| name | hypothesis-driven-data-analysis |
| description | Use this skill when analyzing data to find actionable business solutions. It helps you decide between hypothesis-driven thinking and data mining approaches, ensuring you prioritize logical causation over mere statistical correlation. Apply during strategic planning, product development, or marketing initiatives when you have access to data and a business problem to solve. |
Hypothesis-Driven Data Analysis
When to Use
- You are analyzing data to find actionable business solutions
- You have access to data (Big Data or Small Data) and a business problem to solve
- You need to determine whether to trust correlations or develop hypotheses
- You are working on strategic planning, product development, or marketing initiatives
Prerequisites
- Access to relevant data sources
- A clearly defined business problem
- Understanding of the target customer or market context
Core Workflow
Step 1: Evaluate Data Mining Approach
When using Big Data or AI-driven analysis:
- Identify correlations in the data (e.g., products purchased together, behavioral patterns)
- Assess the risk - correlations may be coincidental or lack logical causality
- Verify the logical "why" before implementing any actions based on the correlation
Warning: Do not rely solely on AI/Big Data outputs without logical verification.
Step 2: Apply Hypothesis Thinking
When developing a hypothesis-driven approach:
- Imagine specific customer scenarios - create concrete customer personas with pain points and desired gains
- Reduce complex problems to simple, meaningful hypotheses
- Test the hypothesis even if it contradicts standard industry practices
Step 3: Decision Logic
Use this decision tree to determine your approach:
- IF Big Data shows a correlation BUT no logical hypothesis exists → Be cautious of implementation
- IF a strong, simple hypothesis exists based on customer empathy → Prioritize this over complex data analysis
- IF both correlation and hypothesis align → Proceed with confidence
Key Constraints
- Correlation does not equal causation - always seek the underlying logical explanation
- Avoid basket analysis without logic - don't act on statistical patterns alone
- Customer empathy is essential - hypotheses must be grounded in real customer needs
Output
An actionable business strategy derived from logical causation rather than just statistical correlation.
Anti-Patterns to Avoid
- Implementing changes based solely on statistical correlations without understanding the "why"
- Over-relying on AI/Big Data outputs without human verification
- Ignoring simple, customer-centered hypotheses in favor of complex data analysis
