Pick a place to practice
| Platform | Where to start | How to use it with CoreTrail |
|---|---|---|
| StrataScratch | Search by question name or ID | Try the familiar problems below, then explain the grain and edge cases without your old solution. |
| LeetCode | SQL 50 study plan | Work through its sequence, keeping a short note on the mistakes you repeat. |
| DataLemur | SQL question list | Choose a weak topic, attempt a problem, then revisit the related concept. |
Choose the SQL dialect before starting. Access, difficulty labels, and available editors can change; the original page is the source of truth. The level labels below are practice guidance. CoreTrail’s linked examples explain related techniques and may use different fixtures or assumptions from the platform question.
A starting list
Try the question before opening the related lesson. Each problem link opens the original platform in the same tab, so the browser’s Back button brings you here.
Find all posts which were reacted to with a heart ↗
Return a post once even when it has several matching reactions.
Revisit the concept →Largest Olympics ↗
Count the right entity and handle tied maxima.
Revisit the concept →Apple Product Counts ↗
Reduce device events to user-level flags.
Revisit the concept →Spam Posts ↗
Distinguish a post denominator from a view denominator.
Revisit the concept →Confirmation Rate ↗
Keep users with no activity and define the rate.
Revisit the concept →Customers Who Bought All Products ↗
Require every product without counting repeated purchases twice.
Revisit the concept →Restaurant Growth ↗
Aggregate to day grain before calculating a rolling metric.
Revisit the concept →Department Top Three Salaries ↗
Compare distinct salary levels and preserve ties.
Revisit the concept →Page With No Likes ↗
Find absent relationships and check null behavior.
Revisit the concept →Histogram of Tweets ↗
Count per user, then count users per bucket.
Revisit the concept →Repeat Purchases on Multiple Days ↗
Separate repeated events from activity on different dates.
Revisit the concept →Top Three Salaries ↗
Rank within departments and apply the stated tie rule.
Revisit the concept →Not sure where to start?
Start with Page With No Likes and Histogram of Tweets. Then try Largest Olympics, Confirmation Rate, and Restaurant Growth. Finish with a ranking problem and Customers Who Bought All Products. This sequence moves from identifying a population to composing several stages.
Once those feel familiar, use the filters to mix topics rather than solving ten nearly identical questions in a row. The mixed drills hide the technique and give progressive hints.
Keep a useful mistake log
After an attempt, record the mistake in one sentence: “I counted events instead of users,” “I removed unmatched rows,” or “I compared the previous observed month.” Add the smallest input that exposes it. Revisit that input a few days later without your original query.
A passing submission checks the platform’s cases. It does not prove that your query is efficient at scale. For that part, use the query-plan labs and compare execution evidence on larger data.