A practical 20-point check to improve an AI prompt before you rely on its output at work.
Better AI output usually begins with a clearer request. Use this scorecard before submitting an important workplace prompt—or when diagnosing an answer that feels generic, incomplete or unreliable.
How to use it: Check each statement that is true of your prompt. Each checked item earns one point. The score helps you improve the prompt; it does not guarantee that the resulting AI output is accurate or appropriate.
Your prompt score: 0/20Start checking the statements below.
0–5: RebuildAdd the purpose, audience and essential context before proceeding.
6–10: ClarifyThe direction is visible, but important inputs and boundaries are missing.
11–15: StrengthenThe prompt is usable; improve evidence, constraints and review instructions.
16–20: Ready to testThe prompt is well structured. Test the output and apply human review.
Important: A high score measures prompt completeness—not truth, safety or fitness for purpose. Always verify consequential outputs using appropriate sources and expertise.
Frequently asked questions
Does a high score guarantee a good AI answer?
No. It improves the conditions for a useful response, but AI can still misunderstand context, invent information or produce inappropriate recommendations.
Should every prompt score 20?
No. A low-risk drafting task may not require every item. Higher-impact work generally needs stronger evidence, constraints and human review.
Can teams adapt this scorecard?
Yes. Add organisation-specific requirements for privacy, security, compliance, brand voice and approval.
