Most people get mediocre results from AI because they treat prompts like casual questions and accept the first answer as final.
The difference between a vague response and a useful one often comes down to how you structure your request—and how many iterations and refinements you apply after that first answer.
A strong prompt reads like a short brief, not a chat. It tells the model what to do, what to focus on, who it’s for, and how the answer should look. Here is how to make your prompt better:
Five parts of a strong prompt
- State the task. “Summarise this news story in five key takeaways.”
- Add specific instructions. What should it focus on? What should it highlight? "Focus on the casualties and government response."
- Provide context. Who is this for? How will you use the output (slide deck, newsletter, interview prep)? ""
- Set constraints. Tone, length, format, what to avoid (“no title case”, “no first person”, “concise and direct”).
- Re-emphasise the goal. Remind the model of the most important requirement before you hit send.
Make it work harder for you
- Assign a role when it helps: “Act as a news editor coach” or “Think like a newsroom managing director.”
- Attach or copy paste examples and structured requests for complex tasks.
- Encourage step‑by‑step reasoning for analysis, interview prep, or documentation. "Generate the news brief first then wait for more instructions to proceed."
Don't skip the human part
- AI is a word machine, not a truth machine.
- Always fact‑check, verify claims, and compare outputs against your source material.
- Treat the first answer as a draft, not a final product.
A simple template you can reuse
Task: [What you want]
Instructions: [Focus and priorities]
Context: [Audience, purpose, intended use]
Source material: [Article, notes, list]
Constraints: [Tone, style, length, exclusions]
Priority: [Repeat the main goal and intended use]
Refine the prompt based on the first response, then save your best version as a reusable template.
