How to Write Better ChatGPT Prompts: 15 Proven Techniques
Stop getting generic AI responses. Learn 15 proven techniques to write better ChatGPT prompts that deliver exactly the results you need, every time.
How to Write Better ChatGPT Prompts: 15 Proven Techniques
Let me guess. You opened ChatGPT, typed something like "write me an essay about marketing," and got back a bland, generic wall of text that sounded like it was written by a bored college student at 2 AM. You stared at it, thought "this AI thing is overrated," and closed the tab.
Here is the truth most people do not realize: ChatGPT is not the problem. Your prompt is.
Think of it this way. If you walk into a five-star restaurant and tell the chef "make me food," you are going to get something random. But if you say "I would like a pan-seared salmon with a lemon dill sauce, medium doneness, served with roasted asparagus" — now you are getting exactly what you want.
The same principle applies to AI. The quality of your output is directly tied to the quality of your input. And learning how to write better ChatGPT prompts is not some obscure tech skill — it is a practical, learnable ability that can save you hours every single day.
In this guide, I am going to walk you through 15 proven techniques that will transform your ChatGPT results from "meh" to "wow, did AI actually write this?" Each technique comes with real before-and-after examples so you can see the difference immediately.
Let us get into it.
---
Why Most People Get Bad Results from ChatGPT
Before we dive into the techniques, let us understand why most prompts fail. There are three common reasons:
- Vagueness. Saying "help me with my business" gives ChatGPT nothing to work with. It does not know your industry, your audience, your challenges, or what kind of help you need.
- No context. AI does not read your mind. If you do not provide background information, it fills in the gaps with generic assumptions.
- Unrealistic expectations. People expect a perfect final product from one prompt. In reality, the best results come from conversations — refining and iterating until you get what you need.
The good news? All three problems are completely fixable. Here are the 15 techniques that will fix them.
---
Technique 1: Be Specific About What You Want
This is the single most impactful change you can make. Vague prompts produce vague results. Specific prompts produce specific, useful results.
Bad prompt:
"Write me a blog post about fitness."
Better prompt:
"Write a 1,000-word blog post about the top 5 strength training exercises for beginners over 40 who have never lifted weights before. Use a friendly, encouraging tone. Include safety tips for each exercise."
See the difference? The second prompt tells ChatGPT exactly what topic to cover, who the audience is, how long it should be, what tone to use, and what extra information to include. There is almost no room for misinterpretation.
The rule of thumb: If your prompt could be interpreted in more than one way, it is not specific enough.
---
Technique 2: Provide Context and Background
Context is the secret ingredient that turns generic AI output into something that actually sounds like it was written for your specific situation.
Bad prompt:
"Write a product description for my shoes."
Better prompt:
"I run a sustainable footwear brand called EcoStep. Our target customers are environmentally conscious millennials aged 25-35. Write a product description for our new running shoe, the EcoStep Trail 3.0. It is made from 100% recycled ocean plastic, weighs 8.5 ounces, and has a carbon-fiber reinforced sole. The tone should be aspirational but not preachy."
When you give ChatGPT context about your brand, your audience, and your values, it can tailor the output to match. Without that context, you get a description that could apply to any shoe from any company.
---
Technique 3: Use Examples to Guide the Output
This is one of the most underrated techniques. When you show ChatGPT an example of what you want, it mirrors that style, structure, and quality remarkably well.
Bad prompt:
"Write me a catchy tagline for my coffee shop."
Better prompt:
"Write me 5 taglines for my specialty coffee shop called Brew & Co. Here are some taglines I like from other brands for reference:
- 'Think Different' (Apple)
- 'Just Do It' (Nike)
- 'Because You're Worth It' (L'Oreal)
I like how these are short, memorable, and emotionally resonant. Create something with that same energy but for a cozy, neighborhood coffee experience."
By giving examples, you are essentially saying "this is the standard I am looking for." ChatGPT uses those examples as a blueprint.
---
Technique 4: Set the Format You Want
Do not leave the format up to ChatGPT. If you want a numbered list, say so. If you want a table, ask for it. If you want bullet points with headers, specify that.
