CSV to PDF Shows Question Marks for Accents
Accents turn into question marks when you convert a CSV to PDF? The file isn't UTF-8. See why it happens and how to re-save it so the PDF reads right.
The short answer: your CSV is not saved as UTF-8
When accented letters come out of a CSV to PDF conversion as question marks or diamond symbols, the CSV file was not saved as UTF-8. The converter reads the file as UTF-8 text, so any other encoding turns letters like e with an accent, u with an umlaut or n with a tilde into a replacement character. The PDF is then faithfully showing a character that was already lost before the page was drawn. The fix is on the CSV side: save the file again as CSV UTF-8, then convert it. Plain English text, digits and punctuation are unaffected, which is why the problem often only shows up on the one row with a customer called Muller or a city called Sao Paulo.
Why Excel's default CSV causes it
The usual source is Excel's plain "CSV (Comma delimited)" option, which writes the file in your system's legacy Windows code page, not UTF-8. In that encoding an accented letter is stored as a single byte that is not valid UTF-8 on its own. Any reader that expects UTF-8 cannot make sense of that byte. Excel hides the problem because it reopens its own files using the same legacy code page, so the sheet looks fine on your screen. The damage only appears when another program, such as a converter, reads the file as UTF-8. Files exported from older accounting packages, CRMs and shop dashboards can have the same issue when they export in a legacy encoding.
What our converter does with those bytes
The CSV to PDF tool decodes the file as UTF-8 in your browser, drops a leading byte order mark if there is one, and replaces any invalid byte with the Unicode replacement character instead of stopping with an error. I checked this against the tool's source and with a quick test: the bytes for the word Cafe with an accented final e, saved in the legacy Windows encoding, decode to Caf followed by the replacement symbol. That is why you get a finished PDF with a bad character rather than a failure message. It also means the damage cannot be undone inside the PDF, because the original letter is not stored anywhere in the output.
How to fix it before converting
Re-save the CSV as UTF-8, then convert the new file. In Excel, choose File, Save As, and pick "CSV UTF-8 (Comma delimited)" from the file type list, not the plain CSV or the MS-DOS variant. In Google Sheets, File, Download, Comma-separated values already exports UTF-8. In LibreOffice Calc, tick "Edit filter settings" when saving and choose Unicode (UTF-8). A text editor works too: open the file in an editor that shows its encoding, set it to UTF-8, and save. Keep the original file untouched in case the export needs to be repeated, and convert only the re-saved copy so a mistake in one step cannot overwrite your source data.
How to check the file before you convert
Open the CSV in a plain text editor, not Excel, and look at the names that contain accents. If they already show a replacement symbol or stray characters there, the file itself is damaged and re-saving the same file will not bring the letters back; you need to export it again from the source system. If the editor shows the letters correctly, the file is fine and the conversion will keep them. This two-minute check tells you whether the fix is a re-save or a fresh export, and it avoids converting the same bad file several times and wondering why the output never changes.
Other things the converter does to a CSV
Three more behaviours are worth knowing when a PDF does not look like your sheet. The delimiter is detected automatically, so semicolon and tab separated files work without any setting. The first row is always drawn as a bold header row. And the converter caps the table at the first 5,000 rows and the first 40 columns, and it shows a warning when it had to cut anything. Quoted fields with commas or line breaks inside them stay in one cell. Very wide tables shrink the text, and the page turns to landscape when portrait would make it too small, so a file with many columns can still be hard to read on paper.
Frequently asked questions
Try it on PixelTools — free, no upload needed