CSV is a table serialization, not just text split at every comma. Quoted cells may contain commas and newlines. Parse it before filtering, deduplicating or summarizing.
Check headers and table shape
Use one row of unique nonempty column names and the same number of cells in each row. The tools reject malformed tables instead of guessing how fields should shift. Preserve the original export so each transformation can be reviewed or reversed.
Choose a comparison convention
Filtering and duplicate removal are case sensitive. Text sorting compares characters, so 10 comes before 2; numeric sorting validates numeric cells. The column summary skips blanks and reports them, but rejects nonblank nonnumeric content rather than quietly converting labels to zero.
Inspect the destination import
Downloads escape prefixes a spreadsheet could interpret as formulas. This may add an apostrophe to text cells, including strings starting with a minus sign. CSV does not preserve spreadsheet types or formatting, and identifiers may lose leading zeros if imported as numbers. Check types and locale settings in the receiving application.
Worked example
Scroll the table horizontally on small screens to see every column.
| Input | Ascending text | Ascending numeric |
|---|---|---|
| 2, 10, 3 | 10, 2, 3 | 2, 3, 10 |
An identifier with leading zeros may need to remain text even when every character is a digit.
Sources and checks
Examples are illustrative and can be checked with the linked tools. Read each tool’s assumptions before applying its result. How we verify methods.