How to Write ChatGPT Prompts That Get Better Answers

1What Makes a Good ChatGPT Prompt?

A good ChatGPT prompt clearly states the task, gives the model the context it needs to understand the situation, and describes the format the answer should come back in. Vague prompts get vague answers because the model has to guess at all three of those things.

The difference between a mediocre and a great ChatGPT response is rarely the model itself — it is almost always how specific and structured the prompt was.

2The Three Parts of a Strong Prompt

Most effective prompts, whether for writing, coding, or analysis, include the same three ingredients.

• Context: relevant background the model needs — who the audience is, what has already been tried, what constraints apply.
• Task: a specific, concrete instruction rather than an open-ended question.
• Format: how the answer should be structured — a list, a table, a short paragraph, a specific length.

3Give the Model a Role

Asking the model to respond as a specific role — an experienced editor, a patient tutor, a skeptical code reviewer — shapes the tone, depth, and priorities of the response more reliably than a plain, roleless request. A role sets expectations for the whole answer in a single phrase.

4Break Complex Requests into Steps

A single prompt asking for five different things at once usually produces a shallow answer to all five. Breaking the same request into a short sequence of prompts — first outline, then draft, then refine — generally produces a stronger result than trying to get everything in one shot.

5Show an Example (Few-Shot Prompting)

Including one example of the format or style you want, known as few-shot prompting, is one of the most reliable ways to steer an answer. Rather than describing the tone you want in words, showing a short sample of it lets the model match the pattern directly.

6Ask for Step-by-Step Reasoning

For tasks involving logic, math, or multi-step decisions, explicitly asking the model to reason through the problem step by step before giving a final answer noticeably improves accuracy, because it reduces the chance of skipping a step.

7Example Prompts: Before and After

Seeing a vague prompt rewritten with context, task, and format makes the difference concrete.

• Vague: Write about databases. Better: Explain what a relational database is to someone who has never used a spreadsheet, in three short paragraphs, using one everyday analogy.
• Vague: Fix my code. Better: Here is a Python function that should return the average of a list but returns zero for empty lists; explain the bug and suggest a fix, keeping the function signature the same.
• Vague: Help me plan a project. Better: Act as a project manager and break this goal into a five-step plan with one measurable outcome per step.

8Common Prompting Mistakes

A few recurring habits quietly undermine otherwise reasonable prompts.

• Asking a question with no context, expecting the model to infer the audience and purpose.
• Stacking too many separate asks into one prompt instead of sequencing them.
• Never specifying a format, then being surprised the answer is too long or too short.
• Giving up after one attempt instead of refining the prompt based on what the first answer got wrong.

9Practicing Prompting as a Skill

Prompting is a learnable, transferable skill, not a trick reserved for a few power users. Practicing the same context-task-format structure across writing, coding, and analysis tasks builds an intuition that carries over to any large language model, not just ChatGPT specifically.