Best practice L1 · Basic prompts informational

Show an Example of 'Good' Instead of Describing It

What it is

A prompting upgrade: instead of describing the output you want in adjectives, paste one concrete example of 'good' and ask ChatGPT to match its style, structure, or tone.

show the targetOutput matches the example3An example of a good result2The task1
One concrete example pins down what adjectives leave vague.

Why it works

Words like 'professional', 'punchy', or 'well-structured' mean different things to you and the model, so it guesses and often misses. One example collapses that ambiguity — it shows the exact length, format, and voice you mean far more precisely than any description. Show-don't-tell is the single cheapest way to raise output quality.

When to use it

Whenever you have or can sketch an example of the result you want — a past piece, a competitor's, even a rough mock-up of the shape.

When not to use it

Genuinely open or exploratory tasks where you want the model's range, and pinning it to one example would narrow it prematurely.

Prompt

Here's an example of the kind of output I want: <paste example>. Now produce <your task> in the same style/structure/tone as that example. Match its format and length; the content is different.

Example

Pasting one product description you love and saying 'write ours like this' beats three sentences of 'make it snappy and on-brand' every time.

Common mistakes

  • Relying on adjectives the model interprets differently than you.
  • Having a perfect example on hand but describing it instead of pasting it.
  • Giving an example that conflicts with your written instructions, confusing the model.

Related