|
Most people use ChatGPT like a search engine: one question, one answer, done. That approach uses a small fraction of what the tool can actually do, and it is the reason so many people find the results generic or disappointing.
The difference between a mediocre result and a genuinely useful one is almost never the model. It is the prompt. The 15 techniques below take you from the basics of writing a good prompt to more advanced workflows, and every single one includes a copyable prompt and a description of what you will get back, so you can see the technique working rather than just reading about it.
You do not need a paid plan or any technical skill. You just need to change how you ask.
Foundational Techniques
1. Give ChatGPT a Role
Telling ChatGPT who to be sets the tone, vocabulary, and depth of its response. A request answered ‘as a financial advisor’ looks very different from the same request answered with no role at all. The role focuses the model on the right kind of expertise.
| COPY THIS PROMPT
You are an experienced financial advisor. Explain the difference between a Roth IRA and a traditional IRA to someone in their first job, in plain language, in under 150 words. |
| WHAT YOU GET
A concise, jargon-free explanation pitched at a beginner, framing each account around take-home pay and future taxes, rather than a generic textbook definition full of terms a first-time saver would not know. |
2. Add Context Every Time
ChatGPT does not know your situation unless you tell it. The more relevant context you give, the more tailored the result. Context is the single biggest lever most people ignore.
| COPY THIS PROMPT
I run a small bakery and want to email customers about a new weekend brunch menu. Tone: warm and casual. Audience: regulars who already love us. Keep it under 120 words and end with a friendly call to visit this weekend. |
| WHAT YOU GET
A short, warm email written for existing fans of the bakery rather than a generic promotional blast, with a natural closing invitation, because the model knew the audience, the tone, the length, and the goal. |
3. Specify the Format You Want
If you do not say how you want the answer structured, you get whatever ChatGPT defaults to. Naming the format, a table, a bulleted list, a step-by-step guide, a short paragraph, saves you from reformatting later.
| COPY THIS PROMPT
Compare three budgeting methods (50/30/20, zero-based, and envelope) for a beginner. Give the answer as a table with columns: method, how it works, best for, main drawback. Keep each cell short. |
| WHAT YOU GET
A clean comparison table you can scan in seconds, instead of three long paragraphs you would have to read and mentally reorganize yourself. |
4. Combine Role, Task, Context, and Format
The four techniques above work best together. A widely used shorthand for this is PTCF: Persona, Task, Context, Format. Including all four in one prompt removes almost all the ambiguity that causes weak results.
| COPY THIS PROMPT
Persona: You are a patient coding tutor. Task: Explain what a ‘for loop’ does. Context: I am a complete beginner who has never coded and learns best with everyday analogies. Format: One simple analogy, then a short plain-English explanation, then a tiny example. Under 200 words. |
| WHAT YOU GET
A beginner-friendly explanation that opens with a relatable analogy, avoids jargon, and ends with one small example, because every part of the request, who, what, background, and shape, was specified up front. |
Because prompt quality drives everything else in this article, it helps to know whether your own prompts actually give ChatGPT enough to work with. You can paste any prompt into the free Prompt Coverage Checker at optimizewithsanwal.com, which scores it for context, clarity, constraints, and structure, and shows you what is missing before you hit send.
Getting Better Answers
5. Ask It to Think Step by Step
For anything involving logic, math, analysis, or multiple steps, adding ‘think step by step’ or ‘show your reasoning’ measurably improves accuracy. It forces the model to work through the problem rather than jumping to an answer.
| COPY THIS PROMPT
A shop buys items for $8 each and sells them for $13. Monthly costs are $2,000. How many items must it sell per month to break even? Think step by step and show your reasoning before giving the final number. |
| WHAT YOU GET
The model lays out the profit per item, divides fixed costs by that margin, and arrives at the break-even quantity, with each step visible so you can check the logic rather than trusting a single number. |
6. Give It an Example of What You Want
Showing ChatGPT a sample of the style or structure you are after, called giving it an example, is often faster than describing it. The model matches the pattern you provide.
| COPY THIS PROMPT
Rewrite these product descriptions in the style of this example. Example: ‘Built for early mornings and long days. This mug keeps coffee hot for hours and fits every car holder.’ Now rewrite: ‘Stainless steel water bottle, 750ml, double-walled, keeps drinks cold.’ |
| WHAT YOU GET
A rewritten description that mirrors the benefit-led, conversational rhythm of your example, rather than a generic spec sheet, because the model had a concrete pattern to follow. |
7. Ask for Multiple Options
Instead of accepting the first answer, ask for several variations at once. This is faster than regenerating repeatedly and lets you pick and combine the best parts.
