AI Prompt Pack for Data, BI and Finance Analysts: 50 Practitioner Prompts + Practice Datasets, SQLite Database and Excel Tools

Most analyst prompt packs are a few pages of one-line prompts such as "act as a data analyst". This one follows the way analysis actually happens. You frame the question, clean the data, write the query, explore it, test it, build the dashboard, and then turn the result into a decision.

WHAT'S INSIDE

• 50 original prompts in 9 workflow stages: Frame the question · Data cleaning and QA · SQL and PySpark · Exploration and stats · Experimentation and A/B testing · Dashboards and BI · Finance and risk analytics · Storytelling and stakeholder memos · Career and interview prep.

• Every prompt comes with a purpose line, fill-in [PLACEHOLDER] fields, a short example output (synthetic data) and a note on which AI tools it suits (ChatGPT, Claude, Gemini, Grok).

• 5 end-to-end workflows that chain prompts and say what to carry into each next step: "Why did the metric drop?", "A/B test from idea to decision", "KPI dashboard people actually use", "Monthly model monitoring (educational)" and "Land an analytics role".

• Built to make the AI check itself. The prompts ask the model to state its assumptions, flag data gaps instead of guessing, give validation queries, and show its arithmetic.

• Test notes. 14 prompts were test-run on synthetic data while the pack was built, and the code they produced was executed. 8 were improved as a result. The other 36 have not been test-run.

• Working files to run the prompts on (see below), so you can check the AI's SQL and numbers against a verified answer.

• 3 formats: a print-quality A4 PDF (60 pages), a Notion import (database CSV plus Markdown pages) and a Google Sheets/Excel file with one row per prompt.

WORKING FILES INCLUDED

• 4 synthetic datasets (CSV, each with a data dictionary): an online store (customers, 32,313 orders, daily traffic by channel), a SaaS subscription business (2,400 accounts, monthly MRR and churn), an A/B test event log (60,000 users), and a generic loans table for education only (5,000 loans with monthly status, plus a deliberately messy raw extract). Column names match the prompts' examples, and each dataset has planted issues to find.

• SQLite practice database (practice.db) with all 10 tables, plus 20 SQL practice questions with an answer key. Every answer query was executed and its result matched an independent pandas calculation.

• 3 Excel workbooks with live formulas: an A/B test calculator (rates, lift, z-test p-value, confidence interval, SRM check, sample size), a KPI dashboard template (month picker, plan vs actual, mix-vs-rate split, 3 charts) and a prompt tracker. They were recalculated in LibreOffice with no formula errors and open in Excel, Google Sheets and LibreOffice.

• "Run it on real files" guide mapping prompts to files, and the generator script (Python, fixed seed) that rebuilds the data.

• All data is synthetic. No real customer, company, lender or employer data was used or imitated.

WHO IT'S FOR

Data analysts, BI developers, product analysts, and finance or credit-risk analysts at junior to senior level. It also suits anyone preparing for analytics interviews.

IMPORTANT

This pack is for education and productivity only. It is not financial, credit, investment or career advice, and it does not guarantee any job, result or income. All examples use synthetic data. Don't paste confidential or customer data into AI tools unless your organisation allows it.

LICENCE

You may use the prompts and everything they generate for personal or commercial work, and publish your own analyses of the practice data (for example in a portfolio). You may not resell, share or redistribute the pack or its files.

Digital download. All sales are final.