Common JSON Errors and How to Fix Them

Two Buckets: Syntax and Shape

Almost every "broken JSON" ticket falls into one of two categories. Syntax errors mean the parser cannot even load the document. Shape errors mean it loaded but does not match what your code expects.

The fixes are different, so diagnose the bucket first.

Syntax Errors

These are the ones that make JSON.parse throw.

// Bad
{
  "a": 1,
  "b": 2,
}

// Good
{
  "a": 1,
  "b": 2
}

// Bad
{ 'name': 'Ada' }

// Good
{ "name": "Ada" }

// Bad
{ name: "Ada" }

// Good
{ "name": "Ada" }

// Bad
{ "note": "Line one
Line two" }

// Good
{ "note": "Line one\nLine two" }

Encoding Errors

Bytes that look like garbage are almost always an encoding mismatch.

• UTF-8 is the only interoperable choice for JSON.
• Do not embed a BOM in JSON files—many parsers reject them.
• Escape non-ASCII characters (\u00e9) or serve as UTF-8 with the correct Content-Type header.
• "é" showing as "é" means UTF-8 was decoded as Windows-1252.

Shape Errors

The JSON parsed, but something downstream broke.

Errors From CSV → JSON

When JSON is generated from CSV, expect these quirks.

• Boolean columns become strings ("true", "false") unless the converter is told to cast.
• Empty cells become empty strings, not null—decide the convention.
• Numeric IDs may be cast to numbers, losing leading zeros.
• Dates in local format get preserved as strings; convert to ISO 8601 in a follow-up step.

Diagnostic Workflow

Work through this list top to bottom.

• Run jq . file.json (or JSON.parse) to catch parse errors.
• Check encoding with file file.json.
• Validate shape against a JSON Schema.
• Diff against a known-good sample.
• For CSV sources, re-run the conversion with typed columns.

Real-World Examples

The most common tickets we all see.

Common Mistakes

Habits that create more errors than they fix.

• Ignoring parser errors by wrapping in try/catch and moving on.
• Hand-editing minified JSON and forgetting a comma.
• Trusting an online "fixer" instead of understanding the error.
• Using JSON5 in production and being surprised strict parsers reject it.
• Skipping schema validation because "the client is trusted".

Best Practices

The habits that keep JSON healthy.

• Lint JSON in CI.
• Validate against a schema at every trust boundary.
• Standardize on UTF-8 without a BOM.
• IDs as strings; dates as ISO 8601.
• Fail loudly on unknown fields.

Why Use Convert CSV Online?

When you suspect the CSV source is the problem, Convert CSV Online is free, browser-based, and needs no account for everyday conversions. Preview the CSV in the Online CSV Editor, fix the offending column, then regenerate JSON. Client-side workflows work on Windows, macOS, and Linux browsers.

Conclusion

Almost every JSON bug is a trailing comma, an encoding mismatch, or a shape you did not expect. Learn the three, then move on to validating with a schema.