2.1 | Prompt Engineering: 6 strategies for better results

What you already know

  • AI models respond to text-based inputs (prompts)
  • The quality of the prompt directly affects the quality of the response
  • Basic principles of prompt formulation
  • Simple AI usage with the provided assistants

What you will learn in this module

  • Six professional strategies for more effective prompting
  • How to clearly formulate goals, format, and context
  • The new prompting paradigm for modern AI models
  • The limits of prompting and how you can overcome them
  • Practical application in your daily work

1. Why this module is important

An AI model is not a mind reader. Good prompts are the key to useful results. And: the rules of the game have changed. GPT-4 models & Co. now think in use cases – not just in words.

„The quality of your prompts determines the quality of the AI’s responses. With the right strategies, you can unlock the full potential of modern language models.“

As a Navigator, you have access to advanced AI functions and LLMs. To use them effectively, you need advanced prompting techniques that go beyond simple input prompts. These six strategies will help you achieve better results and use AI in a more targeted way.

2. The 6 Prompting Strategies for Professionals

1
Clarify Goal & Context

AI models need clear information to generate meaningful answers. State clearly:

  • What do you really want to achieve?
  • Who is the target audience?
  • In what format do you need the result?
Example:

Not optimal: „Write something about leadership.“

Better: „Write a LinkedIn post about 3 leadership mistakes in medium-sized companies. Target audience: CEOs. Tone: motivating.“

2
Role Prompting

Give the AI a role – this changes the perspective, tone, and depth of the answer. By defining a clear role, you create a context in which the AI can respond in a more specialized way.

Example:

„You are an experienced change coach. What would you advise a company that is in the middle of a transformation?“

3
Shot Prompting

Show the AI what a good result looks like. With this approach, you give the AI an example of the desired format or style. Your own texts often work better than external examples.

Example:

„Here is a text that I like: [insert sample text]. Write a similar one about psychological safety in a team.“

4
Activate Step-by-Step Thinking

Ask the AI to think in stages – this is particularly valuable for more complex tasks. This approach helps the AI to respond more logically and structured.

Example:

„Let’s proceed in a structured way. What would be the first three steps for a new HR strategy?“

5
Use Structured Prompts

Use clear formatting: Markdown, subheadings, bullet points. A well-structured prompt leads to an equally well-structured response.

Example:
# Task Create a draft for a training invitation # Target Audience Managers in medium-sized businesses # Tone Collegial, motivating # Length max. 100 words
6
Provide Quality Criteria

Specify how the result should sound or feel. The clearer your quality requirements, the better the AI can implement them.

Example:

„Use active language. No jargon. Goal: to be encouraging.“

3. Extra: The New Prompting Paradigm

With the new GPT models, a short prompt is no longer enough, as GPT now thinks in use cases. This change in perspective fundamentally alters how we should communicate with modern AI models.

Previously:

Short Prompt + Role = Response

Today (GPT-4+):

Format + Goal + Warning + Context = Better Output

Example according to the new schema:

„I need a checklist (Format) to prepare new team leads for AI-supported tools (Goal). It should be free of jargon (WARNING) because the target audience is not tech-savvy. I will use it in the onboarding process (CONTEXT).“

4. Practice: Testing Prompt Variations

Exercise: Prompt Optimization

Choose a real topic from your daily life. Try three of the six strategies with the same goal. Note down:

  • Which strategy yielded the best result?
  • What changed with context or role?
  • Which style works best for you?

5. xpand Tip

Our tip for practical use:

Build your own prompt library. Document effective prompts with their goal, format, and example output. This will make you faster, clearer, and more creative.

6. Knowing the Limits of Prompting

When prompting reaches its limits:

  • For highly specific technical knowledge that emerged after April 2023
  • When processing complex data that requires deeper analysis
  • For topics that require ethical considerations and human judgment
  • When interpreting non-textual information (like graphs, diagrams)

What helps then:

  • Break down tasks into smaller, more manageable parts
  • Incorporate external sources or expert knowledge when needed
  • Use specialized tools for data-intensive analyses
  • Always critically review and validate the results

Tip: Use the Prompting Optimizer Assistant on the xpandAI platform. It helps you refine difficult prompts and points out common pitfalls. This way, you can achieve better results even in challenging situations.

Your Takeaway

  • Prompting is a craft – with these six strategies, you’ll get significantly more out of the AI
  • Modern AI models think in use cases – adapt your prompts accordingly
  • A clear context, defined goal, and specific format will significantly improve your results
  • Test different approaches and build your own prompt library
  • Know the limits of prompting and be aware of when you need alternative approaches