randarium
Datasets

Synthetic Dataset Generator

Create reproducible users, customers, products, orders, events, logs, tickets, inventory, web analytics, or sensor-reading datasets.

Also known as: mock dataset · test data

seeded · synthetic data

Presets

Output

No output yet — set your options and hit .
About this tool, tips & examples

What it does

The Fake Data Generator creates clearly synthetic, seeded datasets — up to 10,000 rows per run — and exports them as CSV, JSON, or NDJSON. Pick a template (users, customers, products, orders, events, logs, tickets, inventory, web analytics, or sensor readings), shape the schema by selecting, renaming, or adding fields, and optionally inject nulls and duplicate rows to make imports realistic. The same seed always regenerates the identical dataset.

Common use cases

  • CSV import testing — generate a fake CSV with exactly the columns your importer expects, including a controlled dose of nulls and duplicates.
  • Demos and dashboards — populate a prototype with plausible-looking users, orders, or events without touching real customer data.
  • Database seeding — reproducible seed data: same seed, same rows, every environment.
  • Load and pagination testing — thousands of rows with one click, exported in whatever format your pipeline consumes.

Settings

  • Template — the domain schema to start from: users, customers, products, orders, events, logs, tickets, inventory, analytics, or sensors.
  • Rows — 1 to 10,000 rows per run.
  • Fields — keep only the columns you list; rename them with old:new; add extra fields on top of the template.
  • Null rate / Duplicate rate — inject missing values and repeated rows at a chosen probability, so your import code meets imperfect data early.
  • Seed — identical seed + settings = identical dataset, byte for byte.

Privacy note

Every value is generated locally in your browser and never uploaded. The output is fictional: names, emails, orders, and readings are synthetic and must not be treated as real people, customers, or operational records — and equally, no real data of yours is ever involved.

FAQ

Is this real anonymized data? No — and that’s a feature. Nothing here derives from real people or records, so there is nothing to re-identify. It is synthetic data generated from templates.

Which export formats are supported? CSV (with an optional header row), JSON, and NDJSON via the shared output controls — copy to the clipboard or download as a file.

Can I make messy data on purpose? Yes. The null rate leaves fields empty at the probability you set, and the duplicate rate repeats rows — both essential for testing validation and deduplication logic. For deliberately malformed values, see the Messy Data and Invalid Data tools.

Can I reproduce a dataset exactly? Yes — the seed pins every row. Share the seed and settings (or a recipe link) and anyone can regenerate the same file.