Dataset license
SQLShed's 9 practice datasets are dedicated to the public domain under CC0 1.0 Universal. Use them for anything, including commercially. No permission, no attribution, no notice to keep.
What you may do with them
- Copy, modify, and redistribute the files, in whole or in part.
- Use them commercially: inside a paid course, a book, a product, a client deliverable, or a benchmark.
- Republish them under a different name, or fold them into a larger dataset of your own.
- Skip the credit entirely. Attribution is welcome and never required.
What it covers
Every file published for the 9 datasets in the
practice dataset catalog,
47 tables and
1,397,461 rows in total. That means
the table data in whichever formats a dataset ships (CSV, Parquet, and
the browser samples), the single-file loaders generated from it
(.sql dumps and
.schema.sql scripts), and
the manifests that describe them.
- Garden Shop · 6 tables, 130 rows
- Movie Rentals · 3 tables, 48 rows
- Website Analytics · 5 tables, 161 rows
- Employee Directory · 4 tables, 57 rows
- Ecommerce Orders · 7 tables, 426,024 rows
- Banking Transactions · 5 tables, 343,840 rows
- SaaS Subscriptions · 6 tables, 259,208 rows
- Marketing Campaigns · 5 tables, 303,609 rows
- Messy Student Records · 6 tables, 64,384 rows
All of it is invented. Every customer, employee, order, payment, movie title, session, and campaign in these files was written for teaching SQL. Nothing is derived from, scraped from, or sampled out of a third-party database, and nothing describes a real person, company, or transaction. That is what makes a public-domain dedication safe to make here: there are no upstream rights to pass on, and no personal data to protect.
What it does not cover
Datasets that are not ours. The free SQL datasets and sample databases listed elsewhere (Chinook, Sakila, Northwind, AdventureWorks, the open data portals, and the rest) are other people's data, hosted on other people's servers. SQLShed links to them and licenses nothing about them. Each has its own terms, and several are not public domain. Check the source before you republish anything from those.
The site's writing. Lessons, reference pages, cheatsheets, mistake and error write-ups, and the dataset guide prose are content, not data, and are not covered by this dedication.
The site's source code. This dedication applies to data files only. It is not a software license, and it does not change the licensing of anything in the SQLShed codebase.
The SQLShed name and logo. CC0 does not waive trademark rights, and says so in section 4(a). Using a dataset does not imply that SQLShed endorses what you built with it.
Why CC0 rather than a permissive license
The intent has always been that these datasets are free to use. The licenses people usually reach for do not quite say that. CC BY would require attribution on every reuse, including inside a product where a credit line has nowhere to go. Share-alike licenses like ODbL would make anything built on a dataset inherit these terms. MIT and Apache are software licenses: they require keeping a notice, and they say nothing about the database right that actually attaches to a published collection of rows in the EU.
CC0 1.0 Universal waives all of that, including the database right, and asks for nothing back. For invented data whose only purpose is to be practiced on, that is the honest position. Where a public-domain waiver is not legally possible, CC0 falls back to an unconditional license with the same effect.
The full legal code is at https://creativecommons.org/publicdomain/zero/1.0/.