Prompting for Structured Outputs — Tables, Comparisons, and Frameworks

Why Structured Outputs Matter

Most business communication requires structure — tables, matrices, scorecards, frameworks, and formatted reports. Yet most people prompt AI for free-form text and then spend time reformatting it. Prompt engineering for structured outputs means getting the right format from the start.

Technique 1: Table Prompting

Define Columns Explicitly

Create a table with these columns: Feature | Option A | Option B | Option C | Winner Include 8 rows comparing [specific features]. For each cell, provide a brief (5-10 word) assessment. Add a final "Total Score" row.

Markdown Table Format

Output as a markdown table. Use pipes (|) and dashes (-) for formatting. Example:

Column 1	Column 2	Column 3
Data	Data	Data

Complex Multi-Level Tables

Create a 2-level table for our project timeline: Level 1: Phase name (row spanning all columns) Level 2: Individual tasks within each phase Columns: Task | Owner | Start Date | End Date | Status | Dependencies Phases: Planning, Development, Testing, Deployment

Technique 2: Decision Matrices

Weighted Scoring Matrix

Create a weighted decision matrix for selecting an AI training provider. Criteria (with weights):

1. Customisation capability (25%)
2. Trainer expertise (20%)
3. HRDF/SSG registration (15%)
4. Post-training support (15%)
5. Price (15%)
6. Client references (10%)

Evaluate Provider A, B, and C. For each: score 1-5, calculate weighted score, and determine overall winner.

Pros/Cons Matrix

Create a pros/cons comparison for 3 options: Option A: Build in-house AI capability Option B: Partner with AI consulting firm Option C: Hybrid (train team + consulting for complex projects) For each option, list exactly 5 pros and 5 cons. Rate each pro/con as High/Medium/Low impact.

Technique 3: Framework Outputs

SWOT Analysis

Conduct a SWOT analysis for [subject]. Format as a 2x2 grid:

	Helpful	Harmful
Internal	Strengths (5 items)	Weaknesses (5 items)
External	Opportunities (5 items)	Threats (5 items)
For each item, provide a 1-sentence explanation and rate its significance (High/Medium/Low).

RACI Matrix

Create a RACI matrix for our AI training rollout project. Roles across the top: CEO, HR Director, IT Manager, Training Provider, Department Heads. Tasks down the side (list 10 key tasks). For each cell: R (Responsible), A (Accountable), C (Consulted), I (Informed), or blank.

Risk Register

Create a risk register for [project] with these columns: Risk ID | Description | Category | Likelihood (1-5) | Impact (1-5) | Risk Score | Mitigation Strategy | Owner | Status Include 10-15 risks, sorted by risk score (highest first).

Technique 4: Scorecard and Dashboard Outputs

KPI Scorecard

Design a monthly KPI scorecard for the HR department. Format: Table with columns: KPI Name | Target | Actual | Variance | Status (🟢🟡🔴) | Trend (↑↓→) Include 12 KPIs covering: recruitment (4), retention (3), L&D (3), compliance (2).

Performance Dashboard

Design a weekly operations dashboard. Include:

1. Header: period, prepared by, distribution list
2. Summary metrics (6 tiles with: metric name, current value, target, trend)
3. Detailed table (15 KPIs with RAG status)
4. Top 3 issues requiring attention (format: issue, impact, action, owner)
5. Upcoming milestones (next 2 weeks)

Technique 5: Checklist Outputs

Pre-Event Checklist

Create a checklist for organising a corporate AI training workshop. Format:

• Task description (Owner) — Deadline Organise into phases: 4 weeks before, 2 weeks before, 1 week before, day of, day after. Include at least 30 items.

Audit Checklist

Create a data privacy compliance checklist for a Singapore company using AI tools. Format: Category → Requirement → Compliance Status (Yes/No/Partial) → Evidence Required → Notes Categories: Consent, Data Collection, Storage, Processing, Transfer, Breach Response.

Pro Tips for Structured Outputs

1. Always specify the format — Do not assume AI will choose the right structure
2. Provide column headers — Name every column explicitly
3. Specify cell content — Tell AI what goes in each cell (brief text, score, status)
4. Request examples — "Show one completed row as an example before filling the rest"
5. Set constraints — "Each cell should be maximum 10 words"
6. Request totals and summaries — "Add a total row and an overall recommendation"

Related Reading

• Prompt Patterns Guide — Comprehensive guide to prompt engineering patterns
• Prompting Internal Documents — Work with internal data sources using structured prompts
• Copilot for Teams, Outlook & Excel — Apply structured prompting in Microsoft Copilot