Bad prompt:
"Tell me about the best programming languages to learn."
Better prompt:
"Create a comparison table of the top 5 programming languages for beginners in 2025. Include columns for: Language Name, Difficulty Level (1-10), Average Salary, Best Use Cases, and Time to Learn Basics. Add a brief 2-sentence summary after the table."
The format instruction alone can transform a wall of text into something organized, scannable, and actually useful.
---
Technique 5: Assign a Role or Persona
This is where things get really interesting. When you tell ChatGPT to act as a specific expert, it draws from a different pool of knowledge and writes in a completely different style.
Bad prompt:
"How should I invest my money?"
Better prompt:
"You are a certified financial planner with 20 years of experience helping middle-class families build wealth. I am 32 years old, married with one child, earning $75,000 per year. I have $15,000 in savings and no debt besides a mortgage. What is a practical, step-by-step investment plan for the next 5 years? Explain it like I have zero investing knowledge."
The role assignment changes the depth, tone, and specificity of the answer dramatically. Instead of surface-level advice, you get expert-level guidance tailored to your situation.
---
Technique 6: Use Constraints to Your Advantage
Constraints are not limitations — they are creative guardrails. When you tell ChatGPT what NOT to do or set boundaries, the output becomes more focused and relevant.
Bad prompt:
"Explain blockchain to me."
Better prompt:
"Explain blockchain technology in exactly 200 words. Do not use any technical jargon. Do not mention cryptocurrency. Explain it using an analogy that a 12-year-old would understand."
Constraints force ChatGPT to be creative within boundaries, which almost always produces better, more focused results than an open-ended request.
---
Technique 7: Iterate and Refine
Here is a mindset shift that will change everything: your first prompt is not your final prompt. The best ChatGPT users treat it as a conversation, not a one-shot request.
Instead of trying to get everything perfect in one prompt, try this approach:
1. Start with a decent first prompt
2. Review the output
3. Say "This is good, but make these changes: [specific feedback]"
4. Repeat until satisfied
Ready to master AI?
Our Complete AI Bootcamp covers prompt engineering, ChatGPT, MidJourney, vibe coding, AI agents and more — with 110+ video lessons and 2,000+ prompts.
Example iteration:
- Prompt 1: "Write a LinkedIn post about the importance of networking."
- Review: Too generic and long.
- Prompt 2: "Make it shorter — maximum 150 words. Make it more personal by starting with a specific story about a time networking changed someone's career. End with a question to encourage engagement."
- Review: Much better, but the tone is too formal.
- Prompt 3: "Keep the content but rewrite it in a casual, conversational tone. Like I am talking to a friend over coffee."
Three iterations, and you have something publishable. That is the power of refinement.
---
Technique 8: Chain Your Prompts
Complex tasks should not be crammed into one prompt. Break them into a sequence of connected prompts where each builds on the last.
Instead of:
"Write a complete marketing plan for my business."
Try this chain:
1. "I run a small bakery specializing in gluten-free pastries. Who would be my ideal target customer? Give me 3 detailed customer personas."
2. "Based on those personas, what are the top 5 marketing channels I should focus on and why?"
3. "For the top channel you recommended, create a detailed 30-day content plan with specific post ideas."
4. "Write the first 3 social media posts from that plan, ready to publish."
Each prompt builds on the previous answer, creating a chain of increasingly specific and useful outputs. The final result is dramatically better than anything you would get from a single prompt.
---
Technique 9: Use Delimiters to Organize Information
When your prompt includes multiple pieces of information — like a text to analyze, instructions, and context — use delimiters to keep everything organized. Delimiters are simply markers that separate different parts of your prompt.
Bad prompt:
"Summarize this text and make it sound professional. Our Q3 revenue grew by 15% compared to Q2, primarily driven by our new SaaS product line which exceeded targets by 22%. Customer acquisition cost dropped from $45 to $31 per user. However, enterprise client retention fell to 87% from 92% last quarter."
