Mock Data Generator

Realistic test data in seconds

Model the schema, not just the rows

The generator starts where your real data starts: a schema. Add a field per column, name it exactly as your table or interface names it, and pick a type — names, emails, UUIDs, dates, numbers, booleans, addresses, companies, lorem text and more. Because the field names carry through to the output keys, CSV headers and SQL columns, the generated rows drop into your code without a rename pass. Set the row count, choose JSON, CSV or SQL, and generate; the output panel gives you copy and download in one click.

Why fake data beats production copies

Seeding development and demo environments with copies of production data is the classic shortcut that turns into a privacy incident: real emails end up in test mailers, real names in screenshots, real records on contractor laptops. Generated data sidesteps the entire problem — it looks convincing in a UI, satisfies foreign-shaped fields like emails and UUIDs, and contains no person. It also demos better: screens filled with plausible names and companies read as finished product, where "test test 123" reads as a prototype.

Make your test data earn its keep

Uniform pretty rows hide bugs; variety finds them. Include the shapes that break layouts and logic: long names that test truncation, addresses with apartment lines, dates near month and year boundaries, zero and negative numbers where they're legal. Generate a small batch first and read it — then scale up to the row count you need (up to 1,000 per run, and nothing stops you from generating several batches). For SQL seeds, set the table name in the bar and the output becomes INSERT INTO statements ready for a migration file.

Step-by-step: from empty schema to seed file

The page opens with a starter schema (id, name, email, age, city, signup date, active flag) so there is something to generate immediately. To model your own table, work down the schema panel: click + Add field for each column, type the exact column name in the left box, and pick a type from the dropdown on the right — id for a 1-based auto-increment, uuid, firstName / lastName / fullName, email, username, phone, city, country, streetAddress, company, jobTitle, date, datetime, boolean, number, age, price, decimal, color, url, word, sentence or paragraph. Remove a field with the × button. Then set Rows (1–1,000), click JSON, CSV or SQL in the format switch — the Table name box only appears for SQL — and press Generate. The output header shows the row count and format; Copy puts the text on the clipboard and Download saves it with a matching .json, .csv or .sql extension.

Real-world use cases

• Seeding a local Postgres or MySQL database. Mirror the users table column-for-column, set the table name, export SQL and paste the statements into a migration or psql session. Because the id type emits consecutive integers, foreign keys in a second table can be generated as number and mapped in range.
• API mocks for front-end work. Generate 50 JSON rows with uuid, fullName, email and datetime, save the file next to your components and point a mock server or MSW handler at it — the UI gets realistic list views before the backend exists.
• Spreadsheet and import testing. CSV with price, decimal, date and country columns quickly reveals how an importer handles decimal separators, quoted commas in addresses and date formats.
• Screenshots and demos. A product screenshot with thirty plausible names, companies and cities looks finished; the same screen with "asdf" does not, and it never exposes a real customer.

How the three exports are built under the hood

All three formats come from the same array of row objects, so a schema tweak changes every export identically. JSON is pretty-printed with two-space indentation — readable in a code review, and trivially minified later if size matters. CSV writes a header row from the field names, then quotes any value containing a comma, double quote or newline and doubles embedded quotes, which is the RFC 4180 convention most importers expect; booleans and numbers are written bare. SQL sanitizes the table name to word characters and underscores, emits one INSERT INTO table (cols) VALUES (...) statement per row, writes booleans as TRUE/FALSE, leaves numeric types unquoted and wraps everything else in single quotes with embedded quotes escaped as ''. If your database expects 1/0 for booleans or a different date literal, a single search-and-replace on the downloaded file is the easiest fix.

Further reading

For the format details behind the exports, see our guides on CSV, JSON and XLSX conversion pitfalls and when to minify or beautify code and data files, or browse all guides.

Frequently asked questions