Skip to main content
Module 1: Getting Started with ChatGPT

Prompting ChatGPT well

OpenAI's current advice: describe the outcome, don't over-engineer.

How you ask shapes what you get, and OpenAI's current prompting guidance has actually shifted — away from elaborate 'act as an expert, think step by step' rituals toward something simpler. Learning the current approach makes you more effective and saves effort.

Describe the outcome, lead with the result. OpenAI's key current advice: state what a good result looks like — the goal, the constraints, and the shape of the final answer — and let ChatGPT figure out how to get there. In their words, describe a process only when the process itself matters; otherwise leave ChatGPT room to search, compare, and adjust its approach. So instead of scripting every step, say what you want the outcome to be. 'Give me a friendly 3-paragraph email declining this, offering to reconnect next quarter' beats a long list of instructions on how to write it.

Don't over-engineer. A genuine shift: the old habit of elaborate prompt rituals is no longer necessary. OpenAI now advises shorter, clearer prompts — a short prompt often works well. Add more structure only for bigger, more complex tasks. So don't feel you need a magic incantation; clear communication of what you want is enough.

The optional structure (for bigger tasks): goal, context, output format, and boundaries.

  • Goal — what you want (lead with this).
  • Context — relevant background, audience, purpose.
  • Output format — the shape you want back (a table, bullets, an email, a word count, a tone).
  • Boundaries — key constraints (what to include or avoid).

For a simple task, just the goal. For a complex one, fill in all four.

Be specific, and specify the format. Vague prompts get generic answers, so be specific about what you want. And explicitly ask for the format — 'as a table,' 'under 150 words,' 'three options,' 'in a professional tone.' Specifying the output shape is one of the highest-value moves.

Iterate. Even with good prompting, the first answer is a starting point. Follow up to adjust — 'shorter,' 'more formal,' 'that opening is generic, try three sharper ones.' Prompting is a conversation, not a one-shot.

The takeaway: OpenAI's current prompting advice has shifted toward simplicity. Describe the outcome, lead with the result — state what a good answer looks like (goal, constraints, final shape) and let ChatGPT choose how to get there; describe a process only when the process matters. Don't over-engineer — the old elaborate rituals are unnecessary; shorter, clearer prompts often work, with more structure only for bigger tasks. For complex work, an optional structure helps: goal (lead with it), context, output format, and boundaries. Be specific and explicitly ask for the format you want (a table, word count, tone) — one of the highest-value moves. And iterate, since the first answer is a starting point. Clear communication beats magic incantations.

Try it

Try the outcome-first approach: take a task and write a prompt that *describes the result you want* (goal, any constraints, the format) rather than scripting every step — and keep it concise. Run it, then iterate once to refine. Notice you don't need an elaborate prompt — clear communication of the outcome, plus specifying the format, is what works now.

Stay in the loop

Enjoying the free lessons? Get an email when we publish new courses and updates — no spam, unsubscribe anytime.

Discussion (0)

Ask a question or share what worked for you. Comments are reviewed before they appear.

Log in to join the discussion and ask questions about this lesson.

No comments yet. Be the first to start the discussion!