An Overview of prompting strategies

Structuring prompts effectively for AI is essential for getting the desired outcomes. Different prompt frameworks can help you craft prompts that are clear, concise, and likely to yield relevant and accurate responses.

Background

These strategies were extracted for a client side UI that I am working on which utilises these structures to render forms based on a strategy notation (more on that later).

My interest in these is prompted by my work on midJourney where using image references rather than prompts I was able to extract a very unique and controlled set of outcomes. My observation from the exercise with MidJourney was that interaction models and a good understanding of the Model itself can significantly improve outcomes.

Here are some common frameworks and strategies for structuring prompts:

1. Role, Task, Format (RTF) Framework

Role: Define the role the AI should take (e.g., “As a teacher,” “Acting as a project manager,” “You are a software developer”).
Task: Clearly state what you want the AI to do (e.g., “explain the process of photosynthesis,” “outline a project plan,” “write a Python function”).
Format: Specify the format in which you want the response (e.g., “in a step-by-step guide,” “as a bulleted list,” “in code with comments”).
Example Prompt: “As a history teacher, explain the causes of World War I in a way that a high school student would understand, using a bulleted list.”

2. Context, Input, Output (CIO) Framework

Context: Provide context to help the AI understand the background or scenario (e.g., “In a business meeting,” “For a beginner in programming”).
Input: Specify any input or data that the AI should consider (e.g., “Given the following data: [data here],” “Using the Python language”).
Output: Define what you want the AI to produce (e.g., “create a summary,” “generate a solution,” “write a brief report”).
Example Prompt: “For a marketing campaign targeting young adults, create a social media post that highlights the product’s eco-friendly features.”

3. Problem, Solution, Execution (PSE) Framework

Problem: Describe the problem or challenge you need help with (e.g., “My website is loading slowly”).
Solution: Ask the AI to provide a solution (e.g., “What can I do to improve the loading speed?”).
Execution: Request actionable steps or advice on how to implement the solution (e.g., “Provide a step-by-step guide”).
Example Prompt: “My website’s loading time is very slow. What can I do to improve it? Provide a step-by-step guide.”

4. Situation, Action, Result (SAR) Framework

Situation: Outline the situation or scenario (e.g., “I’m preparing for a job interview in data science”).
Action: Specify what action you want the AI to take (e.g., “Help me prepare answers to common interview questions”).
Result: Define the desired result or outcome (e.g., “Provide detailed, well-structured responses”).
Example Prompt: “I’m preparing for a job interview in data science. Help me prepare answers to common questions related to machine learning, with detailed explanations.”

5. Who, What, How, Why (WWHW) Framework

Who: Define who the target audience or user is (e.g., “For a beginner programmer,” “For a team of designers”).
What: Describe what you need or want (e.g., “Explain the concept of loops in Python,” “Create a project timeline”).
How: Ask for the method or format (e.g., “using simple language,” “in a visual diagram”).
Why: Provide the reason or purpose to guide the AI’s focus (e.g., “to help them get started quickly,” “to ensure timely project delivery”).
Example Prompt: “For a beginner programmer, explain the concept of loops in Python using simple language to help them get started quickly.”

6. Situation, Task, Action, Result (STAR) Framework

Situation: Describe the situation or context (e.g., “I need to plan a marketing campaign for a new product”).
Task: Define the task or objective (e.g., “Develop a campaign strategy”).
Action: Ask for specific actions or steps (e.g., “Suggest target audiences and key messages”).
Result: Request the desired outcome (e.g., “Create a comprehensive campaign plan”).
Example Prompt: “I need to plan a marketing campaign for a new product. Develop a campaign strategy by suggesting target audiences, key messages, and create a comprehensive plan.”

7. Goal, Obstacles, Methods, Results (GOMR) Framework

Goal: State the goal you want to achieve (e.g., “I want to increase user engagement on our website”).
Obstacles: Identify any obstacles or challenges (e.g., “Our content is not being shared enough on social media”).
Methods: Ask for methods or strategies to overcome the obstacles (e.g., “Suggest ways to make our content more shareable”).
Results: Define the expected results or outcomes (e.g., “Provide a list of actionable tips to improve social media engagement”).
Example Prompt: “I want to increase user engagement on our website, but our content is not being shared enough on social media. Suggest ways to make our content more shareable and provide a list of actionable tips to improve engagement.”

8. Background, Problem, Objective, Solution (BPOS) Framework

Background: Provide background information or context (e.g., “We are a small startup in the tech industry”).
Problem: State the problem or challenge (e.g., “We struggle to attract top talent”).
Objective: Define the objective or goal (e.g., “We want to improve our hiring process”).
Solution: Ask for a solution or strategy (e.g., “Provide tips for attracting top talent, especially in a competitive market”).
Example Prompt: “We are a small startup in the tech industry struggling to attract top talent. We want to improve our hiring process. Provide tips for attracting top talent, especially in a competitive market.”

9. Problem, Analysis, Solution, Action (PASA) Framework

Problem: Define the problem or challenge (e.g., “Our customer churn rate is high”).
Analysis: Request an analysis of the problem (e.g., “What might be causing this high churn rate?”).
Solution: Ask for potential solutions (e.g., “Suggest strategies to reduce churn”).
Action: Request specific actions to implement the solution (e.g., “Provide a step-by-step plan to improve customer retention”).
Example Prompt: “Our customer churn rate is high. What might be causing this, and what strategies can we implement to reduce churn? Provide a step-by-step plan to improve customer retention.”

10. Scenario, Challenge, Approach, Outcome (SCAO) Framework

Scenario: Describe the scenario or context (e.g., “We are launching a new product next quarter”).
Challenge: Identify the challenge or obstacle (e.g., “We need to build brand awareness quickly”).
Approach: Ask for the approach or strategy (e.g., “What marketing strategies should we use to quickly build brand awareness?”).
Outcome: Define the desired outcome (e.g., “Create a plan that will maximize our reach within the first month”).
Example Prompt: “We are launching a new product next quarter and need to build brand awareness quickly. What marketing strategies should we use to achieve this? Create a plan that will maximize our reach within the first month.”

These frameworks can help structure your prompts more effectively, making them clear and actionable, which in turn leads to more accurate and relevant AI responses.