STAR Framework: Situation, Task, Action, Result

S.T.A.R. Framework

The STAR framework prompt engineering method gives you a simple, four-step structure for writing clear AI prompts: Situation, Task, Action, Result. Originally developed for behavioral interviews, STAR has been adapted by the prompt engineering community as one of the most intuitive ways to communicate with large language models. If you can describe where you are, what you need, how to get there, and what the finish line looks like, you can write a great prompt. For more on structured prompting approaches, see OpenAI's prompt engineering guide and Anthropic's prompt engineering documentation.

Introduction

The S.T.A.R. Framework (Situation, Task, Action, Result) is a beginner-friendly approach to prompt engineering that borrows its structure from the STAR method used in job interviews. In interviews, candidates use STAR to give structured answers about past experiences. In AI prompting, you flip the script: instead of describing what you did, you describe what you need the AI to do, grounded in a clear situation and aimed at a specific result.

This framework produces outputs that are:

• Contextually Grounded - Rooted in a specific scenario the AI can reason about
• Goal-Oriented - Focused on a well-defined task
• Methodologically Clear - Guided by explicit steps or approaches
• Outcome-Specific - Targeted at a concrete, measurable deliverable

The S.T.A.R. framework is particularly valuable for:

• Problem-solving prompts where background context is essential
• Decision-making scenarios with multiple factors to weigh
• Analytical tasks that benefit from a step-by-step approach
• Beginners who want a reliable structure without steep learning curves
• Adapting real-world business situations into effective prompts

Origin & Background

The STAR method was introduced by DDI (Development Dimensions International) in 1974 as a behavioral interview technique. Hiring managers discovered that asking candidates to describe a Situation, the Task they faced, the Action they took, and the Result they achieved produced far more useful answers than generic questions. The structure forced specificity and eliminated vague responses. Over five decades, STAR became the standard framework for behavioral interviews at companies ranging from startups to Fortune 500 firms.

Why STAR translates so well to AI prompting:

The same principle applies to language models. Vague prompts produce vague answers. When you give an AI a concrete situation, a focused task, a clear action path, and a defined result format, you eliminate ambiguity and get responses that are immediately useful. The prompt engineering community recognized this parallel and adapted STAR into one of the most accessible frameworks for beginners.

The interview-to-prompting bridge:

In interviews, the weight distribution is roughly 20% Situation, 10% Task, 60% Action, and 10% Result. In AI prompting, the balance shifts: Situation and Result carry more weight because the AI needs thorough context to reason well and precise output specifications to deliver usable results. People who already use STAR for interview prep find the transition to AI prompting nearly effortless. The mental model is identical: set the scene, define the challenge, describe the approach, specify the outcome.

How STAR differs from role-based frameworks:

While RACE leads with a professional Role and ERA leads with an Expertise domain, STAR leads with a Situation. This distinction matters. RACE is ideal when you need expert-level outputs from a specific persona. ERA is streamlined for quick expert consultations. STAR excels when the background context is more important than who is answering. If your prompt depends on understanding a nuanced business scenario, STAR is often the better choice.

How S.T.A.R. Compares

When to choose S.T.A.R.:

• Your prompt depends heavily on understanding background circumstances
• You need the AI to reason through a specific real-world scenario
• The situation involves multiple stakeholders, constraints, or variables
• You want a beginner-friendly framework that still produces strong results
• Your task is analytical, strategic, or decision-oriented

When to use something else:

• For tasks that require a specific professional persona, use RACE
• For quick expert consultations where situation context is minimal, use ERA
• For multi-phase strategic projects with phased outputs, use SCOPE
• For compliance-sensitive content with strict guardrails, use TAG

S.T.A.R. Framework Structure

1. Situation

Describe the current state and background circumstances

The Situation component sets the stage. Provide enough background for the AI to understand the environment, constraints, stakeholders, and any relevant history. Think of it as the briefing a consultant would need before they could give you useful advice.

Good examples:

• "Our e-commerce platform saw a 23% drop in conversion rate after migrating to a new checkout flow last month. Average order value remained stable, but cart abandonment spiked on the payment step."
• "A 50-person marketing agency is losing clients to freelancers who offer lower rates. The agency's strengths are strategy and analytics, but clients perceive the agency as slow."
• "Our mobile app has a 4.2-star rating, but the last 30 reviews mention crashes on Android 14 devices during photo upload."

Bad examples:

• "Sales are down" (no specifics)
• "We have a website" (irrelevant without context)
• "Things are not going well with our product" (too vague to act on)

2. Task

Define the specific challenge or objective to address

The Task component narrows the focus to exactly what needs solving. Keep it to one clear objective. If you have multiple tasks, split them into separate prompts for better results.

Good examples:

• "Identify the three most likely causes of the conversion drop and rank them by ease of investigation"
• "Draft a competitive positioning statement that highlights the agency's strategic advantage over freelancers"
• "Create a bug triage plan that prioritizes the Android 14 crash and estimates fix timelines"

Bad examples:

• "Fix everything" (undefined scope)
• "Make our marketing better" (no specific objective)
• "Help us" (no actionable direction)

3. Action

Specify the approach or steps the AI should take

The Action component tells the AI how to work through the problem. Use specific verbs: analyze, compare, rank, draft, list, evaluate. If there is a methodology or sequence you want followed, spell it out here.

Good examples:

• "Start by analyzing the checkout funnel data step by step, then compare each step's drop-off rate against industry benchmarks, and flag any step where our rate exceeds the benchmark by more than 10%"
• "Review three competing agency positioning statements, identify the common themes, then draft our statement using a differentiation-first approach"
• "Cross-reference the crash reports with Android 14 release notes, check for deprecated API calls in our photo upload module, and list potential fixes in order of implementation speed"

Bad examples:

• "Think about it" (no method)
• "Do your best" (no guidance)
• "Use AI to figure it out" (circular and unhelpful)

4. Result

Define the desired outcome and format

The Result component specifies what the finished output should look like. Include format (table, numbered list, report), length, audience, and any quality criteria. This prevents the AI from guessing what you want and delivering something you cannot use.