| COPY THIS PROMPT
Give me 5 different subject lines for an email announcing a 20% off weekend sale at an online plant shop. Vary the tone: one urgent, one playful, one simple, one curiosity-driven, one benefit-led. Keep each under 8 words. |
| WHAT YOU GET
Five distinct subject lines in clearly different tones, so you can choose the one that fits your brand or blend two, instead of committing to a single option you then have to talk yourself into. |
8. Tell It What NOT to Do
Negative constraints are as useful as positive ones. Naming what to avoid, jargon, hype words, a certain length, prevents the most common problems before they happen.
| COPY THIS PROMPT
Write a 100-word LinkedIn post about finishing a marathon. Do not use hype words like ‘journey,’ ‘grind,’ or ‘blessed.’ No hashtags. No emojis. Keep it plain and human. |
| WHAT YOU GET
A grounded, readable post that avoids the tired language most marathon posts fall into, because you told it specifically what to steer clear of, not just what to include. |
Working in a Conversation
9. Iterate Instead of Restarting
ChatGPT remembers the current conversation. Instead of writing a fresh prompt when something is not quite right, refine what it already produced. Each round builds on the last.
| COPY THIS PROMPT
[After it drafts something] Good start. Make the second paragraph shorter, change the tone to be more confident, and replace the closing line with a question. |
| WHAT YOU GET
A revised version that keeps everything that worked and changes only what you flagged, which is faster and more precise than describing the whole thing again from scratch. |
10. Ask It to Ask You Questions
Flip the usual direction. Have ChatGPT interview you before it produces anything. This surfaces details you would not have thought to include and dramatically improves the result.
| COPY THIS PROMPT
I want to write a cover letter for a marketing role. Before you write anything, ask me up to 6 questions that will help you write the strongest possible letter. Ask them one at a time. |
| WHAT YOU GET
A short interview that pulls out the specific achievements, company details, and motivations that make a cover letter distinctive, so the final draft is built on your real story rather than generic filler. |
11. Have It Critique Its Own Work
After ChatGPT produces something, ask it to review the output against clear criteria. It will often catch weaknesses you would have to spot manually.
| COPY THIS PROMPT
Review the email you just wrote. Is anything unclear, too long, or likely to be misread by a busy manager? List the issues, then give me one improved version. |
| WHAT YOU GET
A short list of specific weaknesses followed by a tightened rewrite, giving you both the reasoning and the improved draft in a single step. |
12. Use It as a Thinking Partner
ChatGPT is useful for pressure-testing your own ideas, not just producing content. Ask it to argue the other side, find holes, or stress-test a plan.
| COPY THIS PROMPT
I am thinking of quitting my job to freelance full time. Play devil’s advocate. What are the strongest arguments against doing this right now, and what questions should I answer before deciding? |
| WHAT YOU GET
A structured set of counter-arguments and decision questions that helps you think more clearly, rather than an answer that simply agrees with whatever you proposed. |
Advanced Workflows
13. Build a Reusable Custom GPT or Saved Instructions
If you do the same kind of task often, save your role, tone, and format once instead of retyping it. Custom Instructions (available to all users in settings) and Custom GPTs (on paid plans) both let you set persistent preferences ChatGPT applies automatically.
| COPY THIS PROMPT
In Custom Instructions, set: ‘Always respond in plain UK English. No hype words. Prefer short paragraphs and bullet points. When I ask for writing, give me two options unless I say otherwise.’ |
| WHAT YOU GET
Every future response follows your standing preferences without you repeating them, turning a five-line prompt into a one-line request because the setup is remembered. |
14. Break Big Tasks Into Layers
For complex work, do not ask for everything at once. Build it in stages: outline first, then expand each section, then refine. Layering produces far better results than a single giant prompt.
| COPY THIS PROMPT
Step 1: Give me an outline for a 1,000-word beginner guide to composting at home. Just the outline. [Then, after reviewing] Step 2: Now write the ‘Getting Started’ section from that outline in a warm, practical tone. |
| WHAT YOU GET
A guide that is coherent and well-structured, because you shaped the skeleton before any prose was written, rather than a long undifferentiated block you have to restructure afterward. |
15. Always Verify Facts, Stats, and Quotes
This is the most important technique of all. ChatGPT can state false information, invented statistics, fake quotes, wrong dates, with complete confidence. Anything factual must be checked against a reliable source before you rely on it. Treat ChatGPT as a fast drafting partner, not a source of truth.
| COPY THIS PROMPT
For any factual claims, statistics, or quotes in your answer, mark them clearly and tell me which ones I should independently verify before publishing. |
| WHAT YOU GET
A response that flags its own factual claims so you know exactly what to double-check, which is far safer than assuming everything stated confidently is accurate. It is not. |
| The Golden Rule
ChatGPT is a drafting partner, not a fact-checker. It generates fluent, confident text whether or not the underlying facts are true. Use it to write faster, brainstorm wider, and think more clearly, but verify every specific fact, statistic, date, and quote against a real source before you publish or rely on it. This one habit prevents the single most damaging mistake people make with AI. |
Mistakes That Waste Your Time
- Sending vague one-line prompts and blaming the model for generic results. The fix is context and structure.
