How to Fix Weird Characters in a CSV
Weird characters usually mean the file's bytes were opened with the wrong text encoding, not that the CSV rows themselves are broken.
Identify whether the source is UTF-8, Windows-1252, UTF-16, or Shift_JIS, check a preview, and convert the decoded text to UTF-8. If the original bytes were already replaced with question marks, conversion cannot reconstruct the lost characters.
Open the CSV Encoding FixerWhat mojibake looks like
Text such as José, ’, or sequences of boxes is often called mojibake. It appears when valid bytes are decoded under a different character map.
Question-mark diamonds can also indicate invalid byte sequences, while plain question marks may mean the exporting system already discarded characters.
Choose the source encoding, not the desired output
The source encoding tells the converter how to read the current bytes. The output encoding is the new format you want—usually UTF-8.
Windows-1252 is common in older Western European exports. Shift_JIS remains common in Japanese business systems. UTF-16 files often contain a byte-order mark or a repeating null-byte pattern.
Why resaving can make corruption permanent
If you open a file under the wrong encoding and save it again, the garbled text becomes the new literal data. Always return to the untouched export when possible.
A preview is essential: confirm names, punctuation, accented text, and Japanese characters before downloading the UTF-8 copy.
How to fix it safely
- Upload the untouched source file to CSV Encoding Fixer.
- Check the detected encoding and preview.
- Choose a different source encoding if the preview is wrong.
- Download UTF-8 with a BOM when Excel compatibility matters.