Learn to ask a database anything.
Read the guide: How to learn SQL for data analysis: an 8 week plan
Direct Answer: Learn to ask a database anything — from your first SELECT to window functions, query plans, and turning a vague business question into a number you can defend.
The relational model, the schema this whole course uses, and your first real queries.
Collapsing thousands of rows into the handful of numbers someone actually asked for.
Every join type with worked results, and the three ways joins produce confidently wrong numbers.
Composing queries out of named steps, and computing across rows without collapsing them.
Fixing messy data in SQL, making queries fast on purpose, and designing tables that stay correct.
Inserts, updates and deletes you can run against a live table, and loads that survive being run twice.
Models other people depend on, tests that fail loudly when the data goes wrong, and the rules for running SQL inside an application.
The analysis people start asking once they trust your models, from subtotals in a single pass to retention curves built out of raw event logs.
Reading the planner's reasoning instead of arguing with it, feeding it better statistics and access paths, and rewriting the queries it can never fix on its own.
What the database actually does when hundreds of sessions touch the same rows at once, and the slow mornings that follow one transaction nobody closed.
Tables that outgrew a single index, engines that charge you by the byte scanned, and the answers that are worth approximating.
Modelling history so that last quarter still reports the same figure it did last quarter, and owning the definitions every dashboard in the company inherits.
Topic checks
Each one opens the next topic
Chapter quizzes
Each one opens the next chapter
50-question final
50 of 100 marks to pass
Certificate
Issued instantly, verifiable
One attempt, 60 minutes. If you don't pass, you can retake once after 30 days.
To learn SQL for data analysis, install a free database such as PostgreSQL, load one small dataset you understand, and learn the language in this order: SELECT and WHERE, then aggregation with GROUP BY, then joins, then CTEs and window functions. At every step, check your answer against something you can count by hand. The syntax takes a few weeks. Learning to notice when a query returns a confident, wrong number is the real skill, and it is the part most tutorials skip.
This guide gives you an 8 week roadmap, the mistakes that catch almost every beginner, and a chapter by chapter look at our free SQL & Data Analysis course, which teaches all of it on one small store database so every lesson builds on the one before.
Spreadsheets are where most analysis starts, and they stay useful. But once the data lives in a company database, someone has to ask that database a question in its own language. That language is SQL, and its core has stayed stable for decades, so what you learn now keeps paying off.
A query is written down, so someone else can read it, rerun it next month and check your logic. That is why analyst roles so often list it, and why our guide on how to become a data analyst in Pakistan puts it straight after spreadsheets.
You do not need to be a programmer. SQL is declarative: you describe the result you want, and the database works out how to get it. If you can build a PivotTable, you already think in the right shapes.
Very little:
Plan on about an hour a day, five days a week.
| Week | Focus | You can answer |
|---|---|---|
| 1 | Tables, keys, SELECT, WHERE, ORDER BY, NULLs, dates | "Which orders were placed in June?" |
| 2 | COUNT, SUM, AVG, GROUP BY, HAVING, medians | "Revenue per category, only categories above a threshold" |
| 3 | INNER, LEFT and other joins, anti-joins, self-joins | "Which customers have never ordered?" |
| 4 | Subqueries, CTEs, window functions | "Each customer's running total and rank" |
| 5 | Cleaning messy data, indexes, query plans, constraints | "Why does this report take minutes?" |
| 6 | INSERT, UPDATE, DELETE, transactions, safe reloads | "Can I fix this data without breaking it?" |
| 7 | Views, data tests, ROLLUP, funnels, retention cohorts | "Monthly retention by signup cohort" |
| 8 | Practice project and review | A written analysis someone else can rerun |
Start with the shape of a table: one row states one fact about one thing, and keys connect tables to each other. Then filter and sort. Spend real time on NULL, because WHERE city <> 'Lahore' quietly drops every customer whose city is missing, and nothing warns you.
Aggregation comes next. Learn exactly what COUNT(*) counts compared with COUNT(column), and why a filter on individual rows belongs in WHERE while a filter on a group's total belongs in HAVING. Learn the median too, because a single huge order can drag an average far away from what a typical customer spends.
Joins are where wrong answers come from. When you join orders to order items, each order appears once per line item, so summing the order total afterwards counts it several times. Revenue looks healthy and is simply false. The fix is to aggregate the line items to one row per order first, then join.
After joins, CTEs let you name each step of a query so you can test them one at a time. Window functions then let you rank, compute running totals and compare a row with the previous one without collapsing your rows. Many analyst interview questions live here.
