The Anatomy of a Good Prompt
The 4-Part Framework
Every great prompt has four ingredients. You don't always need all four — sometimes two are enough. But knowing the framework means you can always diagnose why a prompt isn't working and fix it. Let's break it down.
1. Role — Who should the AI be?
Tell the AI what perspective to take. "You are a senior marketing strategist" or "Act as a patient, experienced teacher." This shapes the AI's vocabulary, depth, and approach. A "doctor" answers differently than a "5-year-old."
2. Context — What's the situation?
Give background information. Who's the audience? What's already happened? What constraints exist? The more context you provide, the more relevant the output. Think of it as briefing a new hire on day one.
3. Task — What exactly do you want?
Be specific about the deliverable. Not "help me with marketing" but "write 5 Instagram caption options for this product photo, each under 100 words, with a call-to-action." Clear task = clear output.
4. Format — How should the output look?
Specify the structure: bullet points, numbered list, table, paragraph, email format, JSON, markdown. Also specify length: "3 paragraphs," "under 200 words," "one page." Format instructions eliminate 90% of back-and-forth.
Remember: R-C-T-F
Role, Context, Task, Format. You don't need all four every time — but when your output isn't great, check which ingredient you're missing. Usually it's context or format.
The Framework in Action
**Role:** You are an experienced HR manager at a mid-size tech company. **Context:** We just launched a new unlimited PTO policy. Some employees are confused about how it works and worried it means "no PTO." The company culture is casual and transparent. **Task:** Write an internal FAQ (5 questions and answers) that addresses common concerns and encourages people to actually use their PTO. **Format:** Use a friendly, conversational tone. Each answer should be 2-3 sentences max. Include one light joke to keep it on-brand.
[The AI would generate a perfectly targeted FAQ that addresses the exact concerns, in the right tone, at the right length, with the right format — because every dimension was specified.]
Why this works: All four ingredients are present. The AI knows WHO it is (HR manager), WHAT the situation is (new PTO policy, confused employees), WHAT to produce (FAQ, 5 Q&As), and HOW to format it (conversational, short answers, include a joke).
Common Mistakes (And Easy Fixes)
Weak Prompt
- •"Write a blog post about AI"
- •"Help me with my resume"
- •"Create a presentation"
- •"Summarize this document"
Strong Prompt (Add R-C-T-F)
- •"Write a 600-word blog post about how nurses use AI in 2026, for a healthcare audience, conversational tone"
- •"Rewrite my resume summary for a product manager role at a SaaS startup, emphasizing my data skills, 3-4 lines"
- •"Create a 10-slide outline for a Q1 sales review, for our VP audience, include key metrics placeholders"
- •"Summarize this doc in 5 bullet points, focus on action items and deadlines, skip background info"
The 2026 Shortcut: Just Tell AI What's Wrong
Here's something that's changed with modern AI: you don't always need to write the perfect prompt upfront. The models in 2026 are much better at follow-up refinement. Get a response, then say: "This is too formal — make it more casual" or "Good direction but I need it shorter and with more specific examples." Iterating is often faster than crafting the perfect prompt from scratch.
Think of your first prompt as a rough sketch. Follow-ups are where you refine. Don't stress about getting it perfect on the first try — stress about knowing what to ask for in the follow-up.
Quick Check
Your AI output is well-written but way too long and too formal. What's the fastest fix?
Key Takeaway
Use the R-C-T-F framework: Role, Context, Task, Format. You don't need all four every time, but checking which one is missing usually fixes a bad output. And don't forget — follow-up refinement is just as powerful as the initial prompt.