Better prompt:
"I need you to do two things with the text below.
Instructions: Summarize the key metrics and write a 3-sentence executive summary suitable for a board meeting.
---
Text to analyze: Our Q3 revenue grew by 15% compared to Q2, primarily driven by our new SaaS product line which exceeded targets by 22%. Customer acquisition cost dropped from $45 to $31 per user. However, enterprise client retention fell to 87% from 92% last quarter.
---
Tone: Professional, concise, data-driven."
Delimiters (like the dashes above) make it crystal clear what is an instruction and what is the content to work with.
---
Technique 10: Ask for Alternatives
Do not settle for the first answer. Asking ChatGPT for multiple options gives you more creative choices and often sparks ideas you would not have thought of on your own.
Bad prompt:
"Write a subject line for my email newsletter."
Better prompt:
"Write 10 subject lines for my email newsletter about our upcoming Black Friday sale on online courses. Give me a mix of styles: 3 that use urgency, 3 that use curiosity, 2 that use humor, and 2 that are straightforward. Keep each under 50 characters."
Now instead of one mediocre subject line, you have 10 options across different styles. You can mix, match, and refine from there.
---
Technique 11: Specify Your Audience
Who is the content for? This single piece of information changes everything about how ChatGPT writes — the vocabulary, the complexity, the examples it uses, and the assumptions it makes.
Bad prompt:
"Explain how machine learning works."
Better prompt for a technical audience:
"Explain how supervised machine learning works to a data science team that is familiar with statistics but new to ML. Include references to concepts like regression, feature engineering, and model evaluation."
Better prompt for a general audience:
"Explain how machine learning works to a small business owner who has never written a line of code. Use everyday analogies and real-world examples of how ML is already being used in retail businesses. Avoid all technical jargon."
Same topic. Wildly different outputs. The audience specification is the difference.
---
Technique 12: Set the Tone
Tone controls how the content feels. Professional, casual, humorous, empathetic, authoritative — each tone creates a different experience for the reader.
Bad prompt:
"Write a social media post about our new product."
Better prompt:
"Write an Instagram caption announcing our new organic skincare line. Tone: playful and confident, like a best friend who happens to be a dermatologist. Use short sentences. Include 2-3 relevant emojis. End with a call to action that feels natural, not salesy. Maximum 200 words."
When you name the exact tone you want — or better yet, describe it with a comparison — ChatGPT nails it much more consistently.
---
Technique 13: Break Complex Tasks Into Steps
When you need ChatGPT to do something complicated, explicitly telling it to think step by step improves the quality of the output dramatically. This works especially well for analysis, planning, and problem-solving.
Bad prompt:
"Help me launch my podcast."
Better prompt:
"I want to launch a podcast about personal finance for people in their 20s. Walk me through this step by step:
Step 1: Help me choose a podcast name (give me 5 options with pros and cons)
Step 2: Outline the format and structure of a typical episode
Step 3: Create a list of equipment I need on a $200 budget
Step 4: Give me 10 episode ideas for the first season
Step 5: Create a launch checklist with specific deadlines for a 30-day timeline
Complete each step thoroughly before moving to the next."
Breaking it into steps keeps the output organized and ensures nothing gets missed. It also makes it easier for you to give feedback on individual sections.
---
Technique 14: Use Negative Prompts
Telling ChatGPT what you do NOT want is just as important as telling it what you do want. Negative prompts eliminate common problems before they happen.
Bad prompt:
"Write a cover letter for a software engineering job."
Better prompt:
"Write a cover letter for a mid-level software engineering position at a startup. Here is what I want to avoid:
- Do NOT start with 'I am writing to express my interest in...'
- Do NOT use cliches like 'passionate,' 'team player,' or 'go-getter'
- Do NOT make it longer than 300 words
- Do NOT use overly formal language
Instead, start with a compelling hook about a specific technical challenge I solved. My background: 4 years experience in Python and React, built a real-time analytics dashboard used by 10,000 users."
The negative prompts prevent all the default, template-style writing that makes cover letters blend into the pile. The result is something fresh and memorable.
