Back to guides
Productivity May 13, 2026 2 min read

Improve Prompts With Before-and-After Examples

How to use before-and-after prompt examples to teach revision patterns and make prompt improvements easier to review.

Before-and-after examples are one of the fastest ways to make prompt improvement concrete. They force you to compare a real change against a real outcome instead of relying on intuition about whether the prompt “feels better.”

When to use this guide

Use this guide when a prompt has clear output problems and you want a disciplined way to improve it without rewriting blindly.

1. Save the exact before state

Keep the original prompt and one representative input. If you change both the prompt and the input, the comparison becomes noisy.

2. Name the problem first

Choose one problem to improve:

  • weak structure
  • missing recommendation
  • invented details
  • inconsistent tone
  • poor handoff

A before-and-after comparison is only useful when you know what you are trying to improve.

3. Make one meaningful change

Examples:

  • add an output contract
  • add uncertainty handling
  • remove decorative filler
  • make the task narrower
  • ask for ranked priorities

One focused change tells you more than five mixed edits.

4. Compare the outputs side by side

Look for:

  • whether the task is clearer
  • whether the output is easier to review
  • whether unsupported claims were reduced
  • whether the answer became more reusable

This is especially helpful for prompts influenced by Lyra Prompt Optimizer or Prompt Architect, where the goal is systematic improvement rather than one lucky run.

5. Write down what changed

A short note like “adding explicit output sections improved consistency but made the answer too rigid” is valuable. It preserves reasoning for the next review.

Review checklist

A good before-and-after workflow should leave you with:

  • the original prompt
  • the revised prompt
  • the same test input
  • a visible improvement goal
  • a short note about what changed

For surrounding vocabulary, see Prompt Iteration, Prompt Refinement, and What a Prompt Optimizer Actually Does.

Related prompts