Analysts who last learn why a query is slow, how to clean text and dates inside SQL, how to change data without regret, and how to write tests that fail loudly when the data goes wrong. Finish with one project: a question about your store data, the queries that answer it, and a short write-up of what you found and what you could not prove.
These are the ones the course builds whole lessons around:
BETWEEN '2024-06-01' AND '2024-06-30' misses almost all of the last day when the column holds times. Use >= '2024-06-01' AND < '2024-07-01' instead.100.0 first when you want a percentage.The SQL & Data Analysis course is rated Beginner to Intermediate and planned as 8 weeks of study. It has 12 chapters of 4 lessons each, 48 lessons in all, and most lessons take about 12 to 15 minutes to read. Every example uses the same four tables: customers, products, orders and order items.
| Chapter | What you learn |
|---|---|
| 1. Thinking in Tables | Keys, SELECT, WHERE, ORDER BY, NULLs, text and dates, UNION and EXCEPT |
| 2. Aggregation: Rows Into Answers | COUNT, SUM, AVG, GROUP BY, HAVING vs WHERE, medians and percentiles |
| 3. Joining Tables | Every join type, fan-out, anti-joins, self-joins, range joins |
| 4. Subqueries, CTEs, and Window Functions | EXISTS, named steps, ranking, running totals, moving averages |
| 5. Clean Data, Fast Queries, Good Schemas | Cleaning in SQL, indexes, query plans, normalisation, constraints |
| 6. Changing Data Safely | INSERT, UPDATE, DELETE, transactions, UPSERT, live schema changes |
| 7. Putting a Query Into Production | Views, data tests, SQL inside applications, finding slow queries |
| 8. Questions That Do Not Fit a GROUP BY | ROLLUP, LATERAL joins, recursive CTEs, funnels and retention cohorts |
| 9. Why the Planner Chose That Plan | Statistics, join algorithms, partial and covering indexes, rewrites |
| 10. Concurrency Under Real Traffic | MVCC, isolation levels, locks, deadlocks, VACUUM |
| 11. Big Tables, Warehouses, and the Bill | Partitioning, column stores, scan cost, approximate counts |
| 12. The Numbers the Company Runs On | Star schemas, slowly changing dimensions, backfills, metric contracts |
Chapters 1 to 4 are the core of analyst SQL. Chapters 5 to 8 make you the person whose numbers others trust. Chapters 9 to 12 go further than most analyst roles need on day one, into the territory where analysis meets data engineering, so treat them as a second pass once the basics are solid.
It suits students, Excel users who keep hitting the limits of their spreadsheets, and junior analysts who can write a query but are not sure they trust the result. It assumes no programming background.
It is not a full analyst career path on its own. If you also want spreadsheets, statistics and presenting results, the Data Analyst Bootcamp covers that wider ground. If you are still more comfortable in spreadsheets, our guide on how to use Excel for data analysis and the Excel for Business Analytics course are a gentler start. Once your queries work, Power BI & Business Intelligence turns them into dashboards.
Each lesson ends with a short quiz you pass to open the next lesson, and each chapter ends with a chapter quiz. You need 70 percent on a chapter quiz to open the next chapter, so you cannot skim ahead without the foundations.
When every lesson is complete and every chapter quiz is passed, the final exam unlocks: 50 questions, 60 minutes, and a pass mark of 50 out of 100. You get one timed attempt. If you do not pass, a retake opens 30 days later. Pass it and your certificate is issued free.
Be honest with yourself about what that certificate proves. It shows you finished a demanding course and passed a timed exam. An employer will learn more from a short project where you explain a query, the check you ran on it, and the mistake you caught.
06 questions
The basics of SELECT, GROUP BY and joins take a few weeks of steady practice. Our course is planned as 8 weeks at about an hour a day. Becoming genuinely reliable, where you catch your own wrong numbers, comes from using it on real questions after that.
Yes. PostgreSQL and SQLite are free to install, and the SQL & Data Analysis course, its quizzes, the final exam and the certificate are all free on Cortexis.
PostgreSQL is a strong choice because it follows standard SQL closely and is free. The course uses it for most examples and notes where MySQL and SQLite differ, so the skills transfer.
Usually not on its own. Most analyst roles also expect spreadsheets, some statistics and the ability to explain a result clearly. SQL is the part that most often separates candidates, so it is worth learning well.
No. SQL is a separate, smaller language and many analysts learn it first. Python becomes useful later for statistics and automation, but it is not a prerequisite for this course.
Window functions calculate across related rows without collapsing them, for example a running total per customer or a rank within each category. Analysts use them constantly, and they come up often in interviews. Ready to start? Open the free [SQL & Data Analysis](/learn/sql-data-analysis) course, run your first SELECT on the store database, and check every answer against one row you can count by hand.