Tip 3: Use format to guide output quality
Why format matters
One of the ways you can guide output quality is to use format. Sometimes in our prompts, we’ll say “output in this format” and tell the model exactly how we want it structured. However, a strict format can restrict the model’s creativity and lead to subpar results.
Example: LinkedIn post for Ramp
In this scenario, a typical LinkedIn post would include hashtags, multiple paragraphs, calls to action, and a conversational tone. If you prompt the model with “output only a title and a paragraph,” the result will diverge from a standard LinkedIn style. It meets the format but sacrifices key elements:
Why restrictive formats can compromise quality
- The model prioritizes matching the requested format over delivering full context.
- Essential components—like hashtags or multiple paragraphs—get omitted.
- Creativity and engagement may be reduced when structure is too narrow.
How to use formatting to boost quality
- Ask the model to output intermediate thinking steps before the final answer.
- For example, first request a list of traits for a strong LinkedIn post.
- Then ask the model to craft the post using those traits as guidance.
- This encourages better reasoning and richer, more complete output.
As you can see, output format is really key to improving or worsening the quality of your content.
