Why emoji and accents break when you open a CSV in Excel
You open a file of comments and "café" has become "café", the hearts are "â¤ï¸" and every row from Brazil is unreadable. The data is fine. Excel read the bytes with the wrong alphabet. Here is why it happens and four ways to stop it.
Updated Nov 19, 2026 · 6 min read
What is going wrong
A CSV is plain text. It stores each character as one or more bytes, and it carries no label saying which system was used to do that. Almost every modern tool writes UTF-8, where é is two bytes and an emoji is four. When you double-click a CSV, Excel has to guess, and on most computers its guess is the older regional code page (Windows-1252 in Western Europe and the Americas), where every byte is one character.
So the two bytes of é are shown as two characters, and you get the familiar damage:
| Written | Shown when UTF-8 is read as Windows-1252 |
|---|---|
| café | café |
| São Paulo | São Paulo |
| niño | niño |
| ❤️ | â¤ï¸ |
| 😍 | 😠|
If your file shows à followed by another symbol wherever an accent should be, this is the cause. Nothing is lost yet: the bytes in the file are still right.
The byte-order mark, and why some CSVs just work
There is one signal Excel does respect. If the file begins with three invisible bytes, EF BB BF, known as the UTF-8 byte-order mark or BOM, Excel reads the whole file as UTF-8. No setting, no wizard.
The CSV you download here starts with that mark, so a double-click shows emoji, accents, Arabic, Cyrillic and Thai correctly. That is true for the Instagram, TikTok, YouTube and Facebook CSV, and for the CSV of each summary view.
The mark can be lost on the way, which is how a file that opened fine yesterday breaks today:
- Someone opened the CSV in a text editor or a script and saved it again without the mark.
- The file went through a tool that rewrites CSVs: a CRM export, a mail merge add-in, a converter.
- It was saved from Excel as CSV (Comma delimited), which is not UTF-8 at all (more on that below).
- The CSV came from another source that never wrote the mark.
One file here has no mark on purpose: the live link for =IMPORTDATA. Google Sheets always reads UTF-8, and a mark at the start would end up inside the first header cell.
Fix 1: open it with Data, From Text/CSV
This works for any CSV, with or without the mark, in Excel 2016 and later on Windows and in current Excel for Mac.
- Open a blank workbook. Do not double-click the CSV.
- Data, Get Data, From File, From Text/CSV. Choose the file.
- In the preview window set File origin to
65001: Unicode (UTF-8). The preview changes as soon as you pick it: check a row with an accent. - Set Delimiter to Comma.
- Set Data type detection to Do not detect data types. This keeps long IDs exact; the reason is in why Excel rounds long comment IDs.
- Press Load.
In older Excel the same choice is in the Text Import Wizard: Data, From Text, then File origin: 65001 on the first step.
Fix 2: skip the CSV
An XLSX file stores text as Unicode by definition, so there is nothing to guess. The Excel workbook holds the same rows as the CSV plus five more sheets, and Copy for Sheets pastes Unicode text straight into Excel or Google Sheets. Use the CSV when another program needs it. For reading in a spreadsheet, the workbook has fewer ways to go wrong.
Fix 3: put the mark back
When you have a CSV without the mark and want it to open by double-click for other people:
- Windows Notepad: open the file, File, Save as, and choose UTF-8 with BOM in the Encoding list.
- Google Sheets: File, Import the CSV (Sheets reads it as UTF-8), then File, Download, Microsoft Excel (.xlsx) and send the XLSX instead.
- Excel: open it with Fix 1, then File, Save As and choose CSV UTF-8 (Comma delimited) (*.csv). Excel writes the mark for you.
The save that destroys emoji for good
Excel offers several CSV types under Save As. Two look alike and behave very differently:
| Save as type | What happens to 😍 and é |
|---|---|
| CSV UTF-8 (Comma delimited) (*.csv) | Kept. The file is UTF-8 with the mark. |
| CSV (Comma delimited) (*.csv) | é is kept only on a computer with a matching code page. Every emoji, and every letter outside that code page, is replaced with a question mark. |
The question marks are written into the file. This is the one case where the damage cannot be undone: export the post again or go back to the original download. A quick test after any save, in a spare cell:
=COUNTIF(E:E, "*~?~?*")
It counts comments containing two question marks in a row, which real comments sometimes have and a wrecked emoji always has. Compare the number before and after saving. If it jumped, the save replaced characters.
Three look-alike problems that are not encoding
- Everything is in column A. In regions where the decimal sign is a comma (most of Europe, Brazil), Excel expects a semicolon between columns when you double-click. The file is fine. Open it with Fix 1 and choose Comma, or select column A and use Data, Text to Columns.
- Emoji show as empty boxes or in black and white. That is the font on the device. Excel on Windows often draws emoji as outlines inside cells, and old systems lack the newest emoji. The characters are intact: paste a cell into a browser to check.
- One comment is spread over several rows. The comment has line breaks and the program that opened the CSV ignored the quotes around it. Excel and Google Sheets handle this correctly. If a script did it, use a real CSV parser instead of splitting on line breaks. A row count check catches it:
=COUNTA(B:B)-1should equal the comments read shown on the export page.
Counting characters when emoji are involved
One side effect to know before you measure comment length. Excel and Google Sheets count most emoji as two characters, and compound ones (flags, skin tones, families) as more:
=LEN("😍")
returns 2 in both. A length filter such as =LEN(E2)>=120 is still a fair test for long comments, but "under 3 characters" is not a reliable test for "one emoji". To find emoji-only comments, strip the letters and digits instead, as shown in how to clean comments in a spreadsheet, and to count which emoji appear most, use the word and emoji frequency guide.
Comments in several languages are common: for research work across languages, see exporting comments for research and sentiment analysis.
FAQ
Why does my CSV show é instead of é?
The file is UTF-8 and Excel read it as an older one-byte code page. Open it through Data, From Text/CSV and set File origin to 65001: Unicode (UTF-8), or use a CSV that starts with a UTF-8 byte-order mark.
What is a BOM?
Three invisible bytes (EF BB BF) at the start of a file that mark it as UTF-8. Excel reads a CSV with the mark correctly on double-click.
My emoji turned into question marks after I saved. Can I get them back?
Not from that file. Saving as plain CSV (Comma delimited) writes question marks in place of the emoji. Go back to the original download or export again, and save as CSV UTF-8 or as XLSX.
Does Google Sheets have this problem?
No. Google Sheets reads CSV files as UTF-8 on import and through IMPORTDATA, so emoji and accents arrive intact.
Why are all the columns in one cell?
Your regional settings make Excel expect semicolons between columns. Import the file with Data, From Text/CSV and choose Comma as the delimiter.