How to write better AI prompts: practical templates that work

A good prompt is not a magic phrase. It is a clear task brief. The model needs to know what you want, what context matters, what to avoid and what format the answer should have.

The basic formula

Use this structure when the first answer matters: role, task, context, constraints, output format and quality check. You do not need every part every time, but the formula helps when the result is vague.

Role: You are a concise product marketer.
Task: Rewrite this landing page headline.
Context: The product is an AI note-taking app for remote teams.
Constraints: Keep it under 70 characters. Avoid hype.
Output: Give 10 options and mark the 3 strongest.

For writing

Rewrite the text below for a professional audience. Keep the meaning, remove repetition, use short sentences, and preserve all facts. Return the result in plain English.
 
Text: ...

For analysis

Analyze this table and find the 5 most important patterns. Separate facts from hypotheses. If data is missing, say what cannot be concluded. End with a short action list.

For images

Create a realistic editorial portrait of a founder in a small design studio, natural window light, 50mm lens, muted colors, clean background, high detail. Avoid text, logos, extra fingers and distorted face.

Common prompt mistakes

• Asking for “the best” without saying best for whom.
• Requesting too many tasks in one message.
• Not giving source material when factual accuracy matters.
• Forgetting to define the output format.

The useful follow-up

After the first answer, ask the model to improve against a specific criterion: shorter, more concrete, more skeptical, more beginner-friendly, more technical, or closer to a source document. Iteration is part of prompting, not a failure of prompting.