Good examples:

• "Deliver a one-page executive summary with a three-row table (Cause, Evidence, Recommended Fix) followed by a prioritized action list with estimated hours per item"
• "Write the positioning statement in under 50 words, then provide three 150-word variations targeted at different client segments: startups, mid-market, and enterprise"
• "Present the triage plan as a Jira-style ticket list with fields for Title, Severity, Estimated Hours, and Assignee Recommendation"

Bad examples:

• "Give me a good answer" (undefined format)
• "Make it comprehensive" (subjective and vague)
• "Write something useful" (no structure specified)

Example Prompts Using the S.T.A.R. Framework

Example 1: Problem-Solving

Prompt:

Example 2: Case Study Analysis

Prompt:

Example 3: Decision-Making

Prompt:

Best Use Cases for the S.T.A.R. Framework

1. Strategic Planning and Analysis

• Business case evaluations
• Market entry analysis
• Resource allocation decisions
• Competitive response planning

2. Problem Diagnosis

• Root cause analysis
• Performance troubleshooting
• Process bottleneck identification
• Customer complaint pattern analysis

3. Decision Support

• Build-vs-buy evaluations
• Vendor selection
• Hiring decisions
• Investment analysis

4. Scenario Planning

• Risk assessment
• Contingency planning
• What-if analysis
• Growth modeling

When NOT to Use S.T.A.R.

STAR is not the right tool for every prompt. Skip it when:

• You need a specific professional persona: Use RACE instead. STAR does not assign a role, so if you need the AI to think like a "senior tax attorney" or "DevOps engineer," RACE is a better fit.
• The task is simple and context-light: If your prompt is "Summarize this article" or "Translate this paragraph," STAR adds unnecessary overhead. Use a simpler framework like APE.
• You need strict compliance guardrails: For content that must stay within legal, brand, or regulatory boundaries, TAG provides explicit guardrail controls that STAR lacks.
• Speed matters more than depth: If you need a quick answer and the situation is obvious, writing out all four STAR components slows you down. ERA's three components may be faster.

Common Mistakes to Avoid

1. Overloading the Situation

Problem: Writing three paragraphs of background when one would suffice.Why it matters: Too much context buries the important details. The AI may fixate on a minor detail and miss the core issue.How to fix: Include only information that would change the AI's recommendation. Apply the test: "If I removed this sentence, would the answer be different?" If not, cut it.

2. Merging Task and Action

Problem: Combining what you want done with how to do it into a single block.Why it matters: When Task and Action blur together, the AI loses clarity on the objective vs. the methodology. You end up with a response that follows a process but does not clearly answer the question.How to fix: Write the Task as a single sentence stating the goal. Write the Action as a separate set of instructions describing the method. If your Task takes more than two sentences, you are probably mixing in Action details.

3. Leaving the Result Undefined

Problem: Skipping the Result component or writing something vague like "give me a good analysis."Why it matters: Without a defined Result, the AI guesses at format, length, and level of detail. You waste time reformatting or re-prompting.How to fix: Specify the exact deliverable: "a numbered list," "a comparison table," "a three-paragraph memo." Include audience and length when relevant.

4. Writing a Generic Situation

Problem: Using broad context like "We are a tech company" instead of specific details that shape the answer.Why it matters: The Situation component exists to calibrate the response. A 10-person startup has different constraints than a 5,000-person enterprise. When you leave the Situation generic, the AI defaults to generic advice that applies to nobody in particular.How to fix: Include the details that would change a consultant's recommendation: team size, budget, timeline, current tools, past attempts, and specific constraints. "We are a 40-person B2B SaaS company with $2M ARR, one DevOps engineer, and a deployment pipeline that breaks twice a month" gives the AI far more to work with.

5. Using STAR for Tasks That Need a Role

Problem: Asking for expert-level output without assigning the AI a professional identity.Why it matters: STAR does not include a Role component. If your task requires domain expertise (legal analysis, medical guidance, financial modeling), the AI defaults to a generalist perspective. You get surface-level answers instead of expert-quality analysis.How to fix: If you find yourself wanting the AI to "think like a senior engineer" or "respond as a tax attorney," switch to RACE which has a dedicated Role component. Reserve STAR for tasks where the situation and objective matter more than the professional lens.

Copy-Paste Template

Use this template for any STAR prompt. Replace the bracketed text with your specifics:

Quick example using the template:

Conclusion

The S.T.A.R. Framework is one of the most accessible entry points into structured prompt engineering. Its roots in behavioral interviewing mean that most people already understand the mental model: set the scene, name the challenge, describe the approach, define success.

Why STAR earns a place in your prompting toolkit:

• The Situation-first approach forces you to think clearly about context before asking
• The four components are simple enough to memorize after one use
• It works across business, technical, and creative domains
• Beginners get strong results without needing advanced prompting knowledge

When to graduate beyond STAR:

As your prompting skills grow, you may find that some tasks benefit from the role-assignment power of RACE or the streamlined efficiency of ERA. STAR remains your go-to for context-heavy, analytical, and decision-oriented prompts. For a broader comparison of frameworks, see the best AI prompt frameworks in 2026.

Your STAR mastery path:

• Start with a familiar business scenario you deal with regularly
• Write out each component separately before combining them
• Notice which component improves your results the most (usually Situation or Result)
• Build a personal library of STAR templates for your recurring tasks

For a deeper dive with more examples and model-specific tips, read the full STAR framework prompt engineering guide.