”
- “Give me 3 more options”
- “Now make it more persuasive”
- “Add a specific example”
The third version is usually better than the first. Treat it like a conversation, not a slot machine.
Iteration also helps you learn what works. After a few back-and-forth exchanges, you start to notice which instructions produce the best results. Next time, you can include those upfront.
Ready-to-Use Templates
Copy these. Fill in the brackets. Get better results immediately.
General Purpose
I need help with [task].
Context: [your situation, background info]
Please give me [format: list/paragraph/steps/options].
Keep it [tone: casual/professional/simple] and under [length].
Writing Help
Write a [type: email/post/message] about [topic].
Audience: [who will read this]
Tone: [casual/professional/friendly/direct]
Length: [word count or "keep it short"]
Include: [specific points to cover]
Problem Solving
I'm trying to [goal] but [obstacle].
Here's what I've tried: [previous attempts]
What are my options? Explain simply.
Learning Something
Explain [topic] like I'm [knowledge level: complete beginner/some background/intermediate].
Use a real example. Avoid jargon.
Decision Making
I need to decide between [option A] and [option B].
My priorities: [what matters most]
My constraints: [budget/time/other limits]
Give me pros and cons for each, then your recommendation.
Review and Feedback
Review this [document/code/plan] for [specific criteria].
Point out weaknesses. Suggest specific improvements.
Be direct. I want honest feedback, not encouragement.
Strategies by Task Type
Different tasks need different approaches. Here’s what works for each.
For Writing Tasks
Creative prompts benefit from constraints, not open-ended freedom. Specify tone, audience, length, and at least one stylistic requirement.
Weak: “Write a blog post about productivity.”
Strong: “Write a blog post about the two-minute rule for productivity. Audience: busy professionals skeptical of productivity advice. Tone: direct, no fluff. One concrete example. 500 words maximum.”
Constraints force better output. Without them, you get generic content that sounds like everything else on the internet.
Additional writing tips:
- Specify what to avoid: “No clichés. No corporate speak.”
- Give a model: “Write in the style of [publication or author]”
- Request a hook: “Start with a surprising fact or question”
For Code and Technical Tasks
Provide complete context: language, framework, what it should do, what’s going wrong.
Effective pattern: “I’m working in Python with FastAPI. This function should return a 400 error with specific field names when validation fails. Currently it returns a generic 500. Here’s the code: [paste code]. Find the issue and show me the fix.”
Include the actual code, not just descriptions. The AI can’t debug what it can’t see.
More technical prompting tips:
- Specify the environment: version numbers, dependencies, OS if relevant
- Describe expected vs actual behavior clearly
- Ask for comments in the code explaining the logic
- Request tests if you want to verify the solution works
For complex problems, ask the AI to explain its approach before writing code. This catches misunderstandings early.
For Analysis and Decisions
Lead with the decision you’re making, not just the information you have.
Example: “I’m choosing between Shopify and WooCommerce for my online store. I sell handmade jewelry, expect 50-100 orders per month, and have no coding experience. Compare them for my situation and recommend one.”
State your criteria upfront. The AI will structure its analysis around what actually matters to you.
For complex analysis:
- Break it into steps: “First summarize the key factors. Then analyze each option. Then give your recommendation.”
- Ask for tradeoffs explicitly: “What am I giving up with each choice?”
- Request confidence levels: “How confident are you in this recommendation? What would change your answer?”
For Brainstorming and Ideas
When you want quantity and variety, say so explicitly.
Example: “Give me 10 video ideas for a YouTube channel about home cooking. Mix practical tutorials, entertainment, and trending topics. Include one unconventional idea I probably haven’t considered.”
Asking for a specific number forces diversity. Asking for something unconventional pushes past obvious suggestions.
Follow-up prompts that help:
- “Which 3 of these have the best potential? Why?”
- “Take idea #4 and develop it further”
- “Give me 5 more ideas, but completely different from these”
Advanced Tips
These techniques take a bit more effort but produce noticeably better results.
Ask for Step-by-Step Reasoning
For complex problems, ask the AI to think through the problem before answering.
Example: “I’m trying to decide whether to hire a contractor or do this renovation myself. Think through the key factors step by step, then give me your recommendation.”
This produces more thorough analysis and catches errors the AI might otherwise make.
Assign a Role
Tell the AI who to be. This shapes expertise level, communication style, and perspective.
Example: “You are a CFO reviewing a startup’s financials. Look at this P&L statement and identify the three biggest concerns you’d raise with the founders.”
Role assignment works especially well for specialized advice, feedback, and analysis.
Ask for Multiple Perspectives
When you want balanced analysis, request different viewpoints explicitly.
Example: “Explain the pros and cons of remote work from three perspectives: the employee, the manager, and the company’s finance team.”
This prevents one-sided answers and surfaces considerations you might miss.
Request Self-Critique
Ask the AI to evaluate its own answer.
Example: After getting a response, follow up with: “What are the weaknesses in this advice? What did you leave out? What assumptions might be wrong?”
This often surfaces caveats and limitations the initial response glossed over.
Quick Tricks
Want clarifying questions first? Add: “Before answering, ask me 2-3 questions to understand my situation better.”
Want multiple options? Add: “Give me 3 different approaches.”
Want honesty about uncertainty? Add: “If you’re not sure about something, say so instead of guessing.”
Want a specific format? Add: “Format your response as [bullet points / numbered list / table / paragraphs].”
Want brevity? Add: “Keep your response under [X] words” or “Give me the short version first, then details if I ask.”
Common Mistakes
Asking too much at once. One task per prompt. Break big requests into sequential steps. “Help me write, edit, format, and translate this document” will produce worse results than handling each step separately.
Skipping audience details. “Write a blog post” produces generic content. “Write a blog post for first-time parents with 5 minutes to read” produces targeted content. Always specify who the output is for.
Accepting the first answer. Iterate. The improvement between attempt one and attempt three is usually significant. Most people give up too early.
Asking for predictions. AI doesn’t know the future. Ask about current patterns and existing information instead. “What will happen” is weaker than “What typically happens” or “What factors should I consider.”
Overloading with instructions. If your prompt is longer than the response you want, simplify it. Conflicting instructions confuse the model. Focus on what matters most.
Being too polite. You don’t need to say please and thank you. It doesn’t affect the output. Use that space for useful instructions instead.
Bottom Line
Three rules: Be specific. Give context. Iterate.
Six templates to start with. Task-specific strategies when you need them. Advanced techniques when you’re ready.
The gap between frustrating AI interactions and productive ones is smaller than it looks. Clear input, clear output. Start with the templates, notice what works, and adjust from there.
Practice makes it automatic. Within a week of deliberate use, you’ll write better prompts without thinking about it.