HomeCasa How to write effective prompts
Ten practical principles to write prompts that work on the first try more often: clarity, specificity, examples, format and review.
There is no secret formula, but there are principles that consistently improve results. These are the ones that make the biggest difference.
1. Be specific
“Make it better” means nothing to an AI. “Make it shorter, with sentences under 20 words and a concrete example in each paragraph” does.
2. Provide the missing context
Think about what a human professional would need to know to do the job well, and add it: audience, goal, situation, data.
3. Explain why
If you say why you want something, the AI can make better decisions. “Use short sentences because people will read this on their phones” works better than “use short sentences”.
4. Show an example
One example is worth many style instructions. If you have a text you like, include it as a reference and say what to imitate (tone, structure) and what not to (content).
5. Ask for a specific format
Table, numbered list, JSON, Markdown, email… A clear format makes the answer easier to review and reuse.
6. Split complex tasks
Instead of asking for a full business plan in one message, ask for the market analysis first, review it, then move on. Each answer will be more careful.
7. Separate instructions from data
When you paste long texts, delimit them with triple quotes, tags or code blocks so the AI does not confuse your text with your instructions:
Summarize the following article in 5 points:
"""
[article]
"""8. Let the AI ask
Add: “If you lack information to do this well, ask me before answering.” It prevents the AI from filling gaps with assumptions.
9. Ask it to flag uncertainty
“List which claims I should verify” or “separate facts from assumptions” reduces the risk of accepting errors as truth.
10. Iterate on the answer
The first result is a draft. Ask for specific changes: “the second paragraph is too technical, rewrite it for someone with no IT background”.
Summary
An effective prompt is clear, specific and complete. To see these principles applied, check the good and bad prompt examples or any prompt in the library.