RandomKit

Fake data generator

Build a table schema, pick a locale, and get realistic mock data for tests, seeds and demos. Export JSON, CSV, NDJSON or SQL inserts. Rows are generated in your browser.

Start from a preset

0 rows × 6 columns

Generating…

Good test data, and why it matters

Hand-typed fixtures like “test test, test@test.com” miss the bugs real data causes: apostrophes in surnames (O’Brien), accented characters, long street names, postcodes with letters, phone numbers with extensions. Switching the locale to German, French or Japanese is a fast way to check that your forms, database collation and CSV exports cope with non-ASCII text.

Emails always use the reserved example.com, example.org and example.net domains, which IANA keeps out of use under RFC 2606. A staging job that sends mail to them can’t reach a real person. Card numbers pass the Luhn check but aren’t real accounts. For payment-gateway testing, see our test credit card numbers page.

Export notes

  • CSV quotes any cell containing commas, quotes or line breaks (RFC 4180), so it imports cleanly into Excel, Google Sheets and COPY … FROM.
  • SQL emits standard INSERT INTO … VALUES statements in batches of 500 rows, with double-quoted identifiers and escaped single quotes. It works in PostgreSQL and SQLite; MySQL needs ANSI_QUOTES mode.
  • NDJSON writes one object per line for streaming loaders such as mongoimport, BigQuery and Elasticsearch’s bulk API.

Your column layout is saved in this browser so it’s there next time. Generated rows aren’t saved.

Questions people ask

What is a fake data generator used for?

Developers and testers use fake (mock) data to fill databases, seed demo environments, write fixtures for unit and end-to-end tests, and try out forms and reports without handling real personal data, which also keeps you on the right side of GDPR data-minimisation rules.

Is the generated test data realistic?

Names, addresses, phone numbers and companies come from the open-source Faker library’s locale data, so they follow real formats for each country. Emails use reserved example domains (RFC 2606), so a test email can never reach a real inbox.

Which export formats are supported?

JSON (an array of objects), NDJSON (one object per line, handy for streaming imports and BigQuery), CSV with proper quoting, and SQL INSERT statements batched 500 rows at a time with quoted identifiers.

How many rows can I generate?

Up to 10,000 rows per run. Generation happens in your browser, so larger sets depend on your device; 10,000 rows with ten columns usually takes well under a second.

What does “describe it with AI” send?

Only the short description you type (for example “SaaS customers with plan and signup date”). The AI suggests column names and types; the rows themselves are still generated locally. If the AI is unavailable, build the schema by hand with the column editor.

Related generators