Home › Guides
How to Write Better AI Prompts: A Beginner's Guide with Examples
The difference between a useless AI answer and a genuinely helpful one is often the prompt. A prompt is simply the instruction you give an AI assistant, but the way you write it shapes everything that comes back. This guide gives you a simple, repeatable framework for writing better prompts, explains why it works, and includes practical examples you can adapt for study, freelancing and small business tasks.
Why prompts matter
AI chat assistants generate responses based on patterns in their training and the context you give them. When your prompt is vague, the assistant has to guess what you want, and it usually guesses something generic. When your prompt is specific, it has the information it needs to produce something tailored. Think of it like briefing a new assistant on their first day: the clearer your instructions, the better the work.
The CRAFT framework
A simple way to remember the parts of a strong prompt is CRAFT: Context, Role, Action, Format and Tone. You do not need all five every time, but checking them helps.
- Context: background information. Who is the audience? What is the situation? What have you already tried?
- Role: who should the AI act as? A tutor, editor, marketing consultant, or critic?
- Action: the specific task. Summarize, compare, draft, rewrite, list, critique.
- Format: how the output should look. Bullet points, a table, 150 words, three options, an email.
- Tone: the style. Friendly, formal, simple, persuasive, neutral.
Weak prompt: "Write about social media marketing."
Stronger prompt: "I run a small home bakery in Lahore that sells custom cakes. Act as a social media consultant. Suggest 10 Instagram post ideas for the next month that would attract local customers. Format as a table with columns for idea, caption hook and best day to post. Keep the tone warm and simple."
The second prompt gives the AI enough to work with, and the output will be much more usable.
Technique 1: Give examples
If you want a specific style, show it. Paste one or two examples of writing you like and say "Match this style." This technique, sometimes called few-shot prompting, is one of the most reliable ways to control output. For example, paste two product descriptions from your store and ask for a new description in the same style for a different product.
Technique 2: Break big tasks into steps
Asking for a complete 2,000 word report in one go often produces shallow results. Instead, work in stages:
- Ask for an outline and refine it.
- Ask for one section at a time, giving notes for each.
- Ask for a review of the whole draft for gaps and repetition.
This keeps you in control and makes it easier to correct the direction early.
Technique 3: Ask it to ask you questions
When you are not sure what details matter, end your prompt with: "Before you start, ask me any questions you need to do this well." The assistant will often ask about audience, goals and constraints, which helps you think more clearly too.
Technique 4: Set constraints
Constraints reduce generic output. Examples:
- "Use no more than 100 words."
- "Avoid jargon; a 12-year-old should understand it."
- "Do not use the words 'unlock', 'elevate' or 'game-changer'."
- "Only use information from the text I pasted below."
The last constraint is especially useful for reducing made-up information when you are summarizing your own documents.
Technique 5: Ask for reasoning or alternatives
For decisions, ask the AI to lay out options with pros and cons rather than giving one answer. "Give me three ways to price this service, with advantages and risks of each." This makes you the decision maker and exposes assumptions you can check.
Technique 6: Iterate instead of starting over
The first answer is a draft. Follow up with targeted feedback: "Make the second paragraph more specific," "Shorter, please," "This sounds too formal, make it conversational," or "Point 3 is wrong because our shop is closed on Sundays." Conversation is where the quality improves.
Practical prompt examples
For students
"I am a first-year university student studying economics. Explain the concept of opportunity cost using two everyday examples from life in a big city. Then give me three short quiz questions to check my understanding, and wait for my answers before revealing the solutions."
For freelancers
"Here is a job post from a client [paste]. Here is my background: [paste]. Draft a proposal under 150 words that addresses the client's two biggest concerns, mentions one relevant project, and ends with a question about their timeline. Tone: confident and friendly, not salesy."
For small business owners
"I sell handmade leather wallets online. Write three versions of a product description for a slim bifold wallet, each under 80 words. Version one should focus on durability, version two on minimalist design, version three on gifting. Do not invent features; the wallet is made of cowhide, holds six cards and has no coin pocket."
Notice how that last prompt includes facts and explicitly tells the AI not to invent features. This protects you from accidentally publishing false claims.
For content creators
"Act as a YouTube scriptwriting coach. I am making a 6-minute video titled 'How I organize my week with free tools.' Create a script outline with a strong 15-second hook, three main sections and a closing call to action. Suggest where to add on-screen examples."
For job seekers
"Here is a job description [paste] and my CV [paste]. List the five most important skills the employer wants, and for each one tell me whether my CV shows it clearly, weakly or not at all. Suggest specific bullet point improvements, but do not invent experience I do not have."
Common prompting mistakes
- Being too vague: "Help me with marketing" gives the AI nothing to work with.
- Overloading a single prompt with ten unrelated tasks.
- Accepting the first answer without review.
- Trusting facts without checking. AI can produce confident but wrong statements, called hallucinations.
- Sharing sensitive personal or business data in public tools.
Checking AI output
Good prompting reduces errors but does not eliminate them. Build a quick review habit:
- Verify facts, numbers, dates and names with reliable sources.
- Check that the output actually answers your question.
- Read it aloud to see whether it sounds natural.
- Make sure it matches your real situation; remove anything invented.
Saving prompts that work
When a prompt gives great results, save it in a notes file or document as a template with placeholders, such as [AUDIENCE], [PRODUCT] and [TONE]. Over time you will build a personal prompt library that saves hours. Many AI assistants also let you set custom instructions or create reusable assistants, so you do not have to repeat background details every time.
Do prompts work the same in every AI tool?
The core principles of clarity, context and examples work across ChatGPT, Gemini, Claude, Copilot, Grok and other assistants. Each tool has strengths and its own features, such as web search, file uploads or image generation, so it is worth testing the same prompt in two tools when the result matters. Image generation tools have their own prompting styles, focusing on subject, style, lighting and composition.
Final thoughts
Writing good prompts is less about secret tricks and more about clear communication. Give context, define the task, specify the format, and keep refining. Combine that with a habit of checking facts, and AI assistants become reliable helpers for studying, freelancing and running a business. Start by rewriting one prompt you used this week with the CRAFT framework and compare the results.
This article is for general information only. Tool features and free plans change often; please confirm details on each provider's official website.