AI Prompts for Data Analysis

I need to analyze [DESCRIBE DATA/SITUATION] to answer this business question: [DESCRIBE QUESTION]. Help me structure the analysis: what is the precise question I am trying to answer, what data do I need, what type of analysis is appropriate (descriptive, diagnostic, predictive, prescriptive), what would a useful output look like, and what are the limitations I should be upfront about?

The business question is: [DESCRIBE QUESTION]. What metrics should I analyze to answer it? For each metric, explain: what it measures, how to calculate it from my data, what a good versus bad value looks like, and what additional context is needed to interpret it correctly. Data available: [DESCRIBE DATA SOURCES]

I ran an A/B test: [DESCRIBE TEST]. Here are the results: [PASTE RESULTS OR DESCRIBE]. Help me analyze whether the result is statistically significant, practically significant (not just statistically), and whether there are segments or confounders I should check before concluding the test was successful or failed.

Before I analyze this dataset, help me identify data quality issues. Here is a sample or description: [DESCRIBE DATA]. What should I check for: missing values, outliers, duplicates, inconsistent formats, implausible values? For each issue found, suggest how to handle it in the analysis.

I want to visualize [DESCRIBE DATA AND RELATIONSHIP]. For each of these scenarios: [LIST SCENARIOS], tell me which chart type is most appropriate and why. What are the most common mistakes people make when visualizing this type of data, and what makes a chart misleading versus informative?

Write Python code using pandas to analyze this dataset: [DESCRIBE DATA STRUCTURE]. I want to: [DESCRIBE ANALYSIS GOALS]. Include: loading the data, cleaning obvious issues, computing the metrics I need, and producing a summary output. Add comments explaining the analytical choices.

Write SQL queries to calculate these business metrics from this database schema: [DESCRIBE SCHEMA]. Metrics needed: [LIST METRICS, e.g. DAU, retention rate, revenue per user, churn]. For each metric, write the query and explain any non-obvious calculation choices, especially around time windows and user definitions.

I want to do a cohort analysis on [DESCRIBE USER BEHAVIOR, e.g. retention, revenue, feature adoption]. My data has: [DESCRIBE TABLE STRUCTURE]. Write the SQL or Python code to: define cohorts by [DIMENSION, e.g. signup week, first purchase month], calculate [METRIC] for each cohort over time, and output a cohort grid I can visualize.

My dataset contains outliers that may be affecting my analysis: [DESCRIBE DATA AND OUTLIER CONCERN]. Write code to: detect outliers using [METHOD, e.g. IQR, z-score, or suggest appropriate method], visualize their distribution, and give me the options for handling them with the pros and cons of each approach for my specific use case.

I need to transform this data from [DESCRIBE CURRENT FORMAT] to [DESCRIBE DESIRED FORMAT] for my analysis. Write the pandas or SQL code to do this transformation. Common transformations: pivot/unpivot, groupby aggregation, rolling windows, merging tables on [KEYS].

My analysis shows a correlation between [VARIABLE A] and [VARIABLE B]: [DESCRIBE FINDING]. Help me think through whether this could be causal or whether there are plausible confounders or alternative explanations. What additional analysis would help distinguish correlation from causation here?

I am about to present this analysis: [DESCRIBE ANALYSIS AND FINDINGS]. Review it for common analytical biases: survivorship bias, selection bias, Simpson's paradox, p-hacking, and confirmation bias. Where might any of these be affecting my conclusions?

My analysis shows this result: [DESCRIBE FINDING WITH NUMBERS]. Is this result statistically significant? Given my sample size of [N], what is the confidence interval? What would I need for this result to be conclusive rather than suggestive? Explain in terms a non-statistician can act on.

Based on this analysis: [DESCRIBE ANALYSIS], what questions am I unable to answer with this data alone? What additional data would I need to be confident in the conclusion? What assumptions am I making that I should be explicit about when presenting this?

My analysis produced this result: [DESCRIBE RESULT]. This seems [SURPRISING/TOO CLEAN/UNEXPECTED]. Help me sanity check it. What are the most common causes of results like this being wrong: data pipeline issues, query errors, definition mismatches, or sampling problems? Walk me through how to verify the result is accurate before presenting it.

Write a clear executive summary of this analysis: [DESCRIBE ANALYSIS AND FINDINGS]. The audience is [DESCRIBE AUDIENCE, e.g. non-technical executives, product team]. The summary should: lead with the answer, not the methodology, explain the key finding in one sentence, provide the supporting evidence, and end with a clear recommendation or next step.

I want to present this data as a story rather than a collection of charts: [DESCRIBE DATA AND FINDINGS]. Help me structure the narrative: what is the central tension or question, what does the data reveal, what is surprising, and what should the audience do differently as a result? Write the narrative arc I can follow in my presentation.

I need to present these findings to two different audiences: [DESCRIBE FINDINGS]. Write a version for technical stakeholders (including methodology and caveats) and a version for executives (focusing on implications and decisions). Each should be no longer than one page.

I am building a dashboard to track [DESCRIBE METRICS AND BUSINESS QUESTION]. Write a brief that defines: which metrics to show, how to define each one precisely, what time dimensions to use, what target or benchmark to show alongside each metric, and what drill-down views are most valuable. The audience is [DESCRIBE].

Based on this analysis: [DESCRIBE FINDINGS], write a clear recommendation memo. Format: one-sentence conclusion, the three strongest data points that support it, the most important caveat or risk, and a specific recommended action with an owner and timeline. This will be shared with [DESCRIBE AUDIENCE].