Best AI Prompts for Data Analysts in 2026 (Query Faster, Explain Better, Automate More)

**Prompt 1: SQL Query Generator** Use this when: you need to write a SQL query from a plain-English description and want a documented, production-ready first draft. Write a SQL query to [describe what the query should return — e.g., 'return the top 10 customers by total revenue in Q2 2026, broken down by product category']. Database: [MySQL / PostgreSQL / BigQuery / Snowflake / other]. Tables available: [list the relevant table names and key columns]. Include comments explaining each major step of the query — what it's selecting, why the joins are structured this way, and what any filters are doing. Format the output so it's readable (consistent indentation, one clause per line). If there are multiple approaches, note the tradeoffs. Why it works: Generating the first draft with comments forces a structured, documented query you can review, edit, and hand off — rather than starting from a blank editor and debugging from scratch.

**Prompt 2: Query Performance Optimizer** Use this when: you have a working query that's running slowly or consuming too many resources and need to identify where to improve it. Review this SQL query and identify performance bottlenecks or optimization opportunities: [PASTE YOUR SQL QUERY HERE] Database: [MySQL / PostgreSQL / BigQuery / Snowflake / other]. Table sizes (approximate): [list tables and row counts if known]. Identify: (1) Any full table scans that could be replaced with indexed lookups, (2) Joins that could be restructured for better performance, (3) Subqueries that could be replaced with CTEs or window functions, (4) Aggregations happening on too large a dataset before filtering, (5) Any syntax patterns specific to [database] that could be optimized. For each issue: describe the problem, explain why it affects performance, and suggest the revised code. Prioritize by impact. Why it works: Performance reviews require knowing what to look for — this prompt gives you a systematic framework for catching the most common issues before they become production problems.

**Prompt 3: SQL-to-Python Pandas Converter** Use this when: you have a SQL query that needs to run in a Python environment (Jupyter, a pipeline, or a script) instead of directly against a database. Convert this SQL query to Python pandas code: [PASTE YOUR SQL QUERY HERE] Assume the input data is available as pandas DataFrames with the same names as the SQL table names. Include: (1) The equivalent pandas operations for each SQL clause (SELECT, FROM, JOIN, WHERE, GROUP BY, ORDER BY, LIMIT), (2) Comments explaining which SQL operation each pandas step corresponds to, (3) Any edge cases where the pandas behavior differs from SQL behavior (e.g., handling of NULLs/NaN, join types). Output clean, readable pandas code with proper variable naming. Why it works: SQL-to-pandas translation is one of the most time-consuming context switches in analytics work — especially when you need to move logic from a database query into a data pipeline or Jupyter notebook.

**Prompt 4: Duplicate Record Finder** Use this when: you need to identify duplicate records in a dataset before loading, reporting, or sending data downstream. Write a SQL query to find duplicate records in the [TABLE NAME] table based on [list the columns that define a duplicate — e.g., 'customer_id and order_date']. The query should: (1) Return all rows where a duplicate exists (not just one representative row), (2) Include a count of how many duplicates exist for each combination of [columns], (3) Order results so the duplicates are grouped together for easy review. Database: [MySQL / PostgreSQL / BigQuery / Snowflake]. Also include a second query that returns only the IDs of rows to DELETE if I want to keep one record per duplicate group (specify which record to keep: the most recent by [date column] or the one with the lowest ID). Why it works: Duplicate detection is a high-frequency data quality task — having a reliable, reusable query template for every table structure saves significant debugging time.

**Prompt 5: Plain-English Query Explainer** Use this when: you need to explain what a SQL query does to a stakeholder, a new team member, or for documentation — without requiring them to read SQL. Explain what this SQL query does in plain English for a non-technical stakeholder: [PASTE YOUR SQL QUERY HERE] The audience is [a business stakeholder with no SQL knowledge / a new team member / a product manager / a C-suite executive]. Your explanation should: (1) Describe what data the query starts with, (2) Explain each major step in plain language — what the query is filtering, grouping, calculating, and sorting, (3) Describe exactly what the output will look like (columns, what each column represents), (4) Summarize the business question this query is answering in one sentence. Avoid technical SQL jargon. If the query has any data quality assumptions baked in, flag them. Why it works: Explaining queries to non-technical stakeholders is one of the most time-consuming communication tasks for data teams — this prompt produces a documentation-ready plain-English summary in seconds.

