If you’re getting average answers, it’s probably the prompt — not the AI.
Why Prompts Matter
If you’ve ever said “ChatGPT gave me a bad answer,” pause and ask: What did I actually ask it?
In most cases, the AI didn’t fail. The prompt did.
Large language models like ChatGPT are powerful — but not psychic. The quality of your input shapes the quality of the response. That’s why it helps to craft your prompts intentionally.
One simple, flexible model can make all the difference.
Introducing the CRAFT Prompting Model
CRAFT is a five-part prompt structure that helps you set the scene, shape the response, and guide ChatGPT toward more relevant and useful output. Thanks to Brian Albert for sharing the CRAFT model.
Each letter stands for a key element:
Letter | Element | What it Does |
---|---|---|
C | Context | Sets the situation or background |
R | Role | Tells ChatGPT who to act as |
A | Ask | Specifies the request or task |
F | Format | Shapes the output structure |
T | Target Audience | Clarifies who the output is for |
Worked Example: A Difficult Conversation at Work
Let’s say you’re a team leader preparing for a tricky conversation with a staff member. Here’s how you can apply the CRAFT model:
C – Context:
“I’m a team leader preparing a difficult conversation with a team member who’s been underperforming.”
R – Role:
“Act as a leadership coach experienced in workplace conflict.”
A – Ask:
“Give me three opening lines to begin the conversation respectfully.”
F – Format:
“Respond in dot points, with a short explanation for each.”
T – Target Audience:
“This is for a mid-level manager with some experience but low confidence.”
Put it all together:
“I’m a team leader preparing a difficult conversation with a team member who’s been underperforming. Act as a leadership coach experienced in workplace conflict. Give me three opening lines to begin the conversation respectfully. Respond in dot points, with a short explanation for each. This is for a mid-level manager with some experience but low confidence.”
That’s a high-quality prompt — and now the AI knows how to help you.
Why CRAFT Works
The CRAFT model works because it reduces ambiguity.
Many people treat ChatGPT like Google — vague, open-ended, and hopeful. But ChatGPT works better when it has structure.
CRAFT gives you:
-
-
Clarity – The model knows what’s going on
-
Relevance – The answers match your scenario
-
Tone – You get output that fits the audience
-
Efficiency – Less time rewriting or re-asking
-
Control – You shape the format and voice
-
CRAFT in Practice – Real Use Cases
Here are a few ways CRAFT helps professionals work smarter with AI:
✅ For Managers
“I’m reviewing a performance issue. Act as an HR partner. Help me outline three coaching strategies. Use dot points. For a manager who’s early in their leadership career.”
✅ For Consultants
“I’m facilitating a strategy session. Act as a workshop designer. Suggest three energiser activities. Format as a table. For a group of senior executives in a government agency.”
✅ For Educators
“I’m teaching graduate students about critical thinking. Act as a curriculum designer. Create a short activity. Use dot points. For adult learners with limited AI experience.”
Pro Tips for Using CRAFT
-
You don’t need to use all 5 elements every time.
Even 3–4 will dramatically improve results. -
CRAFT is adaptable.
It works for writing, analysis, idea generation, coaching, and more. -
Keep refining.
If the response isn’t quite right, tweak your Ask or Format — not everything. -
Build prompt templates.
Save your best CRAFT prompts for repeat use or Projects.
Suggested Prompt Template
Here’s a simple version you can adapt:
“I’m [insert context]. Act as [insert role]. Give me [insert ask]. Format the response as [insert format]. The audience is [insert target].”
Example:
“I’m leading a team offsite. Act as an experienced facilitator. Help me design an icebreaker. Format as a checklist. For a mixed group of introverts and extroverts.”
Internal Links for This Post
-
Understanding Tokens and Context Limits → Explain why well-crafted prompts matter in long threads
-
Using Projects in ChatGPT → Save your best CRAFT prompts in dedicated project spaces
-
From Prediction to Partnership → Reframe prompting as collaborative thinking, not command-and-control
-
AI Boosters for Busy Managers → Discover real-world use cases where CRAFT makes AI work-ready
Final Thoughts
Prompts are not just instructions. They’re a way of thinking.
When you use CRAFT, you shift from “asking questions” to shaping collaboration — and that’s where AI becomes a true thinking partner.
Better prompts = better answers = better outcomes.