How to Convert CSV to JSON

CSV (comma-separated values) is a plain-text table. The first row is often headers. Each following row is a record. Fields are separated by a delimiter (commonly comma, sometimes semicolon, tab, or pipe).

JSON (JavaScript Object Notation) is a structured data format. After conversion you almost always want an array of objects, one object per data row, with property names taken from the header row.

Minimal example

CSV:

name,age,city
Ada,36,London
Lin,29,Taipei

JSON (array of objects):

[
  { "name": "Ada", "age": "36", "city": "London" },
  { "name": "Lin", "age": "29", "city": "Taipei" }
]

Notice age is still a string unless your converter coerces types. Many tools keep everything as strings by default so “00123” zip codes do not become numbers and lose leading zeros.

After conversion, pretty-print and validate with the free JSON formatter validator and our JSON formatter guide.

With “first row is headers” enabled, each header cell becomes a property name on every object. Row 2’s cells map positionally to those names.

Header hygiene

• Unique names — duplicate headers overwrite or collide depending on the parser; rename before convert
• Stable names — spaces and punctuation are legal in JSON keys but annoying in code; prefer first_name or firstName
• No empty header — blank column titles become empty-string keys or auto-generated names; fix the sheet

No-header mode

If the file has no header row, converters often emit arrays of arrays, or invent keys like column1, column2. Array-of-arrays is faithful to the grid but painful for most apps. Prefer adding a header row once in the spreadsheet, then converting to objects.

Column order

JSON object key order is usually preserved by modern engines for string keys you insert in order, but your program should look up fields by name, not by “third key.” CSV order only matters for mapping cells to header names at parse time.

“CSV” is a family of dialects, not one strict law. Wrong delimiter is the #1 silent failure: you get one giant field per line instead of columns.

Delimiter	Common when
Comma ,	U.S./UK Excel exports, many APIs
Semicolon ;	Locales where comma is the decimal mark
Tab	TSV exports, some database dumps
Pipe |	Data that already contains many commas

Auto-detection

Good converters sample the first lines and guess. If your data has more semicolons than commas in the header line, semicolon is a strong guess. Always glance at column count in the preview: 1 column on a 10-field export means wrong delimiter.

Decimals and locales

European-style 1.234,56 numbers inside semicolon-delimited files are still text to a CSV parser. Type coercion rules vary. Keep them as strings unless you deliberately parse numbers with the right locale rules.

Paste into the free CSV to JSON converter, switch delimiters, and watch the live column count until it matches your sheet.

Most web and backend code wants a list of records: [{...}, {...}]. That shape maps cleanly to database rows, table UIs, and map/filter loops.

Other shapes you might see

• Array of arrays — no header semantics; good for pure grids
• Column-oriented object — { "name": ["Ada","Lin"], "age": ["36","29"] } for some analytics pipelines
• Nested JSON — only if a column already contains JSON text or you post-process paths like address.city

Type coercion choices

Optional parsers turn true/false/null/numbers into JSON types. That helps APIs but can damage:

• IDs with leading zeros
• Phone numbers and ZIP+4
• Large integers past Number.MAX_SAFE_INTEGER in JavaScript

When in doubt, keep strings and coerce in your application with explicit rules. After convert, run the free JSON formatter validator to confirm the document is parseable before you ship it.

Never parse production CSV with line.split(',') alone. Real files use RFC 4180-style quoting.

• Commas inside quotes — "Smith, Ada",36 is two fields, not three
• Embedded newlines — a quoted field can span lines; row ≠ one line always
• Escaped quotes — "She said ""hi""" → She said "hi"
• Empty fields — a,,c is three fields; middle is empty string
• Trailing delimiter — a,b, may mean a third empty field
• UTF-8 BOM — Excel sometimes prefixes \uFEFF; your first header becomes "\uFEFFname" if you do not strip it
• Mixed line endings — \r\n vs \n; parsers should accept both

Encoding

Save CSV as UTF-8 when non-ASCII characters matter. Latin-1 misread as UTF-8 produces mojibake in JSON strings. If you must transport binary alongside JSON, that is a different problem — see how Base64 works with the Base64 tool.

For query strings and special characters in URLs (not CSV itself), use the free URL encoder decoder and our URL encoding guide.

Export cleanly, paste or upload, set dialect options, preview rows, then copy or download JSON.

1. Open the free CSV to JSON converter.
2. Paste CSV text (or load the file if the tool supports upload).
3. Confirm whether the first row is headers.
4. Set delimiter: comma, semicolon, tab, or pipe to match the file.
5. Check live preview: row count, column count, and sample objects.
6. Copy the JSON or download the file for your app or API.
7. Validate formatting with a JSON formatter if you need indented output or error locations.

Pre-clean in the spreadsheet

Delete empty trailing columns, fix duplicate headers, and ensure one header row only. Five minutes in Sheets/Excel prevents twenty minutes of key renames in code.

Watch for these when the output “looks fine” but breaks later.

• Wrong delimiter — one property containing the entire line
• Headers treated as data — first object is {"0":"name","1":"age"} style noise or header text as values
• Unquoted commas — columns shift right for some rows only
• Assuming all rows have the same width — short rows need empty defaults for missing trailing cells
• Number coercion on IDs — leading zeros and big integers damaged
• Not validating JSON — trailing commas or bad escapes from a buggy converter

This guide is educational. For regulated data (PII, health, finance), convert locally and follow your organization’s data-handling rules — browser tools still run on a machine you must trust.