Two people ask the same chatbot the same question. One gets gold; the other gets generic mush. The difference is almost always the prompt. Learning how to write good prompts for AI chatbots is the single highest-leverage AI skill — it improves everything you do with these tools, from work emails to holiday planning, without costing a penny.

Here’s the complete guide, from the basic formula to advanced techniques.

How to Write Good Prompts for AI Chatbots: The Core Formula

Every good prompt contains four ingredients:

1. Context: background the AI needs. Who you are, the situation, relevant facts.
2. Task: exactly what you want it to do. Be specific — “draft”, “explain”, “compare”, “brainstorm”.
3. Format: how you want the output. Bullet points? Table? 200 words max? Step-by-step?
4. Tone/audience: formal? Friendly? For a 10-year-old? For a board report?

Compare:

• Weak: “Write about email marketing.”
• Strong: “I’m a freelance florist in Manchester with 500 Instagram followers. Draft 5 Instagram caption ideas promoting Mother’s Day bouquets. Each under 50 words, warm tone, include a call to action. No emojis.”

The strong prompt gives context (who, where, what business), task (draft captions), format (5 ideas, under 50 words), and tone (warm, no emojis). The output difference is night and day.

Techniques That Level Up Your Results

Give examples (few-shot prompting)

Show the AI what “good” looks like: “Here are two headlines I like: [examples]. Write 5 more in the same style.” This beats describing the style in words.

Ask it to ask you questions

“I need a workout plan. Ask me questions one at a time to get the information you need, then create the plan.” This turns a vague request into a guided interview — great when you’re not sure what context matters.

Assign a role

“Act as a patient maths tutor” or “Act as a sceptical investor reviewing my pitch.” Roles focus the AI’s tone and perspective. (Take it with a pinch of salt for genuinely expert roles — it’s still pattern-matching, not expertise.)

Iterate, don’t restart

The first output is a draft. Follow up: “shorter”, “more formal”, “add an example”, “do it as a table instead.” Iteration is where the real value appears — most people quit after one attempt and conclude AI is useless.

Set constraints

Constraints breed creativity and usefulness: word limits, “no jargon”, “assume I know nothing about finance”, “give me exactly 3 options with pros and cons.”

Ask for reasoning

“Explain your reasoning step by step” improves answers to complex questions — and lets you spot where the logic goes wrong.

Common Prompting Mistakes

• Being vague, then blaming the AI. “Write me something about marketing” deserves what it gets.
• One-shot expectations. Treat the first answer as a starting point, always.
• No context about you. The AI doesn’t know you’re a beginner, a CEO, or a teenager unless you say so.
• Accepting the first draft. Especially for anything public-facing — edit ruthlessly.
• Trusting facts blindly. AI invents confidently. Verify anything important independently.

Prompts by Use Case

• Learning: “Explain [topic] to me like I’m 12, then quiz me with 5 questions.”
• Writing: “Here’s my rough draft: [text]. Improve clarity and flow, keep my voice, flag any weak arguments.”
• Decisions: “Help me decide between [A] and [B]. Ask me about my priorities first, then give a structured comparison.”
• Work: “Turn these meeting notes into: decisions made, action items with owners, and open questions.”

See these techniques applied in context: AI for emails, AI holiday planning, AI for CVs, and AI language learning.

Prompting for Specific AI Tools

The core formula works everywhere, but each tool has quirks:

• ChatGPT/Claude/Gemini (chat): conversational iteration shines — treat it as a dialogue, refining across turns. Upload files for context where supported.
• Image generators: visual language matters — style keywords (“cinematic,” “watercolour,” “isometric”), composition terms (“close-up,” “wide shot”), and negative prompts. See free AI image generators.
• AI search (Perplexity, etc.): ask for sources explicitly — “answer with citations” — and specify recency for time-sensitive topics.
• Coding assistants: describe the goal and constraints, paste relevant existing code, and ask for explanations alongside the code so you learn.
• Voice mode: speak naturally but structure your request — “I want to practise French. Correct my mistakes after each message.” Rambling voice prompts get rambling answers.

The Meta-Skill: Thinking Clearly

Here’s the secret nobody mentions: good prompting is mostly clear thinking. The discipline of specifying what you want, why, for whom, and in what format improves your thinking even before the AI responds. Many people report that learning to prompt well made them better at briefing humans too — writing clearer emails, giving better instructions, running better meetings.

It also teaches intellectual humility: when the AI misunderstands you, the fault is usually in your specification, not its intelligence. That feedback loop — specify, observe, refine — is genuinely educational. You’re not just learning to use a tool; you’re learning to think precisely.

Practice exercises

• Take a vague prompt you used this week and rewrite it with the four ingredients. Compare outputs.
• Ask the AI to critique your prompt: “How could I have asked this better?” (Genuinely useful feedback.)
• Try the same task with three different approaches (role-based, example-based, constraint-based) and compare.
• Teach someone else. Explaining prompting cements it faster than any amount of solo practice.

Prompting Pitfalls: When Good Prompts Go Wrong

• Over-specifying can strangle creativity — for brainstorming, leave room: “surprise me” is a valid instruction.
• Under-specifying then complaining about generic output — the most common failure, and always the prompter’s fault.
• Leading the witness: “confirm that X is true” gets confirmation, not truth. Ask neutrally: “what’s the evidence for and against X?”
• Forgetting to iterate — the first answer is a draft. Always.

For a hype-free view of what AI can and can’t do, BBC Technology covers major AI developments with a scepticism that marketing materials lack — useful background for anyone prompting regularly.

Practical Takeaway

Writing good prompts for AI chatbots comes down to a simple habit: context + task + format + tone, then iterate. Show examples, assign roles, set constraints, and never accept the first draft for anything important. Master this and every AI tool you touch gets dramatically more useful — it’s the closest thing to a universal AI skill there is. Start practising on something low-stakes today; within a week it’ll be second nature.