**Prompt 6: Executive Summary from Data** Use this when: you have a data trend or chart result that needs to become an executive-level written summary — not a data dump. Write an executive summary explaining this data trend for a leadership audience: [PASTE YOUR DATA OR CHART SUMMARY HERE — e.g., 'Monthly revenue: Jan $420K, Feb $395K, Mar $412K, Apr $438K, May $467K, Jun $501K. YoY growth: +18%. Top-performing segment: Enterprise, +27% YoY. Lagging segment: SMB, -4% YoY.'] The audience is [the executive team / the CEO / the board]. Include: (1) The headline finding in one sentence — what's the most important thing they need to know, (2) The trend in plain language — what's going up, what's going down, what's changed, (3) The most likely business driver behind the trend, (4) The implication for decision-making — what should they do or watch. Keep the total summary under 150 words. Write in direct, confident prose — no bullet points. Why it works: Converting data output into executive-level narrative is one of the highest-value analyst communication skills — and one of the most time-consuming to do well from scratch.

**Prompt 7: Weekly Data Report Builder** Use this when: you produce a recurring weekly or monthly report for an internal team and want a structured, professional first draft in minutes. Draft a weekly data report for the [DEPARTMENT NAME — e.g., Marketing, Sales, Product, Finance] team using this data: [PASTE YOUR DATA HERE — metrics, tables, or bullet-point summaries] Report structure: (1) Week in review — 2-3 sentences on the overall performance story this week, (2) Key metrics — format as a simple table: metric name, this week's value, last week's value, % change, (3) What's working — 2-3 bullet points on positive trends, (4) What needs attention — 2-3 bullet points on metrics that are underperforming or trending in the wrong direction, (5) Next week focus — 1-2 recommended actions based on the data. Tone: professional, direct, data-forward. Assume the audience is familiar with these metrics but not with the underlying data this week. Why it works: Recurring reports follow a predictable structure — AI drafts the narrative in the same time it takes to format the data table.

**Prompt 8: Dashboard Slack Summary** Use this when: you need to share dashboard metrics in a Slack message or internal update to a non-technical team. Write a Slack message summarizing these dashboard metrics for a non-technical audience: [PASTE YOUR METRICS HERE — e.g., 'This week: 12,400 new users (+8% WoW), $89K MRR (+3% WoW), churn rate 2.1% (vs 1.9% last week), NPS 42 (vs 39 last month)'] The audience is [the sales team / the marketing team / the product team / leadership]. Format for Slack: (1) Lead with the most important number and what it means in plain language, (2) Use bullet points for supporting metrics — each bullet should have the number AND a plain-language interpretation, (3) Flag any metric that needs attention in bold, (4) End with one sentence on what to watch next week. Keep the total message under 150 words. No jargon, no percent signs without plain-language context. Why it works: Dashboard-to-Slack communication is one of the highest-frequency reporting tasks for analyst teams — and one of the easiest to do poorly when you're pressed for time.

**Prompt 9: Chart Type Recommender** Use this when: you know what data you want to visualize but aren't sure which chart type will communicate it most effectively. Suggest the best chart type for visualizing [describe the data you want to show — e.g., 'monthly revenue broken down by 5 product categories over the past 2 years' / 'the distribution of customer lifetime value across 10,000 customers' / 'the correlation between ad spend and new signups by channel']. My audience: [executives / technical stakeholders / a public-facing report]. My tool: [Tableau / Power BI / Looker / Python matplotlib / Google Data Studio / Excel]. Include: (1) Your primary recommendation and why it's the best fit for this data, (2) One alternative chart type and when you'd use it instead, (3) Any data preparation steps required before visualizing (aggregation, normalization, etc.), (4) One design tip specific to this chart type for a non-technical audience. Why it works: Chart selection is a deceptively high-stakes decision — the wrong chart type obscures the insight; the right one makes it undeniable.

