Mycelgrid Guide
What Changes When You Convert JSON to CSV? Rows, Columns, Nested Data, and Data Loss
Understand what a JSON-to-CSV conversion preserves, normalizes, and can make ambiguous before you rely on the exported table.
JSON and CSV represent different shapes
JSON can preserve hierarchy and data types. CSV is a row-and-column text representation. Conversion therefore requires normalization choices; it is not just an extension change.
How rows are created
Arrays of objects naturally become multiple rows. A single object usually becomes one row. Primitive top-level values are placed in a value column. Equal-length top-level arrays inside one object can be transposed by index.
That transpose assumes index alignment only as a structural rule; it does not validate business meaning.
Nested objects and flattening collisions
Flattening converts nested objects into dot-path columns such as customer.name. This is useful for tables, but an original key containing a dot can collide with a generated path. Flattened headers are therefore not guaranteed reversible provenance.
Arrays: JSON text or pipe joining
Arrays can remain JSON text inside one cell or be joined with a literal pipe. JSON text keeps more explicit structure. Pipe joining is easier to scan but becomes ambiguous when an element itself contains |.
Missing, null, and empty values
The current converter does not preserve a distinction between an absent field and an explicit JSON null. Both use the selected null representation. With blank-cell mode, empty strings can also look identical.
CSV serialization and formula guard
Delimiter, newline, header, quoting, and BOM choices affect the exported text. RFC 4180 describes a common CSV form, but CSV implementations vary.
The current formula guard prefixes an apostrophe for guarded text beginning with =, +, -, @, tab, or carriage return. Its scope is implementation-specific and should not be treated as universal spreadsheet sanitization.
When to stop instead of convert
Do not rely on CSV as the only copy when you need exact hierarchy, original types, missing-versus-null state, unambiguous array boundaries, or collision-free source paths. Keep JSON as the authoritative structured source and treat CSV as a derived artifact.