
Fotoğraf: www.Pixel.la Free Stock Photos, Wikimedia Commons (CC0)
6 Rules for Writing Prompts That Get Better Results
Getting much better output from the same model isn't about magic words — it's about structuring the request properly.
Nova AI News Editor
August 18, 2026 · 1 min read
1. State the Role and Purpose Up Front
The difference between "write me some text" and "write a 60-word opening announcement in a friendly tone for the Instagram followers of a newly opened café" is the amount of blank space the model has to guess at. The less blank space, the more accurate the result.
2. Describe the Output Format
Do you want a table, a bullet list, JSON? When you write out the column headings and the ordering, the time you spend editing the output by hand drops to nearly zero.
3. Give Examples
Showing one or two examples is more effective than pages of explanation. Instead of telling the model "in this style," showing it a short passage in that style makes it far easier for it to catch the tone.
4. Say What You Don't Want, Too
Negative constraints like "don't open with a cliché," "don't use overblown adjectives," "don't invent sources" visibly raise the quality of the output, especially in marketing copy.
5. Break Long Jobs into Pieces
Rather than saying "write a 3,000-word guide on this topic" in one shot, asking for an outline first and then section-by-section writing gives a more consistent result. At each step the model has fewer things to balance at once.
6. Have the Model Check Its Own Work
Once you have the output, asking it to "list the unverifiable claims in this text" or "flag the unnecessary repetition" works like a second review pass.
Conclusion
A good prompt is a good brief. Whatever you'd have to explain when handing a job to an agency, you need to explain to the model too.
Related Articles

How Students Should Actually Use AI
Letting AI do your homework kills the learning. Here is how to use it as a tutor instead.
Read more→
Running an AI Model on Your Own Computer
The hardware, tools, and realistic expectations for running a model on your own machine without a cloud subscription.
Read more→
What's the Difference Between Machine Learning and Deep Learning?
We explain two commonly confused concepts in plain language, along with when you should reach for each one.
Read more→Comments
No comments yet — be the first to comment.