**Prompt 10: Data Story Narrative** Use this when: you have numbers that tell a business story but need to translate them into a compelling narrative for a presentation, report, or leadership update. Create a data story narrative that explains [describe the business outcome — e.g., 'why customer acquisition cost increased in Q2' / 'why engagement dropped after the product redesign' / 'the ROI of the new pricing strategy'] using these numbers: [PASTE YOUR DATA HERE] The narrative should follow this structure: (1) The situation — what were we trying to understand or what question prompted this analysis, (2) The finding — what the data actually shows, (3) The explanation — what likely caused this pattern (be clear about what's confirmed vs. hypothesized), (4) The implication — what this means for the business and what decision it should inform, (5) The recommendation — what you'd suggest doing next and why. Tone: confident, analytical, clear — not hedging. Under 250 words. Why it works: Data storytelling is the skill that separates analysts who influence decisions from analysts who produce reports no one acts on — and the narrative structure is learnable and repeatable.

**Prompt 11: Dataset Cleaner Script** Use this when: you have a raw dataset with common quality issues — duplicates, nulls, inconsistent column names — and need a Python script to standardize it before analysis. Write Python code to clean this dataset — remove duplicates, handle null values, and standardize column names: [PASTE A SAMPLE OF YOUR DATA OR DESCRIBE THE STRUCTURE — e.g., 'CSV with columns: Customer ID, first name, Last Name, email address, Phone Number, sign_up_date, order_total. Issues: duplicate rows on Customer ID, nulls in phone_number and order_total, inconsistent capitalization in column names'] The script should: (1) Load the data from a CSV file (use a configurable file path variable), (2) Standardize all column names to snake_case lowercase, (3) Remove exact duplicate rows, (4) Handle nulls: for numeric columns, fill with 0 or column median (specify which); for string columns, fill with 'Unknown', (5) Print a summary before and after cleaning: row count, null counts per column, duplicate count. Include comments explaining each step. Use pandas. Why it works: Dataset cleaning follows the same pattern for 80% of raw data inputs — having a reusable, well-commented script template saves the first 45 minutes of every new data project.

**Prompt 12: DataFrame Merge with Mismatch Flagging** Use this when: you're combining two data sources and need to identify records that don't match cleanly — before bad joins corrupt your analysis. Write a pandas script to merge these two DataFrames on [JOIN COLUMN — e.g., 'customer_id'] and flag any mismatches: DataFrame 1 description: [describe — e.g., 'customer master list from CRM: columns customer_id, name, email, plan_type, signup_date — 15,200 rows'] DataFrame 2 description: [describe — e.g., 'billing records from payment system: columns customer_id, billing_name, email, mrr, last_payment_date — 14,800 rows'] The script should: (1) Perform a full outer join on [join column], (2) Create a 'match_status' column: 'matched', 'left_only' (in DataFrame 1 but not 2), or 'right_only' (in DataFrame 2 but not 1), (3) For matched records, flag rows where [COLUMN] doesn't match between the two sources (e.g., email address or name), (4) Output three DataFrames: matched_clean, left_only, right_only, (5) Print a summary: total matched, total mismatches, left-only count, right-only count. Why it works: Data join quality issues are one of the most common — and most silent — sources of analytical errors. This script surfaces them before they make it into a report.

**Prompt 13: Scheduled Automation Function** Use this when: you perform a repetitive data task regularly (daily export, weekly report pull, monthly reconciliation) and want to automate it with a Python script that runs on a schedule. Create a Python function that automates [DESCRIBE THE TASK — e.g., 'pulling yesterday's sales data from a PostgreSQL database, aggregating by product category and region, and saving the result as a timestamped CSV to a /reports folder'] and runs on a schedule. Requirements: (1) The function should be modular — one function for data extraction, one for transformation, one for output/saving, (2) Include error handling: if the data pull fails or returns zero rows, log the error and send an alert (use Python logging module), (3) Add a main() function that calls the steps in sequence and can be run from the command line, (4) Show how to schedule it using Python's schedule library (for simple use cases) or cron syntax (for production use), (5) Include a config section at the top of the script for any variables that might change (database connection string, file paths, date ranges). Add comments throughout. Why it works: Turning a manual task into a scheduled script once eliminates that task forever — the structure above is reusable for virtually any recurring data workflow.

