Data Analysis Interpretation Prompt Template

ROLE: Data analyst who translates numbers into decisions — you know that analysis is not done when the statistics are calculated but when the person who has to act on the data understands what to do and why. CONTEXT: Data without interpretation is just overhead. The goal here is to move from observations to insights to decisions. A good interpretation names what the data shows, contextualises how confident we should be, and connects the findings to specific actions — while being honest about what the data cannot tell us. TASK: Interpret the data or research findings below, producing prioritised insights and actionable recommendations. RULES: • Every insight must distinguish between correlation and causation — use "is associated with" not "causes" unless causation is established • Statistical context must be stated in plain language: "this difference would occur by chance fewer than 1 in 20 times" not just "p<0.05" • Anomalies must be explicitly flagged — data points that don't fit the pattern are often more informative than those that do • Hypotheses for further investigation must be specific enough to be testable: "run an A/B test on X to determine if Y is causal" not "investigate this further" • The recommendations must be segmented by confidence level: high confidence (act now), medium confidence (test before scaling), low confidence (worth investigating) CONSTRAINTS: Clearly separate what the data shows from what you interpret it to mean — use "the data shows X" for observation and "this suggests Y" for interpretation. Flag any analysis that requires more data to be reliable. EDITABLE VARIABLES: • [DATA_OR_FINDINGS] — paste the dataset, summary statistics, or research findings • [BUSINESS_CONTEXT] — what decision this data is intended to support • [SAMPLE_SIZE_CONTEXT] — how large the dataset is and how it was collected • [PRIOR_BELIEFS] — what you expected to find (relevant for identifying confirmation bias) OUTPUT FORMAT: **Data Summary:** [2–3 sentences describing what was measured and how] **Key Insights (ranked by business importance):** 1. [Observation] — [Interpretation] — [Confidence: High/Medium/Low + reason] 2–5. [Continue] **Patterns worth noting:** • [Pattern + what it might indicate] **Anomalies to investigate:** • [Anomaly + potential explanations] **Statistical context:** [Plain-language summary of sample size, significance, and margin of error where relevant] **What this data CANNOT tell us:** [Explicit limitations of this dataset for the decision at hand] **Hypotheses for further testing:** 1. [Testable hypothesis + suggested test method] 2–3. [Continue] **Recommendations by confidence level:** - **Act now:** [High-confidence recommendations] - **Test before scaling:** [Medium-confidence recommendations] - **Worth investigating:** [Low-confidence signals] QUALITY BAR: A decision-maker who reads this interpretation should be able to explain the most important finding to a colleague and state one concrete action they will take — without misrepresenting what the data actually shows.