Free in-browser tool

Free Dataset Builder for Clean, Reusable Files

Build a clean dataset without guessing at field types. Name one row, add a few real examples, and download files you can open, share, or build on.

  • Catch missing values and duplicate identifiers before they spread
  • Download CSV, JSON, XLSX, and a plain-English README
  • Keep the dataset in your browser instead of uploading it
Build your dataset

Your working file

Build the structure, then test real rows

Start small. A few awkward examples will find problems while they are still easy to fix.

Saved in this browser
1

Name the dataset and one row

One row should represent one clear thing, such as a product, business, query, or event.

2

Define the fields

Give every column one job. Mark fields required when a useful record cannot exist without them.

Which type should I use?

Not sure? Keep text. It is the safest choice for IDs and codes. Use it for phone numbers, ZIP codes, and anything you will not calculate.

string ... Text
Use for words, IDs, codes, or numbers you will not calculate. Example: biz_001 or 02139.
integer ... Whole number
Use for counts that never need a decimal. Example: 12.
number ... Number with decimals
Use when values can include a decimal or need math. Example: 19.95.
boolean ... Yes or no
Use when each value has only two states. Example: true or false.
date ... Date
Use for a calendar date with no time. Example: 2026-10-15.
datetime ... Date and time
Use when the exact time matters. Example: 2026-10-15T14:30:00-04:00.
url ... Web address
Use for a complete http or https web address. Example: https://example.com/page.
3

Test a few real rows

Add records here or paste CSV. Empty rows are ignored when you check and export the dataset.

4

Fix the warnings, then take the files

The check covers field names, required values, data types, and identifiers marked unique.