How to Use These Prompts

These prompts work with ChatGPT (GPT-4o), Claude, or any capable AI assistant. Replace bracketed placeholders with your specific details. For SQL and code prompts, always review and test the output before using it in production.

Privacy note: Never paste real personal data, customer records, or confidential business data into public AI tools. Use anonymised or synthetic data when testing prompts.

SQL Generation Prompts

Generate a query from a description:

"Write a SQL query for [database type, e.g. PostgreSQL / SQL Server / BigQuery] that [describe what you need]. The relevant tables are: [table names and brief descriptions]. Include comments explaining each section."

Optimise a slow query:

"This SQL query is running slowly on a table with [number] rows. Suggest optimisations to improve performance. Here is the query: [paste query]. The table has indexes on: [list indexed columns]."

Explain a complex query:

"Explain what this SQL query does in plain English, step by step. Describe what each CTE or subquery does and what the final output will look like: [paste query]."

Generate a data quality check:

"Write SQL queries to check data quality in a table called [table name] with columns [list columns]. Check for: null values, duplicate records, values outside expected ranges, and referential integrity issues with [related table]."

Python Data Analysis Prompts

Data cleaning script:

"Write a Python script using pandas to clean a dataset with the following issues: [describe issues, e.g. inconsistent date formats in column X, missing values in column Y, duplicate rows based on column Z]. The dataframe is called df and has these columns: [list columns]."

Exploratory data analysis:

"Write a Python script to perform exploratory data analysis on a pandas dataframe called df with columns [list columns]. Include: summary statistics, distribution plots for numeric columns, correlation matrix, and identification of outliers. Use matplotlib and seaborn for visualisations."

Statistical test:

"I want to test whether [describe hypothesis, e.g. there is a significant difference in average order value between customer segments A and B]. My data is in a pandas dataframe with columns [list relevant columns]. What statistical test should I use, and write the Python code to perform it and interpret the results?"

Time series analysis:

"Write Python code to analyse a time series dataset with a date column [column name] and a value column [column name]. Include: trend decomposition, seasonality detection, and a simple forecast for the next [number] periods. Use statsmodels or Prophet."

Dashboard and Visualisation Prompts

Dashboard design advice:

"I need to design a dashboard for [audience, e.g. sales managers / executive team / operations team] that shows [describe what they need to monitor]. What metrics should I include, how should I structure the layout, and what chart types are most appropriate for each metric?"

Chart type selection:

"I want to visualise [describe the data and what you want to show, e.g. the relationship between marketing spend and revenue across 12 months for five product categories]. What chart type would be most effective, and why? Suggest alternatives if the primary option has limitations."

KPI definition:

"Help me define KPIs for a [type of team/function, e.g. customer service / supply chain / marketing] dashboard. For each KPI, suggest: the metric name, how to calculate it, what a good/bad value looks like, and what data sources are needed."

Stakeholder Reporting Prompts

Executive summary:

"Write an executive summary of the following data analysis findings for a non-technical audience. Keep it under 200 words, focus on business implications rather than methodology, and include a clear recommendation: [paste your findings]."

Insight narrative:

"I have the following data findings: [describe findings]. Write a narrative that tells the story of what this data means for the business. The audience is [describe audience]. Use plain language and avoid statistical jargon."

Presenting a negative finding:

"I need to present analysis showing that [describe negative finding, e.g. our new marketing campaign did not improve conversion rates]. Help me frame this constructively for a senior stakeholder audience, focusing on what we learned and what to do next."

Statistical Interpretation Prompts

Explain a statistical result:

"Explain what a [statistical result, e.g. p-value of 0.03 / R-squared of 0.67 / confidence interval of 45-55] means in plain English for a business audience. What does it tell us, and what are its limitations?"

Correlation vs causation:

"I have found a strong correlation between [variable A] and [variable B] in our data. Help me think through whether this is likely to be causal, what confounding factors might explain it, and how I would test for causation."

Sample size guidance:

"I want to run an A/B test on [describe what you are testing]. The current [metric] is approximately [value]. I want to detect a [percentage] improvement with [confidence level] confidence. How large does my sample need to be in each group?"

Tips for Better Results

• Provide context about your industry and the business question you are answering
• Specify your database type, Python version, and key libraries when asking for code
• Ask for explanations alongside code — "explain what each section does"
• Follow up with "What are the most common errors when implementing this?" for practical depth
• Always test generated SQL and code on a sample before running on production data