What makes a good ChatGPT prompt?

A good ChatGPT prompt is specific about the task, defines the output format, specifies the audience and tone, and provides relevant context. The more precisely you describe what you want, the better the output.

What is role prompting?

Role prompting means telling the AI to act as a specific expert or persona before giving your task. For example: 'You are a senior software engineer with 10 years of Python experience.' This primes the model to respond with domain-appropriate depth and vocabulary.

How long should a ChatGPT prompt be?

A good prompt is as long as it needs to be — not shorter, not longer. Simple tasks need 1-2 sentences. Complex tasks may need a paragraph of context, constraints, and format instructions. Don't pad for the sake of length, but don't cut context that helps the model.

Guide

How to Write Better ChatGPT Prompts

Updated June 2026 · 8 min read

The difference between a mediocre AI response and a great one usually isn't the model — it's the prompt. Here are the techniques that consistently produce better output, with real before/after examples.

1. Be specific about what you want

Vague prompts produce vague answers. The more precisely you describe the output, the better.

❌ Weak

Write something about productivity.

✓ Strong

Write a 300-word blog intro about AI productivity tools for busy freelancers. Use a conversational tone and end with a question that hooks the reader.

2. Assign a role

Telling the AI to act as a specific expert dramatically improves technical depth and accuracy. This is called role prompting.

❌ Without role

Review my Python code.

✓ With role

You are a senior Python engineer with expertise in performance optimization. Review the following code for bugs, inefficiencies, and style issues. Explain each problem and suggest a fix.

3. Specify the format

AI models default to paragraphs. If you want a list, table, or structured output — say so explicitly.

Give me a comparison of Claude vs ChatGPT as a markdown table with columns: Feature, ChatGPT, Claude, Winner.

4. Define the audience

The same topic explained to a 10-year-old vs. a domain expert looks completely different. Don't leave this to chance.

Explain how transformers work in AI. Write for someone who understands basic programming but has no machine learning background. Avoid jargon and use an analogy.

5. Give examples (few-shot prompting)

Show the model exactly what you want by including 1-2 examples of the output style. This is one of the most reliable ways to control tone and format.

Write product descriptions in this style:

Example: "The Aeron Chair — engineered for people who take sitting seriously. PostureFit SL keeps your spine where it belongs."

Now write one for: a mechanical keyboard for developers.

6. Add constraints

Tell the AI what NOT to do. Constraints are just as powerful as instructions.

Summarize this article in 5 bullet points. Do not include any statistics. Do not use the word "significant". Keep each bullet under 20 words.

7. Use chain of thought for complex tasks

For reasoning-heavy tasks, ask the model to think step by step before giving the final answer. This reduces errors significantly.

A user has a $500 budget and wants to build a home office AI setup. Think through their options step by step — hardware, software, and subscriptions — then give a final recommendation.

8. Iterate, don't start over

The best prompt is rarely the first one. After getting a response, refine with follow-ups:

"Make it shorter — under 100 words."
"Rewrite the intro to be more direct."
"Add a section on [specific topic]."
"The tone is too formal — make it more casual."

Try our Prompt Generator

Use our free AI Prompt Generator to build structured prompts from scratch — pick your task, tone, audience, and format, and get a ready-to-use prompt in seconds.

Quick reference: prompt template

You are [role/expert].

Task: [what you want the AI to do]
Audience: [who will read this]
Tone: [formal / casual / technical]
Format: [bullet list / table / paragraph / code]
Length: [word count or size]
Constraints: [what to avoid]

Context: [any background info the AI needs]