**Prompt 14: Python Script Reviewer** Use this when: you've written or inherited a Python script and want a code review that explains what it does and surfaces any issues before it runs in production. Review this Python script, explain what it does, and suggest any improvements: [PASTE YOUR PYTHON CODE HERE] Provide: (1) A plain-English summary of what this script does — step by step, (2) Any bugs or logical errors that would cause incorrect output or runtime failures, (3) Performance issues — any operations that are unnecessarily slow or memory-intensive for large datasets, (4) Code quality improvements — any operations that could be simplified, refactored, or made more readable, (5) Missing error handling — any failure modes the script doesn't currently catch. For each issue: describe the problem, the risk, and the suggested fix with revised code. Separate into: Critical (fix before running), Recommended (fix when you have time), Optional (style improvements). Why it works: Code review is the most reliable way to catch silent data errors before they corrupt an analysis — and having a structured review checklist ensures you catch the right things.

**Prompt 15: Regex Pattern Builder** Use this when: you need to extract structured data from messy string fields — phone numbers, email addresses, product codes, transaction IDs, or any pattern embedded in free text. Write a regex pattern to extract [TYPE OF DATA — e.g., 'US phone numbers in any common format' / 'email addresses' / 'product SKU codes in the format ABC-12345' / 'dollar amounts with or without decimal places'] from these strings: [PASTE 5-10 EXAMPLE STRINGS THAT CONTAIN THE DATA — e.g., 'Contact us at (415) 555-1234 or 415.555.1234', 'Order ref: SKU-A1234-B, total: $1,299.99'] Provide: (1) The regex pattern itself, (2) A Python code snippet using re.findall() or re.search() that applies the pattern to a pandas column, (3) Explanation of each component of the regex in plain English, (4) Edge cases the pattern handles and any it intentionally excludes, (5) If there are multiple valid approaches, note the tradeoff between precision and recall. Test against all the example strings provided. Why it works: Writing regex from scratch is one of the most time-consuming data cleaning tasks — AI can generate, explain, and test a pattern in under a minute.

**Prompt 16: CEO-Ready Executive Summary** Use this when: you have a detailed data analysis and need to distill it into a 3-bullet executive summary for senior leadership. Translate these data findings into a 3-bullet executive summary for the CEO: [PASTE YOUR FINDINGS HERE — full analysis, charts, key numbers, context] Requirements: (1) Each bullet should be one sentence — lead with the business implication, not the data point (e.g., 'Customer acquisition cost is increasing faster than revenue per customer, compressing margin' NOT 'CAC increased from $42 to $67 over 6 months'), (2) Each bullet should reference the key number that supports it — but embedded in the business-language sentence, not leading it, (3) The three bullets should flow logically: what's happening → why it matters → what to do, (4) Keep each bullet under 25 words. If there's a recommended action, include it as a fourth line: 'Recommended action: [one specific thing].' Total: under 100 words. Why it works: Executive summaries that lead with business implications — not data outputs — get read, remembered, and acted on.

**Prompt 17: Stakeholder Data Request Response** Use this when: you receive a data request that's unclear, not feasible with current data, or needs scoping before you can begin. Write a response to this stakeholder data request that clarifies what's actually possible with our current data: [PASTE THE REQUEST HERE — e.g., 'Can you pull me a report showing why our conversion rate dropped last month and which customer segments were affected?'] Our current data situation: [describe what data you have, what you don't, and any relevant limitations — e.g., 'We have session data and purchase data but not individual user journey data. Attribution is last-click only. We can segment by cohort but not by individual demographics.'] Draft a professional response that: (1) Acknowledges the business question they're trying to answer, (2) Clarifies what's possible vs. not possible with current data and why (no jargon), (3) Proposes what you CAN deliver — the closest feasible analysis, (4) Asks 1-2 clarifying questions if needed to scope the work, (5) Provides a realistic timeline. Tone: helpful, transparent, not defensive. Why it works: Data requests that aren't scoped properly waste everyone's time — a clear, professional response that reframes what's possible builds trust and avoids rework.

