Claude Academy
Loading

Data analysis with AI

Lesson 645 min
In this lessonBy the end, you’ll be able to
  • Use the Delegation-Diligence loop to systematically validate AI's analytical capabilities for your specific work
  • Apply Description and Discernment to identify patterns in your data while recognizing AI's limitations
  • Build confidence in AI-assisted analysis by testing against data you already understand

Data analysis with AI

Loading

Data analysis with AI · 7 min

SummaryTranscript

This video addresses a critical question that nonprofit professionals face when using AI for data analysis: How do I know I can trust the results? You'll follow Rio, a program director at Valley Veterans Services, as he uses the Delegation-Diligence loop to systematically validate AI's analytical capabilities using past data where he already knows the answers. The video demonstrates how to build confidence through testing, identify capability gaps, and develop an approach you can apply to new data with confidence.

Key takeaways

  • Test AI against data you already understand: Before trusting AI with new analysis, validate it using past data where you know the correct results. If AI can match your known results with the right guidance, you can confidently use it for similar future tasks
  • Use Discernment to identify gaps in AI's reasoning: As you test, note where AI misses important context and what additional Description you need to provide
  • Build validated approaches, not blind trust: Each testing round teaches you what AI does well and where it needs guidance. Document what works so you can replicate it
  • AI can help even if you're not data-savvy: If you're not comfortable with data analysis yourself, AI can help brainstorm solutions, write Excel formulas, and reformat messy data—just keep asking for clarifications so you understand the process
  • Validation builds confidence but doesn't eliminate responsibility: You're still accountable for checking that results make sense and being transparent about AI's role

Exercises

Messaging analysis

This exercise uses lower-stakes data (your own public communications) to practice the Description-Discernment loop for data analysis.

Part I: Gather your data

Collect 10-20 examples of your organization's communications—social media posts, email subject lines, newsletter headlines, or event announcements. Include a mix of what you consider high-performing and lower-performing content.

Part II: Analyze with AI

Share your dataset with AI and ask it to identify patterns:

  • What themes or topics appear in your higher-performing content?
  • What language, tone, or formatting patterns emerge?
  • Are there any gaps between what you communicate and what resonates?

Part III: Apply Discernment

Evaluate AI's analysis:

  • Do the identified patterns match your intuition about what works?
  • What context is AI missing about your audience or goals?
  • Are there patterns AI identified that surprise you?

Reflection:

  • What are you trying to learn from your dataset?
  • How does higher-performing content align with your authentic voice and organizational values?
  • Are you reaching the right audience?

Stretch goal: Use AI to audit how your messaging compares with your organization's stated mission and values, find discrepancies, and create a messaging guide from the analysis.

Lesson reflection

  • How did testing AI against data you already understood change your confidence in using it for new analysis?
  • What gaps or limitations did you identify that will shape how you delegate data analysis tasks in the future?

What's next

In the next lesson, we'll look at workflow automation—how to apply these same principles when AI handles routine tasks on your behalf, freeing up your time for higher-impact work.