What is a Prompt? From Question to Instruction
You understand the difference between a question and an effective prompt.
Why this lesson matters
You open ChatGPT or Claude, type "write an email", and get something back. Sometimes usable, often not. You think: AI doesn't get me.
The AI is fine. The problem is in your question. Or rather: the lack of a good prompt.
The difference between "write an email" and a prompt that delivers exactly what you need? That's the difference between a vague request and a clear instruction. And that difference is what you learn in this lesson.
The difference between a question and a prompt
A question is what you'd ask a colleague at the coffee machine:
- "Could you write something about customer retention?"
- "Make a summary of this report"
- "Help me with a presentation"
A prompt is a structured instruction that tells the AI exactly what you expect:
- Who you are (or who the AI should be)
- What exactly you want
- Who it's for
- How it should look
- What must be left out
A prompt is an assignment, and that's something different from a question. The more specific your assignment, the more usable the result. Think of it this way: when you give an intern an assignment, you don't just say "do something with data" either. You explain what you expect.
Why specific instructions give better output
Keep reading
Leave your name and email address and you can read the rest
You get the whole lesson right away, and every other lesson stays open after that. No password, no confirmation email.
AI models are trained on billions of texts. When you ask a vague question, the model picks the most average answer. It gives you the answer that most often fits similar questions.
That means:
- Vague input → generic, predictable answer
- Specific input → targeted, usable answer
| Vague | Specific | Why it's better |
|---|---|---|
| "Write an email" | "Write a professional email of max 150 words to a customer who hasn't paid his invoice. Tone: friendly but clear." | The AI knows the exact length, tone and context |
| "Make a summary" | "Summarise this report in 5 bullet points for the management team. Focus on financial results and risks." | The AI knows the format, the audience and the focus |
| "Help with presentation" | "Create an outline for a 10-slide presentation on our Q3 results. Audience: the board. Style: based on the numbers, no jargon." | The AI knows the number of slides, the topic and the style |
Examples: bad vs good
Example 1: writing an email
Bad prompt:
Write an email to a customer.Good prompt:
Write a professional email to a customer (Jan de Vries, owner of an SME) who hasn't placed an order in 3 months. Goal: invite him to our new loyalty programme. Tone: warm and personal, not pushy. Max 200 words.Example 2: summarising a report
Bad prompt:
Summarise this.Good prompt:
Summarise the quarterly report below in an executive summary of no more than 300 words. Use this structure:
1. Key result
2. Main deviations from budget
3. Top 3 action points
Audience: a CFO who wants the essentials in 2 minutes.The most common beginner's mistake: assuming the AI "gets what you mean". It doesn't. AI reads exactly what you write. It has no idea what you're thinking. Be explicit.
The rules of thumb for a good prompt
- 1Be specific, say what you want rather than what you don't want
- 2Give context, background helps the AI pick the right register
- 3Define the format, bullet points? A table? Running text? Say so
- 4Set constraints, maximum length, tone, what must be left out
- 5Give an example, show what you expect if you can
Start every prompt with the role you want to give the AI. For example: "You're a financial analyst" or "You're an experienced copywriter". This steers the style and the knowledge level of the answer.
Quiz
Which prompt is most likely to give a usable result?
What's the main difference between a question and a prompt?
Want to keep your progress, get the practice files and sign up for the free evening? Create a free account. All you do is click a link in your email.
Create a free account