Skip to content
TextArray1.41.0
100% local

CSV profiler

Scan a CSV and get per-column type, missing values, unique counts and top values.

Input
Output
Related tools
CSV ⇄ JSON converterDeveloper tools
CSV column validatorDeveloper tools
Sum and number statisticsNumbers
i18n key diffDeveloper tools
TOML to JSON converterDeveloper tools
SQL to JSONDeveloper tools
CSV profiler
Paste CSV data and this tool scans it column by column, reporting what's actually in your data before you build a pipeline, write a schema or trust an import. For every column it estimates a type — integer, decimal, date, boolean or text — and shows the share of values that don't match it, so a column that looks numeric but hides a stray "N/A" or a typo stands out immediately.
Each column report also covers missing cells and their share of the total, how many unique values are present, and a suggestion for whether the column is safe to use as a key: it only qualifies when every row has a value and no two rows share one. Numeric columns get a minimum, maximum and average; text columns get the shortest and longest value length. The most frequent values are listed with their counts, so you can spot a column that's really a category in disguise, or one dominated by a single repeated value.
The delimiter is auto-detected from the data, or you can force comma, semicolon, tab or pipe. Turn off "First row is header" for data that starts straight away, and switch the output between a readable overview and a tab-separated table you can paste into a spreadsheet. Ragged CSV is tolerated: a short row leaves its missing columns marked, and a long row's extra fields are simply ignored rather than breaking the whole scan.
Everything runs locally in your browser — nothing you paste is uploaded anywhere, which matters for CSV exports that often carry real customer or business data. Copy the report, download it as a .txt file, or send it straight into another tool's input to keep working.
FAQ
How is the column type estimated?
Each non-empty value is checked against integer, decimal, boolean and date patterns; the type most values match becomes the column's estimated type, and the tool reports what share of values don't fit it.
What makes a column a "key candidate"?
A column qualifies only when it has no missing values and every value is unique across all rows — the same test a database would apply to a primary key.
What happens with rows that have the wrong number of columns?
A row with fewer fields than the header leaves its missing columns counted as empty; a row with extra fields simply has those extra values ignored, so a ragged file doesn't break the scan.
Can I profile a CSV without a header row?
Yes. Turn off "First row is header" and every row is treated as data; columns are then labelled Column 1, Column 2 and so on.
Is my CSV data uploaded anywhere?
No. The profiler runs entirely in your browser — your CSV never leaves your device, which matters when it contains real customer or business data.