Lesson 9: Prompting for Data Analysis & Research

Analysis Prompts That Deliver Insights

Getting LLMs to analyze data and generate insights requires structured prompts that guide the reasoning process.

The ANALYZE Framework

Mental model: Think of analysis prompts like giving a research brief — the clearer the question and constraints, the more actionable the insights.

Data Analysis Prompts

Basic data analysis:

Analyze this data and provide insights:

[Data in CSV, JSON, or table format]

Please provide:
1. Summary statistics (mean, median, mode, range)
2. Key trends or patterns
3. Anomalies or outliers
4. Actionable recommendations

Format: Executive summary first, then detailed analysis.

Comparative analysis:

Compare these two datasets/periods/options:

[Data set A]
[Data set B]

Analyze:
1. Key differences and similarities
2. Which performs better on each metric and why
3. Statistical significance of differences (if applicable)
4. Recommendations based on the comparison

Present as a structured comparison table with analysis.

Trend analysis:

Analyze trends in this time-series data:

[Data with timestamps]

Identify:
1. Overall direction (growth/decline/stable)
2. Seasonal patterns
3. Inflection points and possible causes
4. Forecast for next [period]

Include confidence level for predictions.

Research Prompts

Literature review:

Conduct a research summary on [topic].

Please:
1. Identify the key themes and debates
2. Summarize major findings from authoritative sources
3. Note areas of consensus and disagreement
4. Identify gaps in current research
5. Suggest directions for future investigation

Cite sources where possible. Note your confidence level for each finding.

Competitive analysis:

Analyze the competitive landscape for [industry/product].

Provide:
1. Key players and their market positions
2. Strengths and weaknesses of each
3. Market trends affecting competition
4. Opportunities and threats
5. Strategic recommendations

Format: Executive summary + detailed breakdown by competitor.

Market research:

Research the [target market] for [product/service].

Include:
1. Market size and growth projections
2. Customer segments and personas
3. Buying behavior and decision factors
4. Price sensitivity analysis
5. Distribution channels
6. Regulatory considerations

Base analysis on available data. Note where you're making assumptions.

Insight Generation Prompts

From raw data:

You are a data analyst examining this dataset:

[Data]

Generate insights that would be valuable for [stakeholder type].

Focus on:
1. What's happening (descriptive)
2. Why it's happening (diagnostic)
3. What might happen (predictive)
4. What to do about it (prescriptive)

Prioritize insights by business impact (high/medium/low).

From reports:

Read this report and extract the key insights:

[Report content or summary]

Please:
1. Identify the 3-5 most important findings
2. Explain why each matters
3. Note any surprising or counterintuitive results
4. Suggest follow-up questions
5. Rate your confidence in each insight (high/medium/low)

Question-Driven Analysis

For exploring data:

I have data about [topic]. Help me explore it by answering these questions:

1. What is the distribution of [key metric]?
2. Are there correlations between [variable A] and [variable B]?
3. What are the top/bottom performers by [metric]?
4. Are there any segments that behave differently?
5. What would happen if [hypothetical scenario]?

For each answer, explain your reasoning and confidence level.

For decision support:

I need to make a decision about [decision].

Here's what I know:
- Option A: [description and pros/cons]
- Option B: [description and pros/cons]
- Option C: [description and pros/cons]

My priorities are: [weighted criteria]

Please:
1. Analyze each option against my criteria
2. Identify risks and mitigations for each
3. Recommend the best option with reasoning
4. Note what additional information would change your recommendation

Analytical Reasoning Prompts

Root cause analysis:

Problem: [describe the problem]

Using the 5 Whys technique:
1. Why did this happen? → [answer]
2. Why did that happen? → [answer]
3. Why did that happen? → [answer]
4. Why did that happen? → [answer]
5. Why did that happen? → [answer]

Then identify:
- Root cause(s)
- Contributing factors
- Recommended solutions
- Prevention strategies

SWOT analysis:

Conduct a SWOT analysis for [subject].

Strengths: [internal positives]
Weaknesses: [internal negatives]
Opportunities: [external positives]
Threats: [external negatives]

For each category:
- List 3-5 items with brief explanations
- Prioritize by impact
- Suggest strategies that leverage strengths/opportunities and address weaknesses/threats

Common Mistakes to Avoid

• Mistake: Asking too broad a question — Fix: Narrow the scope to specific, answerable questions.
• Mistake: Not specifying output format — Fix: Define exactly what the analysis should look like.
• Mistake: Ignoring confidence levels — Fix: Always ask the model to rate its certainty.
• Mistake: Accepting insights without validation — Fix: Cross-reference with other sources.

Professional Tips & Tricks

• Use structured frameworks (SWOT, 5 Whys, etc.) to guide analysis.
• Always ask for confidence levels — it helps you know when to verify.
• For critical decisions, ask for analysis from multiple perspectives.
• Document assumptions the model makes — they may affect conclusions.

Key Takeaways

• Use the ANALYZE framework to structure analysis prompts.
• Be specific about what insights you need and for whom.
• Always request confidence levels for analytical conclusions.
• Use established frameworks to ensure comprehensive analysis.
• Validate critical insights with additional sources.

Next up: Building complete AI workflows with orchestration and chaining.