AI Prompts for Data Analysis

Prompts for cleaning data, explaining charts, finding insights, writing SQL, using Python, and summarizing reports.

Who These Prompts Are For

Data analysts, business owners, marketers, students, and professionals working with data. These prompts help you clean data, find insights, explain charts, write SQL and Python, and summarize findings, with checks so you don’t trust bad output.

How to Use These Prompts

Copy any prompt into ChatGPT, Claude, Gemini, or another AI tool. Describe your data and replace placeholders like [columns] and [goal]. Anonymize sensitive data, test AI-written code on real data, and verify numbers before relying on them.

Data Cleaning Prompts

Plan a data cleanup:

I have a dataset with these columns and issues: [describe columns and problems — missing values, duplicates, inconsistent formats].

Give me a step-by-step cleaning plan: what to fix, in what order, and how to handle each issue (and the trade-offs of each choice, e.g., dropping vs. filling missing values). Flag decisions that could bias the data.

Write data cleaning code:

Write [Python/pandas or SQL] code to clean this data.

Columns: [list]. Issues to fix: [describe]. 

Provide the code with comments explaining each step. Note any assumptions and where I should review the result. Keep it safe — don't drop data silently without flagging it.

Standardize messy values:

I have a column [name] with inconsistent values: [paste examples].

Suggest how to standardize them, and write code ([Python/SQL]) to do it. List any values that are ambiguous and need a human decision rather than an automatic rule.

Insight Prompts

Find insights in data:

Here's a summary of my data: [paste numbers / a table / describe it]. Context: [what it's about, what I care about].

Identify the most important patterns, trends, and outliers. For each: what it suggests and a caveat about what could explain it otherwise. Flag where I'd need more data to be confident. Don't overclaim.

Suggest analyses to run:

I have data about [describe] and want to understand [question/goal].

Suggest the analyses worth running, in priority order, and what each would tell me. For each, note the method and what to watch out for (confounders, sample size, bias). Keep it practical for my skill level: [level].

Interpret results:

Here are my analysis results: [paste]. The question I was answering: [describe].

Interpret them in plain language: what they mean, how confident I should be, and what they don't tell me. Point out any misleading interpretations to avoid.

Chart Explanation Prompts

Explain a chart:

Here's a description of (or data behind) a chart: [paste/describe]. 

Explain what it shows, the key takeaway, and anything notable or surprising. Then flag anything that could be misleading about how it's presented (scale, cherry-picked range, etc.).

Recommend the right chart:

I want to visualize [describe the data and what I want to show] for [audience].

Recommend the best chart type and why, what to put on each axis, and how to keep it clear and honest (avoid distorting scales). Suggest a title that states the takeaway.

SQL and Python Prompts

Write a SQL query:

Write a SQL query for this.

Tables and columns: [describe]. What I want: [describe the result].

Provide the query with comments, and explain what it does. Note any assumptions about the schema. Tell me how to verify it returns the right thing before I trust it.

Write Python for analysis:

Write [Python/pandas] code to [describe the analysis].

Data: [describe columns/structure]. 

Provide commented code, explain each step, and note what to check in the output. Keep it readable, and flag any step where the approach involves a judgment call.

Report Summary Prompts

Summarize findings:

Summarize these analysis findings for [audience — e.g., a non-technical stakeholder]: [paste].

Lead with the key takeaway, then the supporting points, then the recommended action. Plain language, no jargon. Include a brief caveats line about confidence and limitations.

Write a data report:

Write a short data report on [topic] for [audience].

Findings: [paste]. Question it answers: [describe]. 

Sections: executive summary, key findings, what it means, and recommendations. Be honest about limitations and avoid presenting uncertain results as fact.