What ROW_NUMBER does
Number rows inside each group. SQL syntax can vary by database, but the pattern below is a useful starting point for reports and analysis.
Syntax or pattern
ROW_NUMBER() OVER (PARTITION BY customer_id ORDER BY order_date DESC)5 practical examples
Latest order per customer
Number orders from newest to oldest for each customer.
SELECT *
FROM (
SELECT o.*, ROW_NUMBER() OVER (PARTITION BY customer_id ORDER BY order_date DESC) AS rn
FROM orders o
) x
WHERE rn = 1;ROW_NUMBER is the standard latest-record pattern.
Remove duplicate emails
Keep one row per email address.
SELECT *
FROM (
SELECT c.*, ROW_NUMBER() OVER (PARTITION BY email ORDER BY updated_at DESC) AS rn
FROM customers c
) x
WHERE rn = 1;Choose the ordering rule before removing duplicates.
Top order per region
Find the largest order in each region.
SELECT *
FROM (
SELECT region, order_id, total_amount, ROW_NUMBER() OVER (PARTITION BY region ORDER BY total_amount DESC) AS rn
FROM orders
) x
WHERE rn = 1;Partitioning restarts the numbering per region.
Number rows for export
Add a sequential row number.
SELECT ROW_NUMBER() OVER (ORDER BY order_date, order_id) AS row_no, *
FROM orders;This creates a stable row number for a report.
Find second purchase
Return the second order for each customer.
SELECT *
FROM (
SELECT o.*, ROW_NUMBER() OVER (PARTITION BY customer_id ORDER BY order_date) AS rn
FROM orders o
) x
WHERE rn = 2;Changing the filter on rn lets you find first, second or latest events.
Common mistakes to avoid
- Forgetting that SQL dialects vary across PostgreSQL, SQL Server, MySQL, BigQuery and SQLite.
- Using SELECT * in production reports when only a few columns are needed.
- Not checking join keys, duplicate rows or NULL values before trusting results.
FAQ
Will this SQL work in every database?
The idea is portable, but function names and date syntax may vary. Check your database dialect if a function is not recognized.
Should I use this in a report query?
Yes, if the pattern matches the business question and you have checked filters, joins and row counts.
Why does my result have too many rows?
The most common reasons are duplicate join keys, missing filters or grouping at the wrong level of detail.