---
Technique 15: Verify and Fact-Check with AI
This is a meta-technique that most people skip. After ChatGPT gives you information, you can use it to verify that information — essentially asking it to check its own work.
After getting any factual output, follow up with:
"Review the response you just gave me. Are there any claims that might be inaccurate, outdated, or oversimplified? Flag anything that I should double-check with an external source."
You can also use prompts like:
- "What are the counterarguments to what you just said?"
- "Rate your confidence level (1-10) for each claim you made."
- "What important nuances or exceptions did you leave out?"
Important note: This does not replace real fact-checking. ChatGPT can still be wrong about things it is confident about. But this technique catches a surprising number of errors and gives you a starting point for verification.
---
Putting It All Together: A Master Prompt Example
Let us combine several techniques into one powerful prompt. Watch how the techniques stack:
"You are a senior content strategist at a B2B SaaS company [Role - Technique 5]. I need you to write a LinkedIn article about the future of customer success in the AI era [Specificity - Technique 1].
Context: Our company sells customer success software to mid-market companies (100-500 employees). Our audience on LinkedIn is primarily VP-level customer success leaders [Context - Technique 2, Audience - Technique 11].
Requirements:
- Length: 800-1,000 words [Constraint - Technique 6]
- Tone: Thought-leadership style — confident but not arrogant, data-informed but accessible [Tone - Technique 12]
- Format: Start with a bold opening statement, use H2 headers to break sections, end with 3 actionable takeaways in bullet points [Format - Technique 4]
- Do NOT use the words 'synergy,' 'leverage,' or 'game-changer.' Do NOT include generic advice like 'embrace AI' [Negative Prompts - Technique 14]
Here is a LinkedIn article I admire for its style: [paste example] [Examples - Technique 3]
After writing, flag any statistics or claims that should be verified [Verification - Technique 15]."
That single prompt uses 8 of the 15 techniques. The output from a prompt like this will be dramatically different from "write me a LinkedIn article about customer success."
---
Common Mistakes to Avoid
Before you go, here are a few traps that even experienced prompt writers fall into:
- Being too polite or too wordy. You do not need to say "please" or write a paragraph of backstory. Be direct. ChatGPT does not have feelings to hurt.
- Asking for too much at once. If your prompt is 500 words long and covers 10 different things, break it into smaller prompts using the chaining technique.
- Accepting the first output. Always iterate. The first response is a draft, not a final product.
- Not saving your best prompts. When you create a prompt that works brilliantly, save it somewhere. Build a personal prompt library over time.
- Forgetting to specify what "good" looks like. If you do not define success criteria, you cannot blame ChatGPT for missing the mark.
---
The Bottom Line
Learning how to write better ChatGPT prompts is not about memorizing a formula. It is about developing a new way of thinking — a way of communicating that is clear, specific, and intentional. These 15 techniques are your toolkit. You do not need to use all of them in every prompt. But the more of them you apply, the better your results will be.
Start with the technique that addresses your biggest frustration. If your outputs are too generic, focus on Techniques 1 and 2 (specificity and context). If the tone is always off, start with Techniques 5 and 12 (role assignment and tone setting). If you are working on complex projects, master Techniques 8 and 13 (chaining and step-by-step).
The people who get the best results from AI are not the ones with the fanciest tools or the most expensive subscriptions. They are the ones who have learned to communicate with AI effectively. And now, you have the techniques to do exactly that.
Your next step: Pick one technique from this list and try it on your very next ChatGPT conversation. Notice the difference. Then add another technique. And another. Within a week, you will be getting results that make other people say "how did you get AI to do that?"
That is the power of great prompts.
Written by Saad A
AI Expert Instructor with experience at Deloitte, PwC, BMO, and Microsoft. Teaching 24,318+ students worldwide.
Ready to master AI?
Our Complete AI Bootcamp covers prompt engineering, ChatGPT, MidJourney, vibe coding, AI agents and more — with 110+ video lessons and 2,000+ prompts.