- Starting a new chat every time instead of refining within the same conversation, which loses all the context you built up.
- Accepting the first answer. The first draft is a starting point, not the finished product.
- Trusting facts without checking. Confident tone is not evidence of accuracy.
- Sharing sensitive personal, financial, or confidential information. Treat anything you type as potentially retained; do not paste passwords, private client data, or secrets.
Key Takeaways
- The quality of ChatGPT’s output is determined mostly by the quality of your prompt, not the model or the plan.
- The four foundations are role, context, format, and specificity. Combined, they are often called the PTCF framework.
- For anything analytical, ask it to think step by step. For style, give it an example to follow.
- Work in a conversation: iterate instead of restarting, have it ask you questions, and have it critique its own work.
- Save time on repeat tasks with Custom Instructions or a Custom GPT, and break big tasks into layers.
- Always verify facts, statistics, and quotes. ChatGPT states false information with full confidence.
- Never paste passwords, confidential, or sensitive personal data into any AI chat.
Frequently Asked Questions
How do I get better answers from ChatGPT?
Give it more to work with. Tell it what role to play, add context about your situation, specify the format you want, and be specific about the task. A prompt that includes all four produces dramatically better results than a vague one-line request. Then refine the answer in the same conversation rather than starting over.
What is the PTCF framework?
PTCF stands for Persona, Task, Context, and Format. It is a simple structure for writing effective prompts: tell ChatGPT who to be (Persona), what to do (Task), the background it needs (Context), and how to present the answer (Format). Including all four removes the ambiguity that causes weak results.
Does telling ChatGPT to ‘think step by step’ actually help?
Yes, for tasks involving logic, math, analysis, or multiple steps. Asking the model to show its reasoning before answering measurably improves accuracy, because it works through the problem rather than jumping straight to a conclusion. For simple factual or creative requests, it makes less difference.
Can I trust the facts ChatGPT gives me?
Not without checking. ChatGPT can state false information, invented statistics, fake quotes, and wrong dates with complete confidence. Always verify any specific factual claim against a reliable source before relying on it. Use ChatGPT as a drafting and thinking partner, not as a source of truth.
Do I need a paid ChatGPT plan to use these techniques?
No. Almost all of these techniques work on the free plan, since they are about how you write prompts, not which model you have. The main paid-only feature mentioned here is Custom GPTs, but the free Custom Instructions setting covers most of the same benefit for saving your preferences.
Is it safe to put personal information into ChatGPT?
Be cautious. Avoid pasting passwords, financial details, confidential work data, or sensitive personal information into any AI chat, as you cannot fully control how that data is stored or used. For general questions and drafting, it is fine; for anything private or confidential, leave the specifics out.
Conclusion
Using ChatGPT effectively is a learnable skill, and it has almost nothing to do with your subscription tier. The people who get the most from it are the ones who understand that a good prompt is a collaboration: you supply the role, context, format, and specifics, and the model supplies the speed and range.
Start with the four foundations, role, context, format, and specificity, because they alone will transform your results. Add step-by-step reasoning for analytical tasks, learn to iterate inside a conversation, and save your preferences for work you repeat. Then, above everything else, verify the facts.
Pick two or three techniques from this list and use them today. That is genuinely all it takes to go from generic answers to output you can actually use.
Must Reads
- Claude vs ChatGPT: Which Is Better in 2026?
- Who Owns ChatGPT? OpenAI’s Ownership Explained
- How to Cancel Your ChatGPT Subscription (All Devices)
About the Author
I’m Sanwal Zia, an SEO strategist with more than six years of experience helping businesses grow through smart and practical search strategies. I created Optimize With Sanwal to share honest insights, tool breakdowns, and real guidance for anyone looking to improve their digital presence. You can connect with me on YouTube, LinkedIn, Facebook, Instagram, or visit my website to explore more of my work.
Disclaimer
All information published on Optimize With Sanwal is provided for general guidance only. Users must obtain every SEO tool, AI tool, or related subscription directly from the official provider’s website. Pricing, regional charges, and subscription variations are determined solely by the respective companies, and Optimize With Sanwal holds no liability for any discrepancies, losses, billing issues, or service-related problems. We do not control or influence pricing in any country. Users are fully responsible for verifying all details from the original source before completing any purchase.