Data Analysis Prompt Examples

Guides AI through a systematic exploratory data analysis, covering distributions, correlations, outliers, and initial hypotheses from a dataset.

Perform an exploratory data analysis on the following dataset.

Dataset description: [WHAT THE DATA REPRESENTS]
Columns: [LIST COLUMNS WITH TYPES]
Row count: [NUMBER]
Time period: [DATE RANGE]

Analyse the following:

1. **Data Quality**:
   - Missing values per column (count and percentage)
   - Duplicate rows
   - Obvious data entry errors or impossible values
   - Data type mismatches

2. **Distributions**:
   - Summary statistics for numeric columns (mean, median, std, min, max, quartiles)
   - Value counts for categorical columns
   - Identify skewed distributions

3. **Relationships**:
   - Correlations between numeric variables (flag strong correlations > 0.7)
   - Cross-tabulations for key categorical variables
   - Any obvious segments or clusters

4. **Temporal Patterns** (if time-series):
   - Trends over time
   - Seasonality
   - Anomalous periods

5. **Initial Hypotheses**: Based on the above, suggest 3-5 hypotheses worth investigating further.

6. **Data Preparation Recommendations**: What cleaning or transformation would improve this data for modelling?

Present findings with specific numbers and clear visualisation recommendations.

Takes raw metrics data and generates an executive-friendly interpretation with trends, anomalies, and recommended actions.

Interpret these business metrics for [PERIOD] and provide an executive summary.

Metrics:
[PASTE METRICS DATA - e.g., revenue, users, conversion, churn, etc.]

Previous period comparison:
[PASTE PREVIOUS PERIOD DATA IF AVAILABLE]

Targets:
[PASTE TARGETS IF AVAILABLE]

Provide:

1. **Headlines** (3 bullet points): The most important things leadership needs to know

2. **Performance vs Target**: For each metric:
   - Actual vs target (% variance)
   - Trend direction and magnitude
   - Traffic light status: Green (on track) / Amber (watch) / Red (action needed)

3. **Anomalies**: Anything unusual that warrants investigation

4. **Root Cause Hypotheses**: For any metrics significantly off target, suggest 2-3 possible causes to investigate

5. **Recommended Actions**: Specific actions to take based on the data

6. **Forward Look**: Based on current trends, what should we expect next period?

Write for a non-technical executive audience. Lead with insights, not data.

Analyses A/B test results including statistical significance, practical significance, segment analysis, and clear recommendations.

Analyse the results of this A/B test.

Test details:
- Hypothesis: [WHAT WE EXPECTED]
- Primary metric: [METRIC NAME]
- Duration: [HOW LONG]
- Sample size: Control [N], Variant [N]

Results:
- Control: [METRIC VALUE]
- Variant: [METRIC VALUE]
- Relative change: [X%]
- p-value: [VALUE]
- Confidence interval: [RANGE]

Secondary metrics:
[LIST SECONDARY METRICS AND RESULTS]

Segment data (if available):
[RESULTS BY SEGMENT: device, geography, user type, etc.]

Analyse:
1. **Statistical Significance**: Is the result statistically significant at 95% confidence? Address sample size adequacy.
2. **Practical Significance**: Is the effect size large enough to matter for the business?
3. **Novelty Effects**: Any evidence this is a temporary novelty effect vs sustained change?
4. **Segment Analysis**: Does the effect vary significantly across segments? Any segments where the variant is worse?
5. **Secondary Metric Impact**: Any concerning movements in secondary metrics?
6. **Recommendation**: Ship / Don't ship / Extend test — with clear reasoning
7. **Caveats**: Limitations of this analysis and what to monitor post-launch

Performs cohort analysis on customer data to identify retention patterns, revenue trends by cohort, and factors that predict long-term customer value.

Perform a cohort analysis on the following customer data.

Data provided:
- Customer signup dates
- Monthly revenue per customer
- Churned date (if applicable)
- Acquisition channel
- Plan type

Analyse:

1. **Retention Curves by Cohort**: Month-over-month retention for each signup cohort. Identify:
   - Which cohorts retained best/worst
   - The critical "drop-off" month
   - Whether retention is improving over time

2. **Revenue by Cohort**: Average revenue per customer by cohort month. Track:
   - Revenue expansion within cohorts
   - Time to reach stable revenue
   - LTV trends across cohorts

3. **Acquisition Channel Performance**: Retention and LTV by acquisition channel. Which channels bring the most durable customers?

4. **Early Indicators**: What actions in months 1-2 predict long-term retention? (Feature adoption, engagement patterns, etc.)

5. **Churn Patterns**: When do customers typically churn? Is there a "danger zone" month?

6. **Recommendations**: Top 3 actions to improve retention based on the cohort patterns.

Present key findings with tables showing retention rates by cohort month.

Systematically checks a dataset for quality issues including completeness, consistency, accuracy, and timeliness, producing an actionable quality report.

Conduct a data quality audit on the following dataset.

Dataset: [NAME/DESCRIPTION]
Expected schema: [DESCRIBE EXPECTED COLUMNS AND TYPES]
Business rules: [LIST ANY KNOWN VALIDATION RULES]
Data source: [WHERE THE DATA COMES FROM]

Check for:

1. **Completeness**:
   - Missing values by column (count, percentage, pattern)
   - Are missing values random or systematic?
   - Impact of missing data on analysis

2. **Consistency**:
   - Values that contradict each other within a row
   - Format inconsistencies (dates, phone numbers, addresses)
   - Referential integrity issues

3. **Accuracy**:
   - Values outside expected ranges
   - Statistical outliers (> 3 standard deviations)
   - Known business rule violations

4. **Timeliness**:
   - Most recent data point
   - Any gaps in time series
   - Stale records that should have been updated

5. **Uniqueness**:
   - Duplicate records
   - Near-duplicate records (fuzzy matches)

For each issue found, provide:
- Severity: Critical / Warning / Info
- Count of affected records
- Example records
- Recommended remediation action

End with an overall data quality score (0-100) and a prioritised remediation plan.

Analyses survey responses including quantitative scores and free-text comments, identifying key themes, correlations between responses, and actionable recommendations.

Analyse these survey results comprehensively.

Survey: [SURVEY NAME/PURPOSE]
Respondents: [NUMBER] out of [TOTAL INVITED] (response rate: [X%])
Period: [WHEN CONDUCTED]

Quantitative responses:
[PASTE SUMMARY SCORES FOR EACH QUESTION]

Free-text responses:
[PASTE FREE-TEXT COMMENTS]

Demographic breakdowns (if available):
[SCORES BY DEPARTMENT, TENURE, ROLE LEVEL, etc.]

Provide:

1. **Overall Summary**: Key headline findings in 3 bullet points

2. **Quantitative Analysis**:
   - Scores vs benchmarks or previous survey
   - Highest and lowest scoring areas
   - Statistically significant differences across demographics

3. **Qualitative Theme Analysis**:
   - Top 5 themes from free-text comments
   - Sentiment by theme
   - Representative quotes for each theme

4. **Correlation Insights**: Do respondents who score low on X also score low on Y?

5. **Action Plan**: Top 5 recommended actions, prioritised by impact and feasibility

6. **Communication Draft**: A brief summary suitable for sharing results back to participants

Note any limitations due to sample size or response bias.