One column, entered by different people or exported from different tools, ends up with 3/4/2026, 04-Mar-26, and 2026/3/4 all in the same file. Drop the CSV, tell it which convention the ambiguous numeric ones use, pick one output format, and get a column every tool will read the same way. Nothing leaves your browser.
plainkit found columns that look like dates. For each one, say how to read a plain numeric date like 3/4/2026 — that's the only part it can't tell from the text alone — then choose one output format for everything.
A CSV has no date type — every value is just text, so nothing stops one row from being typed as 4-Mar-2026 and the next as 03/04/2026, whether because two people entered it, or two tools exported it differently. Sorting, filtering, and importing into another system all need the whole column in one shape.
Dates with a month name (4 March 2026, Mar-04-26) are unambiguous — the tool reads them correctly on its own. Plain numeric dates like 3/4/2026 are not: written by a US convention that's the 4th of March; written day-first, as most of the world including Canadian government forms writes it, that's the 3rd of April. There is no way to tell from the text alone, which is why you're asked once per column, not asked to guess.
A 2-digit year (26) is read as 2000–2069 for 00–69 and 1900–1999 for 70–99 — right for anything recent, worth a second look for older records.