Writing Clear Prompts

Writing clear prompts is like telling a friend exactly what you want. The clearer you are, the less guessing the model has to do.

Checklist for clear prompts:

1. Role + Task: Tell the model who to be and what to do in one short sentence.
2. Constraints: Specify length, tone, and exact format (e.g., “3 bullet points”).
3. Examples: Show 1–3 examples when style matters.
4. Step requests: For hard tasks, ask the model to show steps or an outline first.
5. Negative rules: Say what not to do when needed (e.g., “Do not use technical jargon”).

Vague vs Clear (easy comparison):

• Vague: “Write a product description.”
• Clear: “You are a marketing writer. Write a 60–80 word product description for a lightweight hiking backpack. Use active verbs and include features: waterproof fabric and 20L capacity. End with a 1-line tagline.”

Small, precise changes like adding a role, a word count, and required features usually make outputs much better.

Quick workflow (fast loop):

1. Ask the model with a clear prompt.
2. Check the result against your rules (format, length, tone).
3. If it’s wrong, fix the prompt (add constraints, example, or ask for steps) and try again.

Common mistakes:

• Forgetting to say how the answer should look.
• Using vague adjectives like “good” without specifics.
• Expecting the model to read long context you didn’t provide.

Tip: Keep a short prompt notebook with the version that worked and why — it’s the fastest way to repeat success.