What Is JSON?

JSON objects wrap key-value pairs in curly braces. Keys must be double-quoted strings; values can be strings, numbers, booleans (true or false), null, arrays, or nested objects. Arrays list values in square brackets with comma separators. Strings use double quotes exclusively—single quotes are invalid.

JSON does not allow comments, trailing commas, undefined values, or unquoted object keys. Functions, dates, and undefined from JavaScript do not serialize to JSON without explicit conversion. Strict parsers reject any deviation, which is why pasted JavaScript object literals often fail validation until you remove trailing commas and swap quotes. Validators that pretty-print also minify—both operations require parseable input.

Numbers include integers and floats; scientific notation is allowed. Strings support Unicode and standard escape sequences (\n, \t, \", \\). Boolean true and false are lowercase literals. null represents an intentional absence of value, distinct from an empty string or zero.

There is no native date type—APIs typically use ISO 8601 strings like "2026-05-23T14:30:00Z". Binary data appears as Base64-encoded strings inside JSON rather than raw bytes. When exporting to CSV, remember that JSON types do not map perfectly: booleans become TRUE/FALSE text and null becomes an empty cell unless your converter specifies otherwise.

An array of uniform objects—[{ "id": 1, "name": "Ada" }, { "id": 2, "name": "Ben" }]—maps cleanly to spreadsheet rows where each object becomes one row and each key becomes a column. This is the easiest JSON shape for CSV conversion and Excel import.

A single nested document—{ "company": { "name": "Acme", "offices": [...] } }—requires flattening before most tabular tools can consume it. Pagination wrappers like { "data": [...], "meta": { "page": 1 } } are common in REST APIs; extract the inner array before converting. Null fields, empty arrays, and missing keys create sparse CSV output if not handled explicitly.

REST APIs send JSON in HTTP bodies with Content-Type: application/json. Clients POST JSON payloads to create resources and GET JSON responses to read them. GraphQL returns JSON with a data envelope and optional errors array. Webhooks deliver event JSON to your endpoint; validating structure before processing prevents silent data corruption.

Typical workflow: fetch an API endpoint, receive a JSON array of records, pretty-print or validate with a formatter, optionally flatten nested fields, then convert to CSV for finance or operations teams in Excel. Config files (package.json, tsconfig.json, CI pipelines) use the same syntax, so the skills transfer between runtime API data and developer tooling. Log sample payloads in staging—not production—to debug schema drift without exposing customer PII.

Minified JSON removes whitespace for wire transfer; formatted (pretty-printed) JSON adds indentation for human review. Both are semantically identical if the underlying structure matches. Online formatters detect syntax errors with line-level feedback, which saves time when debugging large webhook payloads.

Schema validation (JSON Schema, OpenAPI models, or language-native types) enforces required fields and data types at build time or runtime. Even without formal schemas, checking for expected top-level keys—data, results, items—before conversion prevents empty CSV exports when an API changes its envelope structure.

Never eval() JSON or treat parsed JSON as executable code. JSON is data, not a scripting language—though historically some libraries confused the two. When pasting JSON from email or tickets into online tools, remember it may contain personal data subject to privacy policies; client-side formatters that do not upload payloads reduce exposure compared to cloud validators that store inputs.

Large JSON files can exhaust browser memory when pretty-printed. Stream or chunk multi-gigabyte exports on the server side rather than in a tab. For public APIs, rate-limit and authenticate endpoints that return sensitive JSON rather than relying on obscurity of field names. Redact tokens before sharing formatted samples in support tickets.