How to Convert CSV to JSON (Free, In Your Browser, No Code)

You've got a CSV file. You need JSON. Maybe it's for a REST API, a configuration file, a NoSQL database, or a front-end app that reads JSON directly. Here's how to do the conversion correctly — including the parts that most converters get wrong.

What CSV to JSON conversion actually does

A CSV file is fundamentally a flat table: rows and columns, every row has the same fields. JSON is a hierarchical key-value format. Converting between them means turning each row into a JSON object, using the header row as the keys.

Given this CSV:

name,age,city
    Alice,32,New York
    Bob,28,London

The natural JSON output is an array of objects:

[
    { "name": "Alice", "age": 32, "city": "New York" },
    { "name": "Bob",   "age": 28, "city": "London"   }
    ]

Simple enough. But there are decisions to make: Should numeric values stay as strings or become actual JSON numbers? Should you nest by a key column? How do you handle empty cells? These choices depend on what you're doing with the JSON.

Converting in your browser with CSV Splitter Online

CSV Splitter Online has a CSV ↔ JSON tab that handles conversion without uploading your file to any server.

The type coercion question

This is the decision that trips up most CSV-to-JSON conversions. CSV stores everything as text — there's no distinction between the string "42" and the number 42. JSON has explicit number and boolean types. When converting, you need to decide which columns should become which type.

The safe default is to keep everything as strings — it's lossless and can't produce incorrect type conversions. The problem is that downstream code consuming your JSON might expect age: 28 (a number) and break when it gets age: "28" (a string), especially in JavaScript where "28" + 1 === "281".

CSV Splitter Online's auto-number-coercion converts any value that parseFloat() correctly identifies as a number. This handles integers, decimals, and negative numbers. It doesn't touch values that look numeric-ish but aren't — like phone numbers, ZIP codes, or product SKUs that start with digits. Those stay as strings.

Nested JSON: when you need a lookup table

Sometimes you need JSON that's keyed by one of your columns rather than a flat array. For example, a product catalog where you want to look up products by SKU:

// Array of objects (default)
    [
    { "sku": "A100", "name": "Widget", "price": 9.99 },
    { "sku": "B200", "name": "Gadget", "price": 24.99 }
    ]

    // Nested by key (key column: "sku")
    {
    "A100": { "name": "Widget", "price": 9.99 },
    "B200": { "name": "Gadget", "price": 24.99 }
    }

The second structure lets you do O(1) lookups in JavaScript: catalog["A100"].price instead of searching through an array. Use this when the JSON will be used as a configuration file or lookup table, and the key column has unique values.

Handling empty cells

Empty cells in CSV are ambiguous: they could mean "no value", "null", "zero", or "not applicable" depending on context. In JSON you have options:

• Emit empty string: "email": "" — cleaner, always present
• Emit null: "email": null — semantically correct, but some systems handle null differently from missing keys
• Omit the key entirely: { "name": "Alice" } with no email key — smallest output, but consuming code needs to check for undefined

CSV Splitter Online outputs empty cells as empty strings by default. If you need nulls, a quick post-process in JavaScript: obj[key] = obj[key] === "" ? null : obj[key].

Converting CSV to JSON with Python

For scripted or bulk conversions:

import pandas as pd
    import json

    df = pd.read_csv('data.csv')

    # Array of objects — most common
    records = df.to_dict(orient='records')
    with open('output.json', 'w') as f:
    json.dump(records, f, indent=2)

    # Nested by key (e.g. by 'sku' column)
    nested = {row['sku']: {k: v for k, v in row.items() if k != 'sku'}
    for row in records}
    with open('output_nested.json', 'w') as f:
    json.dump(nested, f, indent=2)

The built-in Python approach without pandas:

import csv
    import json

    with open('data.csv', newline='', encoding='utf-8') as f:
    reader = csv.DictReader(f)
    records = list(reader)

    with open('output.json', 'w', encoding='utf-8') as f:
    json.dump(records, f, indent=2, ensure_ascii=False)

csv.DictReader uses the header row as keys automatically. ensure_ascii=False keeps special characters (accented letters, etc.) readable in the output file rather than escaping them to \u00e9.

Converting JSON back to CSV

CSV Splitter Online also handles the reverse: paste JSON into the JSON tab and get a CSV out. This works for arrays of flat objects — nested JSON needs to be flattened first.

In Python: pd.DataFrame(records).to_csv('output.csv', index=False) — if your JSON is an array of objects with consistent keys, pandas flattens it automatically.

Common problems and fixes