Fix garbled CSV characters in Excel: UTF-8, BOM, and import settings

First check whether the original characters are intact

If a name such as Zoë, Montréal, or 김민수 is correct in the source but garbled only after opening the CSV in Excel, check how Excel is interpreting the file’s encoding. Downloading UTF-8 with a byte order mark, or importing the file through Excel’s Text/CSV flow, can address that mismatch.

If the source already contains question marks or replacement characters, adding a BOM cannot recover the lost text. Return to a correct source export before making another copy.

Other symptoms need different fixes:

Symptom	Likely area to inspect	Useful next action
Accented or non-Latin names are garbled	Encoding	Select UTF-8 during import or use a UTF-8 BOM download
An entire record appears in one column	Delimiter	Match comma, semicolon, or tab to the source
An address creates an extra column	Quoting	Check quotes around the field containing the comma
A note creates extra rows	Embedded line breaks	Check that the complete multiline field is quoted
0012 becomes 12	Automatic type conversion	Import the identifier column as Text

Do not use a global find-and-replace on the data until you know which problem you are solving.

Choose the download or import route deliberately

In Tooliv’s JSON and CSV converter, paste text or open/drop one UTF-8 .json, .csv, or .tsv file. Input is limited to 5,000,000 bytes. Choose JSON to CSV and the intended delimiter. Enable Add UTF-8 BOM for Excel (CSV download) when you want that encoding hint in the downloaded file.

The Copy button copies the CSV text. If you paste that text into another editor and save a file, the editor’s encoding settings determine the new file. That is a different path from downloading with the BOM option enabled.

Microsoft’s UTF-8 CSV guidance describes normal opening for UTF-8 files saved with BOM and an alternative through Data → Get Data → From File → Text/CSV. The exact labels vary across versions and platforms. In the preview, confirm the characters and delimiter before loading.

The BOM identifies the text encoding. It does not identify a column as an account number, repair malformed CSV, or preserve a leading zero against a spreadsheet’s automatic number conversion.

Try a sample with names, commas, and quotes

Paste this JSON into Tooliv with the comma delimiter selected:

[
  {"id":"0012","name":"Zoë","note":"Montréal, QC"},
  {"id":"0013","name":"김민수","note":"\"Ready\" confirmed"}
]

The conversion functions produce this CSV:

id,name,note
0012,Zoë,"Montréal, QC"
0013,김민수,"""Ready"" confirmed"

The comma inside Montréal, QC belongs in one cell. The doubled quotation marks encode the quotation marks around Ready. RFC 4180 documents these common CSV quoting rules.

This exact sample was run through Tooliv’s conversion functions, and converting the CSV back returned the original strings. The BOM-enabled download-content function also produced the UTF-8 byte prefix EF BB BF. These are function-level checks; they do not claim that the file was opened in every Excel version.

A note containing a line break needs quotes around the entire field:

id,note
0014,"Line one
Line two"

The parser returns one record with a line break inside note; it does not treat the two lines of that field as separate records. That case was also checked with the conversion function. Splitting text at every comma or newline would not reproduce this behavior.

Set identifier types before accepting Excel’s conversion

Confirm that 0012 remains 0012 after import. Microsoft’s guidance on leading zeros and large numbers describes importing these values as Text and notes Excel’s automatic number conversions.

In the import or transformation stage, set identifier columns to Text before loading values that could be coerced. If an earlier step has already removed zeros or rounded a long number, changing the display format afterward cannot reliably reconstruct the original value. Reimport from the unchanged CSV.

Keep quantities as numeric values when the destination needs arithmetic, but make that decision column by column. Postal codes, product codes, and account identifiers are often text even when they contain only digits.

Check delimiter and header assumptions

Tooliv defaults to commas and offers semicolon and tab options. Opening a .tsv file selects tab; for other input, confirm the delimiter yourself. A CSV header must be nonempty and unique, and each record must have the same number of fields as the header.

The table preview shows up to ten data rows and twelve columns. Use it to spot shifted fields, then check the complete downloaded dataset. Replacing every separator blindly can damage punctuation inside values or decimal notation. Non-UTF-8 files must be correctly re-exported first; this file reader does not guess legacy encodings.

Nested JSON objects and arrays stay as JSON text in a single CSV cell. CSV-to-JSON keeps every value as a string, and empty cells do not restore distinctions between null, a missing property, and an empty string. The JSON-versus-CSV guide shows that round trip explicitly.

Review external values before opening them as spreadsheet data

Spreadsheet programs can interpret certain cell contents as formulas. OWASP explains why ordinary CSV quoting is not sufficient protection and why saving and reopening can affect escaping. A BOM is also unrelated to formula execution.

For unfamiliar exported data, review formula-like cells and use an import path with the intended text types. Do not treat a successful conversion as a security check.

Finish by comparing row counts, headers, representative names, identifiers, empty values, and multiline notes in the actual destination. For a larger handoff involving both prose and data, follow the document and data review workflow.