CRAFT Framework
Context · Role · Action · Format · Target
Role-based framing with a clear target audience. Great for marketing, copywriting, and creative work.
- Definition
- The CRAFT framework (Context · Role · Action · Format · Target) is a prompt engineering structure that breaks your AI request into 5 discrete fields. It is best suited for marketing copy and content creation.
The 5 Fields
Context
Background the AI needs — what project, product, or situation this relates to.
Role
The persona the AI should adopt — e.g., expert copywriter, senior developer, UX researcher.
Action
The specific task — what you want the AI to do.
Format
How the output should be structured — e.g., numbered list, email, code block, table.
Target
The intended audience for the output. Defines vocabulary, depth, and angle.
Real Example
Scenario: Writing a LinkedIn post for a new AI product launch
Context: PromptQuorum is launching a multi-AI comparison tool. Role: You are a B2B SaaS marketing expert. Action: Write a LinkedIn post announcing the product. Format: 3 short paragraphs + 5 hashtags. Target: Startup founders and AI-curious professionals.
When to Use CRAFT
- ✓Marketing copy and content creation
- ✓Creative writing and storytelling
- ✓Any task where a specific expert persona improves quality
- ✓Writing targeted at a defined reader
- ✗Tasks requiring fine-grained tone/style separation (use CO-STAR)
- ✗Sequential step-by-step processes (use RISEN)
- ✗Tasks where few-shot examples are the key signal (use TRACE)
Frequently Asked Questions
What does CRAFT stand for?
CRAFT stands for Context, Role, Action, Format, and Target — five fields that combine role-based framing with a clear audience definition.
How is CRAFT different from CO-STAR?
CRAFT uses a Role field (giving the AI a persona) while CO-STAR splits style and tone separately. CRAFT is simpler; CO-STAR gives more voice control.
What is CRAFT best used for?
CRAFT excels at marketing copy and creative writing where assigning an expert persona and defining a target audience significantly improves output quality. After building CRAFT prompts, evaluate their quality using systematic frameworks at https://www.promptquorum.com/prompt-engineering/how-to-evaluate-prompt-quality.