“Write me a marketing strategy” isn’t a prompt. It’s a wish.
Prompt engineering is the skill of communicating with AI in a way that produces repeatable, useful results. And contrary to what the LinkedIn gurus say, you can learn it in a single day.
Three rules of a good prompt
1. Context
AI doesn’t know your company, your client or your industry. You have to tell it.
Bad: “Write an email to a client.”
Good: “You are an account manager at an advertising agency. Write an email to a client who hasn’t responded to our proposal in 2 weeks. Tone: professional, but human.”
2. Format
Tell the AI what the answer should look like.
- How many paragraphs?
- Bullet points or continuous text?
- Formal or informal?
- In which language?
3. Constraints
The best prompts have clear boundaries.
- “Maximum 200 words”
- “Use only the data from the attached document”
- “Don’t use technical jargon”
A practical exercise
Take any repetitive task from your work and try to describe it as a prompt. Not one prompt, but three different ones. Compare the results. Repeat with a better prompt.
That’s the whole secret: iteration.
What’s next?
In my training sessions, we do this live, with participants’ real tasks. Not textbook examples, but your everyday work.
The result? After 4 hours, people walk out with ready-made prompts they can use on Monday morning.