Click the headers you want and get just those columns back. The order you click in is the order you get, so this reorders as well as filters — which is usually what an import actually needs when its column order differs from your export. Output as CSV, as a plain list (the form you want when the goal is a column of email addresses to paste somewhere), or as JSON objects keyed by header. A unique filter removes repeated values, so pulling one column and asking for unique gives you the distinct values in it. Columns are picked after a real CSV parse, not by splitting lines on commas, which is what keeps quoted fields containing commas under the right header. Whole folders process at once; nothing is uploaded.
| Format | Use it for |
|---|---|
| CSV | Re-importing with fewer or reordered columns |
| Plain list | A column of values to paste somewhere |
| JSON | Feeding a script, keyed by header |
No. Parsing and extraction happen in your browser, which is the point when the column you are removing is the one with personal data in it.
Yes — the output follows the order you click in, not the order in the file. Click email first and name second and that is the order you get.
Because a quoted field can contain a comma. Splitting on it puts every later value under the wrong header for that row, and the result looks plausible enough that the mistake usually survives into whatever you import it into. A full parse avoids this.
The missing column comes out as an empty string rather than shifting the other values left. In a real export a short row almost always means trailing empty fields, not misaligned ones.
On the output rows after selection. So if you extracted one column, you get that column's distinct values; if you extracted two, rows are unique on the pair.
Columns are picked by index after a full RFC 4180 parse, not by splitting each line on the delimiter, which matters the moment any field contains a quoted comma — line-splitting puts the wrong values under the wrong header from that field onward, and the output looks plausible enough to go unnoticed. Selection order is preserved, so clicking email then name emits those two columns in that order; this reorders as well as filters, which is usually what a database import actually needs. Ragged rows are padded rather than skipped: a row with fewer cells than the header yields an empty string for the missing column instead of shifting values left, because in a real export a short row means a trailing empty field far more often than it means a misaligned one. The unique filter compares whole output rows after selection, not the original rows, so extracting one column and asking for unique values gives you the distinct values of that column. JSON output keys objects by the header row, and list output joins the selected cells with a space, which is the form people want when the goal is a column of email addresses to paste somewhere. Everything runs in your browser and no file is uploaded.