Writing good prompts: how to ask clearly and get better answers
The single biggest difference between people who find AI chatbots useless and people who find them indispensable is usually not the tool — it's the quality of the questions. A "prompt" is just the message you send, and small improvements to how you write it produce dramatically better answers. Here's how.
Rule 1: Be specific
Vague prompts get vague answers. Compare:
- Weak: "Write an email to my landlord."
- Strong: "Write a polite but firm email to my landlord asking him to fix a leaking kitchen tap. It's been dripping for three weeks and I reported it by phone on September 5th. Keep it under 150 words."
The second prompt works better because it answers the questions the chatbot would otherwise have to guess: tone, purpose, key facts, and length. Whatever details you put in, you get back in the answer.
Rule 2: Give context
Chatbots start with a blank slate every conversation. Tell them what they'd need to know if you were explaining the task to a helpful stranger:
- Who this is for. "Explain this for a 12-year-old" and "explain this for my accountant" produce completely different — and both correct — answers.
- Your situation. "I'm a beginner cook with a small kitchen and no fancy equipment" steers advice toward recipes you'll actually make.
- The goal. Are you informing, persuading, entertaining, apologizing? Say so.
Rule 3: Ask for a format
Don't accept a wall of text when a table or list would do. Try:
- "Give me the pros and cons in a two-column table."
- "List the steps as a numbered checklist."
- "Summarize this in 3 bullet points, then give me one paragraph of detail."
- "Keep the whole answer under 200 words."
Rule 4: Show an example
If you want a particular style, paste a sample: "Write the announcement in a tone like this: [paste your example]." A few sentences of example teach the chatbot more than a paragraph of instructions. This works for humor, formality, brand voice — anything where "how it sounds" matters as much as "what it says."
Rule 5: Iterate — the first answer is a draft
This is the habit that changes everything. Treat the first response as a starting point and follow up:
- "That's too formal — make it warmer."
- "Good, now cut it in half."
- "You missed the main point: the meeting is about the budget. Try again."
- "Explain the second point in more detail."
Every follow-up narrows the gap between what you got and what you wanted. Two or three rounds of back-and-forth routinely beat one perfect first prompt.
Rule 6: One job per prompt
"Write my wedding speech, plan the honeymoon, and design the seating chart" gets you three mediocre results. Ask for one thing, get it right, then move on. The chatbot doesn't get bored — you can afford to go one step at a time.
What not to worry about
You don't need magic words, elaborate formatting, or "prompt engineering" courses. Plain, clear instructions in your own words are exactly what these tools respond to best. Write the way you'd brief a competent colleague, and you'll be in the top tier of users immediately.
Common beginner mistakes
- Asking yes-or-no questions when you want an explanation. "Is this a good deal?" gives you a guess. "What factors should I consider when judging this deal?" gives you a framework you can reuse forever.
- Forgetting to say what "good" means. "Write a birthday message" is fine, but "write a birthday message for my 70-year-old mother — warm, a little funny, no jokes about aging" is ten times better.
- Quitting after one bad answer. If the first response misses, say exactly what was wrong and try again. Most "the AI is useless" stories end at attempt one; most "the AI is amazing" stories include attempts two and three.
- Over-explaining the technology, under-explaining your need. You don't need to know how the model works. You do need to know what you want. Spend your words on the outcome, not the mechanism.