How to Craft a Prompt

Master the essential elements of effective prompt engineering

The Anatomy of a Great Prompt

A well-crafted prompt combines multiple elements to guide the AI toward producing exactly what you need. Think of it as a recipe — each ingredient serves a specific purpose.

Essential Elements

1

Clarity: Be Specific

Vague prompts produce vague results. Avoid open-ended questions and instead provide clear, specific instructions.

❌ Vague:

"Tell me about customer service."

✓ Specific:

"List 5 best practices for reducing customer support ticket response times in SaaS companies."

2

Actor: Define the Role

Tell the LLM who they should be acting as. This shapes the perspective, expertise level, and tone of the response.

Examples:

  • "You are an experienced software architect with expertise in microservices..."
  • "Act as a friendly customer support agent for a fitness app..."
  • "You are a detail-oriented financial analyst reviewing quarterly earnings..."
3

Context: Provide Background

Supply relevant background information that helps the AI understand the situation, audience, or domain.

Example with Context:

"Our company is a B2B SaaS startup with 50 employees selling project management software to construction firms. We've noticed a 30% churn rate after the first year. Analyze potential causes and suggest retention strategies."

The context (industry, size, product, problem) allows the AI to provide targeted, relevant advice.

4

Constraints: Set Boundaries

Provide guidance on limits to ensure outputs are appropriately scoped and formatted.

Common Constraints:

Length:

  • • "Limit to 3 paragraphs"
  • • "Maximum 500 words"
  • • "Provide exactly 5 bullet points"

Tone:

  • • "Use language a minimum wage employee would understand"
  • • "Professional, not casual"
  • • "Avoid technical jargon"

Example:

"Explain cloud computing in 2 paragraphs using simple language that a retail store manager with no tech background would understand."

5

Intent: Explain Your "Why"

Explaining why you're asking helps the AI understand the ultimate goal and provide more relevant answers.

Without Intent:

"Explain how APIs work."

With Intent:

"Explain how APIs work because I need to brief my non-technical executive team on why we should invest in building an API for our product."

The second version will likely focus on business value and executive-level benefits, not technical implementation details.

6

Small Pieces: Break Down Complex Tasks

Don't ask the AI to do huge tasks in one go. Instead, ask it to break down the work, then iterate through each piece.

Planning Approach:

"I need to create a comprehensive marketing plan for launching a new mobile app. First, break this down into 5-7 major components we should address. Then we'll tackle each one individually."

This approach produces better results than asking "Write a comprehensive marketing plan" in a single prompt.

7

Examples: Show Don't Just Tell

Provide sample inputs/outputs or items to include. This is called "few-shot prompting" and dramatically improves accuracy.

Example with Samples:

"Generate product descriptions in this style:

Input: 'Wireless Bluetooth Earbuds'

Output: 'Experience crystal-clear audio on the go with our premium wireless earbuds. 8-hour battery life. Perfect for workouts and commutes. Available in 3 colors.'

Now write descriptions for: 'Laptop Stand' and 'USB-C Hub'"

8

Output Format: Specify the Structure

Clearly state how you want the output formatted — this ensures you get results that are immediately usable.

Format Options:

  • Text format: "Provide the answer as a numbered list" or "Format as a table with columns for Name, Description, Price"
  • Data format: "Output as CSV" or "Format as JSON with keys: id, name, status"
  • Visual format: "Create a flowchart" or "Generate a diagram showing the process"
  • Document format: "Format as a professional email" or "Structure as a Word document with headings"

Putting It All Together

Complete Example Prompt:

[Actor] You are an experienced HR consultant.

[Context] Our tech startup (150 employees) is struggling with employee retention in our engineering department.

[Intent] I need to present recommendations to our executive team next week.

[Clarity] Identify the top 5 causes of engineering turnover in tech startups and provide specific, actionable solutions for each.

[Constraints] Keep each solution to 2-3 sentences. Use language suitable for executives without HR backgrounds.

[Output Format] Format as a table with columns: Cause | Impact | Solution | Estimated Cost

Ready to Craft Better Prompts?

Let us help you design prompts that consistently deliver high-quality results