Output Formatting & Structure
Control exactly how AI presents its responses — tables, lists, JSON, markdown, and more.
Why Format Matters
The same information presented in different formats can be dramatically more or less useful. A 500-word paragraph about project risks is harder to act on than a table with columns for Risk, Likelihood, Impact, and Mitigation.
Always tell the AI how you want the output structured. If you don't specify, you get whatever the model defaults to — usually verbose paragraphs.
Common Format Instructions
Tables: "Present as a markdown table with columns: [Column1], [Column2], [Column3]"
Bullet points: "List the top 5 points as bullet points, each under 20 words"
Numbered steps: "Provide step-by-step instructions, numbered 1-10"
JSON: "Return the data as a JSON object with keys: name, category, priority, description"
Code blocks: "Write the code in a Python code block with comments explaining each section"
Comparison: "Compare these in a two-column table: Pros on the left, Cons on the right"
Executive summary: "Start with a 2-sentence executive summary, then provide detailed analysis"
Controlling Length
Be explicit about length:
- "Respond in exactly 3 sentences"
- "Keep your response under 200 words"
- "Write a comprehensive analysis (1,000+ words)"
- "Give me the TL;DR in one paragraph"
Without length instructions, AI tends to be verbose. It's almost always better to ask for shorter responses and then request more detail on specific points.
Ask ChatGPT to explain machine learning three ways: as a paragraph, as a table comparing types, and as a numbered list of steps to learn it. Notice how the same content becomes more or less useful depending on format.
- ✓Always specify output format — don't let the AI default to paragraphs
- ✓Tables are underused and extremely effective for comparisons and data
- ✓Be explicit about length — AI defaults to verbose
- ✓JSON output is valuable for structured data you'll process further