What shape is the JSON?

Your choice. An array of objects uses the header row as keys, which is what most APIs expect; an array of arrays preserves the raw grid including the header row.

Which sheet does it convert?

The first one by default, and you can switch sheets before converting — each sheet becomes its own JSON document rather than being merged.

How are dates handled?

Excel stores a date as a number counting days from 1900, so a naive conversion produces values like 45678 instead of a date. Dates are read as the workbook stores them and written in a readable form instead.

Do I get formulas or their results?

The results. A workbook caches the last computed value of every formula, and that value is what ends up in the JSON — the formula text itself is not carried over.

What happens to merged cells?

Only the top-left cell of a merged range holds the value; the rest are genuinely empty in the file. Merged headers are the usual reason a converted sheet has blank-looking columns.

Will leading zeros survive?

Only if the column is stored as text in the workbook. If Excel stored a postcode or product code as a number, the zeros were already gone before the file reached this tool.

Is my workbook uploaded to a server?

No. The file is read and converted by your browser, on your device. This matters for spreadsheets in particular, which routinely hold payroll, customer lists or financial figures.

Excel to JSON Converter — XLSX & XLS to JSON

How it works

Select or drop an .xlsx or .xls workbook.
Pick the sheet and the JSON shape you want.
Download the JSON. The workbook is read on your device and never uploaded.

The two JSON shapes, and which one you want

Given a sheet like this:

        A          B         C
1    name       role      salary
2    Ada        engineer  90000
3    Grace      admiral   95000

An array of objects treats row 1 as keys. This is what almost every API, database import and JavaScript codebase expects:

[
  { "name": "Ada",   "role": "engineer", "salary": 90000 },
  { "name": "Grace", "role": "admiral",  "salary": 95000 }
]

An array of arrays keeps the raw grid, header row included. Use it when column order matters, when the sheet has no header row, or when duplicate column names would collide as object keys:

[
  ["name", "role", "salary"],
  ["Ada", "engineer", 90000],
  ["Grace", "admiral", 95000]
]

Excel dates are numbers, and that is the main trap

A spreadsheet does not store 2026-03-14. It stores 46095 — a count of days since 1 January 1900 — plus a display format telling Excel to render it as a date. A naive conversion reads the number and writes 46095 into your JSON, which then looks like plausible data and is not.

This tool reads the cell’s format alongside its value and writes dates in a readable form instead. Two related quirks worth knowing if you ever do this conversion elsewhere: the epoch is off by one because Excel deliberately reproduces a Lotus 1-2-3 bug that treats 1900 as a leap year, and workbooks saved by older Mac versions of Excel count from 1904 instead, so the same number means a date four years apart depending on the file.

What else does not survive the way you expect

Formulas become their results. A workbook caches the last computed value of every formula, and that cached value is what you get. If the file was written by a tool that never calculated, those cells can be empty.
Merged cells hold one value. Only the top-left cell of a merged range contains anything; the others are genuinely empty in the file. Merged title rows are the usual reason a converted sheet has a header of null values.
Leading zeros are already gone. If a postcode or product code was stored as a number, Excel dropped the zeros when the file was saved. No conversion can recover them — the information is not in the file.
Displayed values are not stored values. A cell showing 3.14 may hold 3.14159265. Rounding is a display format; the JSON gets the full number.
Nothing about appearance carries over. Colours, fonts, conditional formatting, column widths, comments and charts have no JSON representation and are dropped.

Header rows that are not header rows

Real workbooks often start with a title, a blank row and a logo before the actual table begins. Converting with the object shape then produces keys like Sales report Q3 and __EMPTY_1. Delete everything above the real header row before converting — the conversion trusts the first row absolutely, and cleaning up afterwards in JSON is far more work than deleting three rows in Excel.

The same applies to trailing content. A totals row at the bottom becomes a data record indistinguishable from the rest, and a footnote in column A becomes a row with one populated field.

Several sheets

Each sheet becomes its own JSON document rather than being merged, because sheets in one workbook rarely share a schema — merging them would produce records with mostly-empty fields. Convert the sheets you need one at a time and combine them afterwards if you actually want one array.

.xls as well as .xlsx

Both are read. .xlsx is the modern zipped XML format; .xls is the pre-2007 binary one, still produced by plenty of legacy exports and capped at 65,536 rows. If an .xls export looks truncated, that limit is the first thing to check — the data may have been lost before you received the file.

Frequently asked questions

What shape is the JSON?: Your choice. An array of objects uses the header row as keys, which is what most APIs expect; an array of arrays preserves the raw grid including the header row.
Which sheet does it convert?: The first one by default, and you can switch sheets before converting — each sheet becomes its own JSON document rather than being merged.
How are dates handled?: Excel stores a date as a number counting days from 1900, so a naive conversion produces values like 45678 instead of a date. Dates are read as the workbook stores them and written in a readable form instead.
Do I get formulas or their results?: The results. A workbook caches the last computed value of every formula, and that value is what ends up in the JSON — the formula text itself is not carried over.
What happens to merged cells?: Only the top-left cell of a merged range holds the value; the rest are genuinely empty in the file. Merged headers are the usual reason a converted sheet has blank-looking columns.
Will leading zeros survive?: Only if the column is stored as text in the workbook. If Excel stored a postcode or product code as a number, the zeros were already gone before the file reached this tool.
Is my workbook uploaded to a server?: No. The file is read and converted by your browser, on your device. This matters for spreadsheets in particular, which routinely hold payroll, customer lists or financial figures.