Open a CSV the wrong way and the text turns to mojibake — Excel guesses the encoding and guesses wrong. This tool detects the encoding, shows you a preview, and lets you confirm or override it before converting to Excel; on the way back it writes UTF-8 (with a BOM by default, so Excel opens it clean). Every cell stays text, so 007 never becomes 7. Runs entirely in your browser — the file is never uploaded.
This conversion exists so a plain-text CSV and a spreadsheet each stay readable — without the encoding and type surprises in between.
A plain-text table: rows of values separated by a comma — or a semicolon (common in European locales) or a tab. It carries no encoding label, so whoever opens it has to guess: a file saved as Shift-JIS, GB18030 or EUC-KR, or as UTF-8 without a BOM, is exactly what shows up garbled in Excel.
The grid translators, reviewers and analysts are comfortable in — filter, sort, bulk-edit. The catch: double-click a .csv and Excel picks the encoding and the delimiter for you, mangles anything non-UTF-8, and re-types values that look numeric — a leading-zero 007 becomes 7, a long ID turns into scientific notation.
What this tool guards is that hand-off: it detects the encoding and delimiter, shows you a preview, and converts only once you've confirmed — so nothing is silently mis-decoded. Going to Excel, every cell is written as text (no type inference, so 007 and long IDs survive). Coming back to CSV it emits UTF-8, with a BOM by default so Excel re-opens it without mojibake — or without one for git and programmatic pipelines. It's a generic table converter: no localization structure is assumed, and real newlines inside cells stay real.