JSON Alternatives: When to Use What

JSON is the right default for most situations. But there are legitimate cases where another format is a better fit. Understanding the alternatives — what they add, what they cost — helps you make an informed choice rather than a cargo-cult one.

JSON5 extends JSON with several features that JavaScript developers often miss:

• Trailing commas in objects and arrays
• Single-quoted strings
• Unquoted object keys
• Line and block comments
• Multi-line strings

{
  // Server configuration
  host: 'localhost',
  port: 3000,
  allowedOrigins: [
    'https://example.com',
    'https://app.example.com',  // trailing comma OK
  ],
}

Use when: You control both the writer and the reader, and want the ergonomics without the YAML complexity. Not compatible with JSON.parse() — requires the json5 library.

JSONC is essentially JSON plus // and /* */ comments. It is not a formal standard but is widely recognized — VS Code uses it for settings.json and tsconfig.json.

Use when: Configuration files that developers edit by hand and want to annotate. Supported by VS Code, TypeScript's compiler, and the jsonc-parser library.

YAML is a superset of JSON (any valid JSON is valid YAML 1.2) with significantly more human-friendly syntax for config files:

• Comments with #
• No quotes required for simple strings
• Indentation-based structure instead of braces
• Multi-line strings without escaping

# Server configuration
host: localhost
port: 3000
allowedOrigins:
  - https://example.com
  - https://app.example.com

Use when: Configuration that humans write and read frequently — Kubernetes manifests, GitHub Actions, Docker Compose, Ansible. Avoid for API payloads — YAML's whitespace sensitivity makes it error-prone to generate programmatically.

TOML is designed specifically for config files. It is less ambiguous than YAML and more readable than JSON for nested structures:

[database]
host = "localhost"
port = 5432
name = "mydb"

[features]
darkMode = true
notifications = false

Use when: Application configuration files, especially in Rust (Cargo.toml) and Python (pyproject.toml) ecosystems.

MessagePack is a binary serialization format that encodes the same types as JSON but in a compact binary representation — typically 20–50% smaller than JSON with faster parsing.

Use when: High-volume inter-service communication where bandwidth or latency matters and human readability is not needed. Common in game servers, IoT, and real-time systems.

Google's Protobuf requires you to define a .proto schema file. It generates strongly-typed code for your language, producing small binary messages.

Use when: Internal microservice communication where you need strong typing, versioning, and maximum throughput. The schema requirement is a feature — it catches mismatches at compile time.

An IETF standard (RFC 8949) similar to MessagePack. Designed to be decodable without a schema, like JSON, but in binary.

Use when: IoT and constrained devices where you need binary efficiency but cannot use Protobuf's schema requirement.

If you have data in one of these formats and need it as JSON, the JSON Converter handles the most common conversions. For a detailed side-by-side comparison of JSON with XML and CSV specifically, see JSON vs XML vs CSV.