Key Takeaways
- Good prompts are delegation in text form: context, task, format, quality criteria.
- The most common mistake is one-liner thinking – brief briefings produce arbitrary results.
- Leaders have a natural advantage: they can delegate – prompting is the same competence.
- Reverse prompting turns the tables: the AI asks you the right questions first.
The sentence “AI only delivers mediocrity here” rarely says anything about the AI – and a lot about the briefing. My favorite exercise in workshops: the same task, once as a one-liner, once as a proper briefing. The quality difference convinces faster than any slide.
What Separates a Good Prompt from a One-Liner?
The same thing that separates good delegation from a shout across the hallway: Who am I, what is this about, what exactly do I need, in which format, for which audience, what does a good result look like? Those who give a new employee such a briefing get usable work – with AI it is exactly the same. That is why I like to say in keynotes: prompting is leadership competence in text format.
How Does Reverse Prompting Work?
Instead of writing the perfect briefing yourself, let the AI ask: “Ask me all the questions you need to solve this task excellently.” The result surprises even professionals – the AI uncovers requirements you had not thought of yourself. For beginners it is the fastest path to top results, for teams a wonderful shared aha moment in the workshop.
How Do You Scale Prompting Across the Organization?
- Build a prompt library: share and maintain the team's ten best briefings centrally.
- Lead by example: leaders who show their own prompts lower the barrier more than any guideline.
- Practice with real tasks: no training example motivates like one's own annoying routine task.
Which Five Building Blocks Belong in Every Good Prompt?
Anyone who uses these five building blocks by habit no longer needs a prompt collection from the internet:
- Role: who should answer – a specialist lawyer, a sales director, a critical editor?
- Goal and context: what do you need the result for, who reads it, what is at stake?
- Material: your own figures, drafts or minutes instead of general world knowledge.
- Format: length, structure, tone – ideally with one example.
- Quality criterion: how will you recognise a good result? Say it in advance, not afterwards.
The fifth point is the most frequently forgotten – and the most effective. It turns a vague request into an assignment with a yardstick.
How Do You Build a Prompt Library That Actually Gets Used?
Most collections die because they sit in the wrong place and belong to nobody. Four points change that:
- Start with ten real tasks from everyday work, not with a hundred theoretical ones.
- Store the collection where the work happens – in the team channel, not in an intranet archive.
- Name one owner per department who reviews contributions and cleans up.
- Do a spring clean once a quarter: whatever nobody has used for three months goes out.
Which Mistakes Do Leaders Make When Prompting?
Four patterns show up regularly: the request stays vague because it feels clear in your head. There is no example, although a single good example does more than three paragraphs of explanation. There is no second round – yet quality almost always emerges from following up. And sensitive content ends up in services that were never approved for it. The first three cost quality; the fourth can cost considerably more.
Planning an event? As a keynote speaker for artificial intelligence, digitalization and Microsoft 365, I bring technology topics to the stage in a way that sticks – hands-on, entertaining and immediately actionable. Book an AI keynote: content, formats and fee at a glance.
Frequently Asked Questions (FAQ)
Don't prompt techniques become outdated constantly?
The tools change, the principle remains: clear context and clear expectations deliver better results – that was true for every assistant and is true for every AI generation.
Do different departments need different prompt trainings?
The basics are identical, the examples are not: sales, HR and accounting ignite on completely different use cases. Good workshops therefore work with the participants' real tasks.
What about sensitive data in prompts?
Clear guardrails belong in every prompt training: which data classes may go into which tool? With enterprise solutions and anonymized examples, almost every use case can be mapped cleanly.
How long does it take to prompt well?
The basics settle after roughly two hours of deliberate practice on your own tasks. Routine follows within a few weeks – provided there are real occasions and not only training examples.
Do we need separate templates for every tool?
No. The building blocks work across tools. Differences mainly concern handling your own files and the approvals – clarify that once per tool, not per prompt.