**Prompt 18: Analytics Presentation Outline** Use this when: you're presenting quarterly or monthly analytics results to leadership and need a structured outline before building slides. Draft a presentation outline for sharing Q[QUARTER] analytics results with the leadership team. Business context: [brief description of your company, business model, and what the analytics team tracks]. Key metrics for this quarter: [list the 5-8 metrics you'll cover]. Significant findings: [list 2-3 major observations from the data — what changed, what's surprising, what needs action]. Build a 10-slide outline with: (1) Title and agenda, (2) Quarter in review — the headline story in 3 bullet points, (3-7) One slide per major metric area (structure each as: metric + trend + business interpretation + recommendation), (8) What's working — top 3 positive findings, (9) What needs attention — top 3 risks or underperforming areas, (10) Recommended actions and next quarter priorities. Include 2-3 talking points per slide. Why it works: Analytics presentations that tell a coherent story — rather than a dashboard tour — change decisions. Starting with an outline forces the narrative before the slides.

**Prompt 19: Statistical Significance Explainer** Use this when: you need to explain a statistical concept like p-values, confidence intervals, or sample size to a non-technical colleague or stakeholder without losing them. Write a one-paragraph explanation of statistical significance for a non-technical marketing manager who needs to understand whether a recent A/B test result is actionable. The result: [describe the test — e.g., 'Variant B had a 12% higher click-through rate than the control, p=0.08, n=1,200 total visitors']. The paragraph should: (1) Explain what statistical significance means in plain English — without the phrase 'statistical significance' if possible, (2) Interpret this specific result in business terms — is this result reliable enough to act on or not, (3) Explain what they should do next: run the test longer, act on the result, or investigate differently. Under 150 words. No formulas. No jargon without immediate plain-English translation. Why it works: Statistical literacy explanations that start with the business decision — not the math — are the ones stakeholders remember and act on.

**Prompt 20: Data Request Template** Use this when: you need data from another team — engineering, product, finance, or operations — and want a clear, professional request that gets you what you need without back-and-forth. Create a data request template I can send to other teams when I need data from them. The template should include fields for: (1) The business question I'm trying to answer (1-2 sentences), (2) The specific data I need — table name or data source, specific fields/columns, date range, filters, (3) The format I need it in — CSV, database access, API, shared dashboard, (4) How I plan to use it — the analysis or report it will feed into, (5) The deadline and why it's time-sensitive if applicable, (6) Any questions I have about data definitions, data quality, or known issues with this data. Make the template professional and easy to fill in quickly. Include a brief covering note I can customize for different teams. Why it works: Ambiguous data requests create back-and-forth that wastes both teams' time — a clear, structured request gets you the right data faster.

**Prompt 21: Resume Bullet Point Rewriter** Use this when: you have a list of job responsibilities and need to rewrite them as achievement-focused bullet points that quantify impact and use strong action language. Rewrite these job responsibilities as achievement-focused bullet points for my data analyst resume: [PASTE YOUR CURRENT RESPONSIBILITIES — e.g., 'Responsible for building and maintaining reporting dashboards. Worked with marketing and product teams on data analysis. Wrote SQL queries to pull data for business teams. Created visualizations in Tableau.'] For each bullet: (1) Start with a strong past-tense action verb (built, designed, reduced, automated, accelerated, etc.), (2) Include the scale or context — how many dashboards, how many stakeholders, what business function, (3) Quantify impact where possible — if I haven't provided numbers, suggest what metrics to include, (4) Keep each bullet under 20 words. Output 2-3 improved versions of each bullet so I can choose the best one. Why it works: Most analyst resumes describe duties rather than impact — quantified achievement bullets are what hiring managers remember and what get you to the interview stage.

