CSV vs Excel: Which Format to Use and When

CSV vs Excel is a choice between a plain-text table and a full spreadsheet document. A CSV stores values and nothing else. Unlike an .xlsx workbook, it carries no formulas, no formatting, no second sheet — which is exactly why every CRM, bank, and analytics tool can read one, and why so many people lose data moving between the two.

Below: what each format holds, which one to reach for, what gets destroyed in conversion, and how to convert either direction without handing the file to a server.

What each format actually stores

The two formats answer different questions. A CSV answers "what are the values?" An .xlsx file answers "what does this document look like and how does it calculate?"

A CSV is text you can read without a spreadsheet program at all:

first_name,last_name,phone,country
Ada,Lovelace,+442071234567,UK
Grace,Hopper,+12025550143,US

RFC 4180 registers text/csv and defines the rules that make that file parseable: each record sits on its own line delimited by a line break, fields are separated by commas, and any field containing a comma, a line break, or a double quote must be wrapped in double quotes — with a literal double quote written twice to escape it. That is the entire specification worth memorizing. There is more on the format's history and edge cases in our explainer on what a CSV file is.

An .xlsx file is a different animal. Microsoft describes the Excel Workbook format as the default XML-based file format for Excel, and it is a compressed package of XML parts: one per worksheet, plus shared strings, styles, defined names, and relationships. It stores the formula in a cell and the cached result. It stores that a column is currency with two decimals. It stores charts, conditional formatting rules, and the fact that row 14 is frozen.

The row and column ceiling is worth noting because it is a property of the worksheet, not of your data: Microsoft's published limits are 1,048,576 rows by 16,384 columns, with 32,767 characters per cell. A CSV imposes none of those — the constraint moves to whatever program is reading it.

Where CSV is the better choice

Reach for CSV whenever the file is in transit rather than at rest.

Imports and exports. Every CRM, email platform, accounting package, and ad tool reads CSV; a good number read nothing else. When you export contacts from one system to load into another, the CSV is the interchange format and the spreadsheet is an optional detour.

Large datasets. A million-row export as .xlsx can take a minute to open and hundreds of megabytes of memory, because the program is reconstructing a document. The same data as CSV is a stream of lines.

Anything a script or database will read. Loading a CSV needs a parser; loading an .xlsx needs a library that understands the Open XML package. Bulk-load utilities in Postgres, MySQL, and BigQuery all take CSV directly.

Version control and diffs. Two CSVs can be compared line by line. Two .xlsx files are binary archives — a diff tells you only that they differ.

Long-term archiving. Plain text stays readable when the software that produced it is gone. That is why public bodies publish open data as CSV.

Where Excel is the better choice

Reach for a workbook whenever the file is a working document rather than a payload.

Anything with a formula. A CSV cannot store =SUM(B2:B400) as a live calculation. Save the workbook as CSV and the formula is replaced by whatever number it evaluated to at that moment.

Multi-sheet models. Budgets, forecasts, and reports that reference a lookup tab need a container that holds more than one table.

Presentation. Number formats, conditional formatting, frozen headers, and charts exist so a human can read the numbers. None of them survive a text export.

Data types you want preserved. A workbook records that a column holds dates, so 03/04/2026 stays a date. In a CSV it is seven characters, and how they are read depends on the reader's locale.

Collaboration in place. Comments, tracked changes, and protected ranges are document features. A CSV has no room for them.

What breaks when you convert between them

This is where most of the pain in the CSV vs Excel question actually lives, and both directions have a failure mode.

Excel → CSV loses everything except values. Microsoft's own documentation on supported file formats is blunt about it: if you save a workbook in any text format, all formatting is lost, and the CSV option "saves only the active sheet". A five-tab workbook becomes one table, silently. Formulas collapse to their results. Charts, images, and comments are gone.

CSV → Excel silently rewrites your data. When Excel opens a .csv it uses the current default data-format settings to interpret each column, as Microsoft's import and export guidance explains. In practice that means leading zeros vanish from postcodes and account numbers, long identifiers turn into scientific notation, and text that resembles a date gets converted to one. Nothing warns you. The fix is to import through Data > From Text/CSV and set the affected columns to Text before the data lands — the same route covered step by step in our guide to opening a CSV in Excel.

The separator is not always a comma. In locales where the comma is the decimal separator, Excel writes and expects a semicolon as the list separator. A file exported in Berlin can open in London as one long column.

How to choose in practice

Answer one question: is this file going somewhere, or is it being worked on?

• Going somewhere — a CRM import, a database load, a partner handover, an archive: CSV. Export it, clean it, send it.
• Being worked on — calculations, multiple tabs, a report someone reads: Excel. Keep the workbook as the source of truth.
• Both — most real workflows: keep the workbook, export a CSV each time you need to move data out. Treat the CSV as a snapshot, not the master copy.

One rule prevents most damage: never round-trip. Opening a CSV in Excel, editing it, and saving it back as CSV runs your data through Excel's conversion logic twice. If the file only needs deduplicating or reformatting, do that without opening it in a spreadsheet at all.

Converting either way without uploading the file

Converting between the two formats normally means a web converter that wants your file on its server — an odd thing to accept for a spreadsheet full of customer names, emails, and phone numbers.

It is avoidable. Sigmera converts CSV to Excel and Excel back to CSV with client-side code that runs inside your browser tab, so the file is read from your device and never transmitted. There is no upload, no server copy, and no processor to name in a GDPR record — because there is no transfer. The same applies to the cleaning steps you would otherwise do in Excel and risk on the way back out: removing duplicate rows from a CSV keeps the file as text throughout, so leading zeros and long IDs come out the way they went in. Every tool is free to preview and the plans differ only in how many finished files you download.

Where Excel genuinely wins — formulas, multi-sheet models, charts — keep using it; our comparison of Sigmera and Excel is candid about that split. The point is not to replace the spreadsheet. It is to stop letting a format conversion decide what your data says.