How to convert Google Sheets to JSON without Apps Script

No. File > Download in Google Sheets offers Microsoft Excel, OpenDocument, PDF, Web page, CSV and TSV, with no JSON option, and the CSV download covers only the tab you have open.

So you either convert that CSV into JSON objects or read the cells through Google's APIs and build the keys yourself. The 4 routes compare like this:

A sheet shared as Anyone with the link has a CSV export URL that Google does not officially document: https://docs.google.com/spreadsheets/d/<id>/export?format=csv&gid=<gid>. Put it after https://rows.page/ and rows.page opens it as a table.

1. In Google Sheets, set sharing to Anyone with the link.
2. From the address bar, copy the <id> (the part after /d/) and the <gid> of the tab (the number after #gid=).
3. Open https://rows.page/https://docs.google.com/spreadsheets/d/<id>/export?format=csv&gid=<gid>.
4. Click Export and pick JSON for one array of objects, or JSONL for one object per line.

Google's export link allows cross-origin requests, so your browser downloads the CSV straight from Google instead of through the rows.page proxy, and nothing is stored on rows.page. The link has no file extension, so rows.page reads the format from the bytes and opens it as CSV.

Before exporting, trim the data in the grid. Type a filter such as price:>20 or -name:* (rows where name is empty), or run SQL that starts with SELECT, WITH or FROM against the table data. Export writes the current view: every matching row of a filter, or up to 1,000,000 rows of a SQL result.

rows.page fetches links without your Google login, so a sheet that is not shared publicly will not load from its export link. Use File > Download > Comma Separated Values (.csv) instead, drop the file on rows.page and click Export > JSON.

The file is read by DuckDB-WASM inside your browser tab, up to 1.5 GB, and is not uploaded. DuckDB detects the delimiter and the header row for you.

For a quick slice, copy the cells in Google Sheets, click Paste data on rows.page, paste them into the box and click Open. Copied cells arrive tab-separated and open as a table you can export as JSON.

For an app that reads a sheet in code, Google's official route is the Sheets API v4 spreadsheets.values.get. It returns a ValueRange object with range, majorDimension and values, where values is an array of row arrays rather than objects keyed by header. Map values[0] onto the other rows yourself:

// Example ValueRange response from spreadsheets.values.get
const response = {
  range: "Sheet1!A1:C3",
  majorDimension: "ROWS",
  values: [
    ["id", "name", "price"],
    ["101", "Widget", "25"],
    ["102", "Gadget", "40"]
  ]
};

const [headers, ...rows] = response.values;
const jsonObjects = rows.map(row =>
  Object.fromEntries(headers.map((key, i) => [key, row[i] ?? null]))
);

The API keeps your app connected to the live sheet, but you handle authentication, quotas and types yourself. Values come back as formatted strings by default, and empty trailing cells are left out of each row, which is why the mapper falls back to null.

Apps Script does the conversion inside Google: click Extensions > Apps Script, read the sheet's values into a 2D array, build objects from the header row and call JSON.stringify(). Deploy the script as a web app if you want a URL that returns the JSON.

It suits custom logic tied to one spreadsheet. The cost is a script to write and maintain for each sheet, while the export link in rows.page needs no code.

If a person needs the latest JSON, bookmark the https://rows.page/https://docs.google.com/... link. Google marks the export as no-store, so every visit downloads the current sheet.

If a script needs a stable URL, save the JSON and push it from a terminal:

# First push: prints the dataset link, an owner link and the command for the next version
curl -T data.json https://rows.page

# Push a new version (v2, v3) to the same link with the rpt_ owner token
curl -T data.json -H "Authorization: Bearer rpt_..." https://rows.page/<id>

Each hosted dataset serves https://rows.page/<id>/raw.json (latest version, no-cache) and pinned versions such as https://rows.page/<id>/v1/raw.json, both with access-control-allow-origin: *. Without an account a link lasts 7 days and holds up to 20 versions; sign in with Google to keep it, and the Free plan keeps 10 datasets.