Here's the most common mistake people make with AI: they type a vague question, get a vague answer, conclude that AI isn't very useful, and move on.
The problem isn't the AI. It's the prompt.
The difference between a good prompt and a bad one is the difference between getting something you'd actually use and getting something that needs starting from scratch. This guide shows you exactly what makes the difference — and gives you a framework you can use from today.
The vague vs specific problem
Let's look at the most common example — asking for help with an email.
The difference isn't length — it's specificity. The second prompt tells AI who this is for, what the situation is, what tone to use, and how long it should be. Four pieces of information that make an enormous difference.
The CRAFT framework
CRAFT is a simple framework for building better prompts. You don't need all five elements every time — but the more you include, the better the result.
What's the situation? Background information that helps AI understand what you actually need.
Who should AI be? "Act as a experienced HR manager" or "Write this as a financial journalist" focuses the response dramatically.
What do you want AI to do? Be specific — "write", "summarise", "compare", "explain", "generate five ideas".
How do you want the output? A bullet list, a table, three short paragraphs, a 200-word email — tell it.
How should it sound? Formal, casual, direct, empathetic, technical, plain English — specify it.
CRAFT in practice
Here's the same request written with CRAFT applied:
You could use this as-is and get something genuinely usable on the first attempt. That's the goal.
Iterate — don't start over
The best results usually don't come from one perfect prompt. They come from a conversation.
If the first response isn't quite right, don't delete and start again. Tell AI what to change:
- "Make it more concise — cut it to 80 words"
- "The tone is too formal — make it sound more human"
- "Add a section at the end about next steps"
- "The second paragraph is good but the first doesn't land — rewrite just that part"
This iterative approach — treating it as a conversation rather than a single request — is one of the biggest differences between people who find AI genuinely useful and people who don't.
Common mistakes to avoid
Being too vague
We've covered this — but it's worth repeating because it's the most common issue. If your prompt is one sentence with no context, you'll get a generic response.
Asking multiple questions at once
If you ask AI to do five different things in one prompt, the results will be uneven. Break complex tasks into steps.
Not specifying the audience
"Explain machine learning" gets a different response than "Explain machine learning to a 50-year-old who runs a small business and has no technical background". The second will be far more useful.
Accepting the first draft
AI's first response is a starting point, not a final product. Always read critically and ask for improvements.
Build a personal prompt library — a simple document where you save prompts that worked well. Over time it becomes one of your most valuable working tools. When you find a prompt that gets great results, save it with a note about what it's for.
A prompt to try right now
Open ChatGPT or Claude and try this. Paste it in exactly — then adapt it for your own situation:
Notice how much more useful the response is compared to just asking "what skill should I learn?".
- Specificity is the single most important factor in prompt quality
- CRAFT: Context, Role, Action, Format, Tone — use as many as needed
- Treat AI as a conversation, not a single request — iterate
- Specify the audience when the level of explanation matters
- Build a prompt library and save what works
Want to go much deeper on prompting?
Our AI Essentials and AI Advanced courses cover prompting in detail — including chain prompting, expert personas, and building your own prompt library. Verified certificate included.
View AI Essentials →