Anatomy of a Great Prompt

Why Most People Get Bad Results from AI

The #1 reason people are disappointed by AI isn't the technology — it's the prompt. Vague inputs produce vague outputs. Specific inputs produce specific, useful outputs.

The CRAFT Framework

Great prompts have five elements. You don't need all five every time, but the more you include, the better your results.

C — Context: What's the situation? What background does the AI need? "I'm a marketing manager at a B2B SaaS company targeting mid-market CFOs." R — Role: Who should the AI act as? "Act as an experienced copywriter who specialises in financial services." A — Action: What exactly do you want it to do? "Write three subject lines for a cold email campaign." F — Format: How should the output be structured? "Present each as a numbered option with a brief explanation of why it would work." T — Tone: What voice or style should it use? "Professional but not stuffy. Confident, not pushy."

Putting It Together

Bad prompt: "Write me an email subject line." Good prompt: "I'm a marketing manager at a B2B SaaS company targeting mid-market CFOs. Act as an experienced copywriter who specialises in financial services. Write three subject lines for a cold email campaign about our automated expense reporting tool. Present each as a numbered option with a brief explanation of why it would work. Tone: professional but not stuffy."

The difference in output quality is dramatic.

Common Mistakes

1. Too vague: "Help me with my presentation" → "Help me create a 10-slide presentation structure for a board meeting about Q3 results." 2. No format specified: AI will guess how to structure the output. Tell it what you want. 3. No context: AI doesn't know your situation unless you tell it. The more relevant context, the better. 4. Asking for too much at once: Break complex requests into steps.