Why Excel Ruins Your CSV Files (and How to Stop It)
The real, documented excel csv problems — stripped zeros, dates, scientific notation, encoding — and how to stop excel ruins csv mistakes for good.
You export a clean CSV, open it in Excel to take a quick look, save, and send it on. Somewhere in that round-trip, ZIP codes lost their leading zeros, an order number turned into 1.23457E+14, and a column of part numbers became dates. You did not change anything — Excel did, the moment the file opened.
This is not a bug, and it is not Excel being bad software. Excel is a calculation app, and a genuinely great one for analysis. The problem is that CSV is plain text, and Excel insists on interpreting that text as spreadsheet data on the way in and reformatting it on the way out. When the file itself is the deliverable, that helpfulness is exactly what you do not want.
Here are the real, documented ways it goes wrong — and how to stop it.
The documented ways Excel mangles CSVs
It strips leading zeros
Open a CSV with a ZIP code column and 02134 becomes 2134. Product IDs like 00042 become 42. Excel formats those columns as numbers by default, and numbers do not carry leading zeros. The text on disk is intact until you hit Save — then Excel writes its numeric version back, and the original is gone.
It converts long numbers to scientific notation
Order numbers, account numbers, IMEIs, and credit card numbers are long digit strings, not quantities you would ever do math on. Excel does not know that. A 16-digit value gets shown as 1.23457E+14, and if you save, the rounded scientific form can be written back to the file — permanently dropping the trailing digits. You cannot recover them.
It turns values into dates
This is the famous one. Excel auto-detects anything date-shaped and converts it. The gene name SEPT1 becomes a September date; MARCH1 becomes a March date. A part number like 1-5 becomes a date range. A measurement of 3/4 becomes the fourth of March. This is so well documented in genomics that researchers renamed human genes to stop Excel from corrupting their data.
It changes date formats by locale
A date written as 03/04/2026 means March 4 in the US and April 3 in much of the rest of the world. Excel reads it according to the machine's locale, then rewrites it on save. Open a file on a US laptop and a European one and you can end up with two different, silently swapped dates from the same source.
It drops rows past 1,048,576
An Excel worksheet maxes out at 1,048,576 rows. Open a larger CSV and the rows beyond that line are simply not imported — no error, no warning. If you save, you have just truncated your file. (More on this in CSV files too big for Excel.)
It guesses encoding wrong
Excel does not reliably detect UTF-8. Open a file full of accented names, currency symbols, or non-Latin text and you get é where é should be, or worse. Save it and the broken encoding gets baked in. The fix is to be explicit about encoding rather than letting an app guess — see fixing CSV encoding.
It re-quotes on save
Even when the displayed values look fine, Excel rewrites the file's structure on export: which fields get wrapped in quotes, how line endings are written, whether a BOM is added. A diff between your original export and Excel's re-saved version can light up across rows you never touched, which is a headache for version control and for any downstream system expecting a specific format.
What you had vs. what Excel made it
| What you had | What Excel made it |
|---|---|
02134 (ZIP code) | 2134 |
00042 (product ID) | 42 |
1234567890123456 (account number) | 1.23457E+15 |
SEPT1 (gene name) | 1-Sep |
3/4 (ratio) | 4-Mar |
1-5 (part number) | 5-Jan |
03/04/2026 (US date) | 04/03/2026 on a EU locale |
café (UTF-8) | café |
| Row 1,500,000 | (dropped — not imported) |
Every one of these is real, reproducible behavior, not an exaggeration. The pattern is the same throughout: Excel guesses what you meant, applies it on open, and writes it back on save.
How to stop it
The reliable fix is not a setting buried in an import wizard. It is to stop opening CSVs in a tool whose whole job is to reinterpret data. View and edit delimited files in a dedicated CSV editor that treats them as text, and only writes back the export options you explicitly choose.
CEESVEE is built for exactly that. It is a free, open-source (MIT) CSV and delimited-file viewer and editor for Windows, macOS, and Linux, built with Tauri, Rust, and React. The difference that matters here:
- It does not reinterpret your data. A leading zero stays a leading zero. A long number stays a long number.
SEPT1staysSEPT1. Nothing is silently converted to a date or rounded into scientific notation. - It round-trips faithfully. Parse, edit, save returns the same data you opened. The delimiter and encoding are auto-detected, with correct BOM handling, and Save / Save As make export explicit: you choose the delimiter, encoding, quoting style, line endings (LF or CRLF), and whether to include a BOM. Nothing changes behind your back.
- It handles files Excel cannot. A Rust core with a virtualized grid opens and scrolls 1,000,000-row, 100 MB+ files — well past Excel's row cap — with no rows quietly dropped.
- It still does the editing work. Cell editing, multi-column sort, find and replace (plain or regex), tabs, live stats, and light/dark themes. And it is 100% local: no cloud, no accounts, no telemetry. Your files never leave your machine.
For the everyday workflow of opening, fixing, and re-saving delimited files, see editing CSV without Excel and the case for a free Excel alternative for CSV work.
Where Excel still belongs
To be fair: when you actually need to analyze data — formulas, charts, pivot tables — Excel is excellent, and CEESVEE deliberately does none of that. The two are complementary. Clean and prepare the CSV in a tool that respects the format, then open it in Excel for analysis if you need to. The mistake is using a calculation engine as a CSV editor and being surprised when it edits your CSV.
Stop the mangling
If Excel keeps rewriting your exports, the fix is one download away.
- Download CEESVEE — free, open source, fully local.
- Open the file Excel keeps breaking; the delimiter and encoding are detected for you.
- Edit if you need to, then save with full control over delimiter, encoding, quoting, line endings, and BOM — so the file you save is exactly the file you meant to save.
Frequently asked questions
Why does Excel strip leading zeros from my CSV?
Excel treats CSV columns as numbers by default, and numbers have no leading zeros. So 02134 becomes 2134 and 00042 becomes 42 the moment the file opens. The data on disk is fine until you save, at which point Excel writes the reformatted version back.
Why does Excel turn numbers into scientific notation?
Long numeric strings like order numbers, IMEIs, or credit card numbers exceed the digits Excel shows in a General-format cell, so it displays them as something like 1.23457E+14. If you save, the rounded scientific value can be written back, permanently losing digits.
Why does Excel change my values into dates?
Excel auto-detects anything that looks date-like — gene names such as SEPT1 or MARCH1, fractions like 3/4, and ranges like 1-5 — and converts them to serial dates. This is well documented in genomics, where it has corrupted published datasets.
How do I open a CSV without Excel reformatting it?
Use a dedicated CSV editor that does not reinterpret values. CEESVEE is a free, open-source (MIT) viewer and editor for Windows, macOS, and Linux that keeps your data as-is and only writes back the export format you choose.
Can Excel open a CSV with more than a million rows?
No. A worksheet caps at 1,048,576 rows, and any rows past that are silently dropped on import. A tool with a virtualized grid, like CEESVEE, opens 1,000,000-row, 100 MB+ files without that limit.