The Core Principle: Specificity Beats Brevity
ChatGPT generates output based entirely on what you tell it. Vague prompts produce vague answers. Specific, contextual prompts produce specific, useful answers. Most people use 5–10 words when 30–50 would get dramatically better results. The extra 30 seconds of prompt writing saves many minutes of editing and re-prompting.
How to Write Better Prompts
- 1
Add context about who you are and what you need it for
“I am a high school teacher preparing a lesson” produces different output than “I am a marketing manager writing a report.” Tell ChatGPT your role, your audience, and why you need the output. This context shapes tone, vocabulary and level of detail appropriately.
- 2
Specify the format you want
Ask for a numbered list, a table, bullet points, a formal email, a casual text message, a step-by-step guide, or a summary in plain language. Without a format instruction, ChatGPT makes assumptions that may not match your needs. “Give me a bullet-point summary of the key risks, suitable for an executive who will read it in 30 seconds” is far better than “summarise the risks.”
- 3
Give examples of what you want (few-shot prompting)
If you want output in a specific style, show an example: “Here is an example of the tone I’m looking for: [paste example]. Now write something similar about [topic].” ChatGPT is exceptionally good at matching style and tone when given an example to follow.
- 4
Ask it to think step by step for complex tasks
For reasoning-heavy tasks (analysis, planning, problem-solving), adding “think through this step by step” or “reason through this before giving your answer” significantly improves accuracy. This is called chain-of-thought prompting and produces noticeably better results for complex questions.
- 5
Iterate and refine rather than accepting the first output
The first response is often a starting point. Follow up: “Make it more concise,” “Add a section on X,” “Rewrite the second paragraph in a more confident tone,” or “This is good but the ending feels weak — can you strengthen it?” ChatGPT remembers the full conversation, so refining iteratively is faster than rewriting prompts from scratch.
Practical Use Cases
- Drafting emails and messages: Describe the situation, recipient relationship, tone and goal
- Editing and improving your writing: Paste your draft and ask for specific improvements
- Research starting points: Get an overview, then verify details from primary sources
- Brainstorming: Generate options, names, ideas or angles you had not considered
- Code: Explain what you want the code to do, in what language, with what constraints
- Learning: Ask it to explain complex topics at different levels: “Explain compound interest as if I am 12 years old”