**Prompt 22: Senior Data Analyst Cover Letter** Use this when: you're applying for a senior data analyst role and need a cover letter that opens with impact, not interest. Write a cover letter for a Senior Data Analyst role at [COMPANY TYPE — e.g., a Series B SaaS startup / a Fortune 500 consumer goods company / a fintech firm] emphasizing [YOUR STRENGTHS — e.g., 'SQL expertise, stakeholder communication, and Python automation experience']: [PASTE THE JOB DESCRIPTION HERE] My background: [describe your experience — years in analytics, tools used, domain expertise, notable achievements with numbers if possible]. Tone: direct, confident, specific — not humble. Under 350 words. Open with your most relevant achievement or the strongest match between your experience and the role requirements — not 'I am writing to apply.' Close with a clear statement of what you can deliver in the first 90 days. Why it works: Data analyst cover letters that lead with a relevant achievement get interviews; letters that lead with enthusiasm don't.

**Prompt 23: Interview Prep for Data Analyst Roles** Use this when: you're preparing for a behavioral interview for a data analyst role and want to practice structured, compelling answers. Generate 10 behavioral interview questions for a data analyst role and give me the ideal STAR-format answer structure for each. The role is at [COMPANY TYPE / industry]. My experience level: [years, domain, tools]. For each question: (1) State the question, (2) Explain why interviewers ask it — what they're actually evaluating, (3) Provide a STAR framework specifically for this question: what Situation and Task details to include, what Action demonstrates the right skill, what Result to emphasize, (4) Give an example of a weak answer and a strong answer for contrast. Focus the 10 questions on: technical-to-business communication, stakeholder management under ambiguity, working with imperfect data, influencing decisions with analysis, and cross-functional collaboration. Why it works: Behavioral interview preparation is highest-leverage when you understand what the question is actually testing — not just what story to tell.

**Prompt 24: LinkedIn Summary for Data Analysts** Use this when: you're updating your LinkedIn profile to attract opportunities — recruiter outreach, internal visibility, or freelance consulting work. Write a LinkedIn summary for a data analyst with [X] years of experience specializing in [DOMAIN — e.g., 'e-commerce analytics and customer lifetime value modeling' / 'healthcare data and clinical reporting' / 'B2B SaaS product analytics and funnel optimization']. My target audience: [recruiters at tech companies / hiring managers in fintech / people who might hire me for freelance analytics consulting]. My differentiator: [what makes you different from other analysts — e.g., 'I specialize in translating complex data into decisions non-technical leaders can act on' / 'I bridge the gap between data engineering and business stakeholders']. Tone: professional, specific, first person — no buzzwords. Under 220 words. Include a closing line that signals what I'm open to or available for. Why it works: LinkedIn summaries that describe your specific domain and differentiator — not just 'data-driven professional' — appear in recruiter searches and generate inbound opportunities.

**Prompt 25: 90-Day Analytics Transition Plan** Use this when: you're a data analyst who wants to move into a data science or analytics engineering role and need a structured learning and positioning plan. Create a 90-day learning plan for a data analyst who wants to transition into a [data science / analytics engineering / senior analytics / ML engineering] role. My current skills: [list your current tools and strengths — e.g., 'SQL (advanced), Python (intermediate), Tableau, some pandas and statsmodels']. The target role requires: [describe the skills gap — e.g., 'machine learning (sklearn, model deployment), dbt for analytics engineering, stronger Python for production pipelines']. Structure the plan as: (1) Days 1-30: Foundation — the 2-3 skills to learn first and the specific resources to use (courses, books, projects), (2) Days 31-60: Application — 1-2 projects to build that demonstrate the new skills, with project brief descriptions, (3) Days 61-90: Positioning — how to update your resume, LinkedIn, and portfolio to reflect the new direction; 2-3 specific job titles to target and companies to research. Include checkpoints for each phase. Why it works: Career transitions fail most often not from skill gaps but from unclear sequencing — this prompt creates a specific, executable plan rather than a vague learning list.