Why are good prompts important?
A prompt is your request or instruction to the AI. The clearer and more specific you formulate, the more precise and useful the answer will be. Think of the prompt like a work instruction for a very capable but literal-thinking assistant.
The goal: With the right formulation, you save time and immediately get the desired result..
The Basic Rules for Effective Prompts
1. Be specific instead of vague
❌ Bad: "Help with numbers"
✅ Better: "Create a summary of Q3 sales figures with the key metrics"
❌ Bad: "Analyze this"
✅ Better: "Show the main trends of the last 6 months in the customer data"
2. Provide context
Why context is important: The AI better understands what it should work with.
Examples of good context:
"Based on the uploaded Excel file..."
"For our nursing team..."
"In the context of our product launch..."
3. Define the desired format
Format specifications help the AI understand how you need the result:
"As a table with 3 columns"
"In 5 bullet points"
"As an email draft"
"As a presentation slide"
"In simple language for patients"
The Ideal Prompt Structure
A perfect prompt consists of three elements:
1. WHAT should be done (Task)
"Analyze the sales data"
"Create a summary"
"Compare the two documents"
2. HOW should the result look (Format)
"As a table with the top 10 results"
"In 3 concise paragraphs"
"As a bullet-point list"
3. FOR WHOM is the result intended (Target audience)
"For management"
"For new employees"
"For customers without technical knowledge"
Example of a complete prompt:
"Analyze the sales data from the Excel file (WHAT) and create a clear table with the 5 most successful products (HOW) for the monthly presentation to management (FOR WHOM)."
Common Mistakes and How to Avoid Them
Too vague formulations
❌ Problem: "What does it say?"
✅ Solution: "Summarize the content of the contract in 5 bullet points and highlight the most important conditions"
❌ Problem: "Do something with the data"
✅ Solution: "Create a chart from the survey data showing customer satisfaction by age group"
Missing context
❌ Problem: "Write a text"
✅ Solution: "Write a blog post about our new product feature for our website visitors, approx. 300 words"
Unclear expectations
❌ Problem: "Analyze the document"
✅ Solution: "Identify the 3 most important risks in this project plan and suggest solutions"
Using Prompt Templates
basebox offers you help: In the sidebar you'll find prompt templates for common use cases.
Typical templates:
Document analysis: "Summarize the document and..."
Data evaluation: "Analyze the table and show..."
Text improvement: "Revise the text for..."
How to use templates:
Click on a suitable template
Adapt it to your specific situation
Add your context
Advanced Prompt Techniques
Step-by-step instructions
For complex tasks:
"1. Analyze the sales data of the last 12 months
2. Identify the 3 strongest trends
3. Create action recommendations for the next quarter"
Provide examples
Show the AI what you want:
"Create product descriptions in the style of: 'Our clinic combines...' - use a similarly emotional tone"
Assign roles
Let the AI take a perspective:
"Act as a marketing expert and evaluate this campaign"
"From an employee's perspective: How understandable is this instruction?"
Iteratively Improve Prompts
If the result doesn't fit:
Be more specific: Add more details
Adjust format: Change the output form
Expand context: Provide more background information
Ask follow-up questions: "Can you explain that in more detail?"
Practical Examples by Use Case
Document analysis:
✅ Good: "Extract all dates and deadlines from the contract and list them chronologically"
Data evaluation:
✅ Good: "Show me the 5 months with the highest revenue and explain possible reasons"
Text creation:
✅ Good: "Write a professional email response to the customer that politely but firmly explains our position"
Presentation preparation:
✅ Good: "Create 3 key messages from this report for a 5-minute presentation to investors"
⚠️ Tip: Use the prompt templates in the basebox sidebar as a starting point and adapt them to your specific needs.
Experiment: Try different formulations - you'll quickly develop a feel for effective prompts.
Need help with prompt formulation? → Contact Support
