Convert Excel to JSON

An .xlsx file is not an opaque binary — it is a ZIP archive of XML parts in the SpreadsheetML format. When you drop one here, the converter does in JavaScript what a spreadsheet application does natively: it walks the ZIP central directory, inflates the parts it needs — the worksheet and the sharedStrings table — with the built-in DecompressionStream API, and scans the worksheet XML cell by cell. Everything happens on your device — no server ever sees the file, and the Network tab in DevTools shows no upload while it runs.

Real workbooks store text in two ways, and both are handled. Excel normally pools every distinct string in a sharedStrings part, with each cell holding an index into that pool; other producers write inline strings directly inside the cell element. Stored and deflate-compressed ZIP entries are both supported, so files written by Excel, Google Sheets, LibreOffice, and most export libraries read correctly. Conversion targets the first worksheet — sheet1, or the first worksheet part found if the workbook names its parts differently.

The practical consequence is privacy: this is safe for spreadsheets you would not feed to a random upload form — salary data, customer exports, financial models. It is also why conversion is instant; there is no upload round trip, just local CPU.

The contract is the same one the CSV converter follows: the first row of the worksheet is the header, each header cell names a column, and those names become the keys of every object in the output. Every following row becomes one object, so a header plus fifty data rows produces an array of fifty objects. Column count comes from the widest row, so a stray value to the right of the header still gets a column instead of being dropped.

Headers are cleaned up deterministically. A blank header cell is named by its position — a blank fourth column becomes column_4 — and surrounding whitespace is trimmed. Duplicates cannot survive in JSON, because an object cannot hold the same key twice, so the second email column becomes email_2 and a third would become email_3. Both columns keep their data; nothing is silently merged.

Empty cells and short rows become null, and every object carries the complete key set, so downstream code can rely on a key existing even when the cell was blank. If you control the sheet, name columns the way you want the JSON keys to read — lowercase, snake_case, no spaces — since renaming a header once in the spreadsheet beats renaming a key on every object afterwards.

Cells in an .xlsx carry a type marker, and the converter honors it: numeric cells become JSON numbers, boolean cells become true and false, and text becomes strings. Formula cells are read as their cached result — the file stores the last value the spreadsheet computed, so a SUM cell arrives as 4200, not as formula text.

Dates deserve an honest explanation. Excel has no date type in storage: a date is a number counting days since December 30, 1899 — an epoch inherited, leap-year quirk and all, from Lotus 1-2-3 — with the time of day as the fractional part. A cell that displays 2023-07-16 actually stores 45123, and this converter reports the stored value rather than guessing which locale-dependent display format to apply. Date columns arriving as serial numbers is by design, not an accident.

There are two clean workarounds. In the spreadsheet, add a helper column like =TEXT(A2,"yyyy-mm-dd") — or format the column as Text before entering the dates — so the file stores real strings. In code, convert a serial with new Date(Date.UTC(1899, 11, 30) + serial * 86400000), which turns 45123 back into July 16, 2023.

.xls, the Excel 97–2003 format, is BIFF — a stream of binary records inside a compound document. It shares nothing with the ZIP-of-XML structure of .xlsx, and reading it would mean shipping a second, much heavier parser, so this tool does not attempt it. Dropping an .xls produces a clear error instead of garbage output.

The fix is a single re-save. In Excel: File → Save As, then pick Excel Workbook (*.xlsx). LibreOffice Calc offers the same filter in its Save As dialog. In Google Sheets: File → Download → Microsoft Excel (.xlsx). In Apple Numbers: File → Export To → Excel. The re-saved file drops straight in.

It also helps to know what does not carry over from any workbook: only the first worksheet is converted, and only values — styles, colors, column widths, and merges are ignored, so a merged range contributes its value at the top-left cell and null elsewhere. If your source is actually CSV, the CSV to JSON converter is the shorter path. Once the JSON is out, the editor takes over: format it, validate it, walk it as a tree or mind map, or hand it to the AI copilot to reshape.