XLSX vs CSV: which information survives?

An XLSX file can preserve a workbook’s sheets, formulas, formatting and tables. A CSV file carries rows of plain-text fields instead—not the workbook around them. Exporting may retain the displayed data, but not its calculation rules or presentation; importing can then reinterpret the text. Microsoft’s workbook structure documentation describes the separate worksheet, formula, table and formatting elements that make XLSX more than a grid of values.

The example below follows Microsoft’s documentation for Excel for Microsoft 365 on Windows, checked on 6 November 2026. It is an illustration, not a report of a hands-on export test.

The difference in one workbook

Imagine an orders.xlsx workbook containing three sheets:

• Orders: a named Excel table called Sales, with blue headers, alternating row colours and filter buttons.
• Prices: reference prices, with B2 holding 19.90 and B3 holding 7.50.
• Summary: a formula, =SUM(Orders!F2:F3), showing the combined order total of £62.30.

Here, “table” means an Excel table object, not merely some cells arranged in rows and columns.

The Orders sheet displays this data in columns A to F:

Code	Customer	Order_date	Quantity	Unit_price_GBP	Total_GBP
00127	North, West	2026-11-06	2	19.90	39.80
00408	Elm Ltd	2026-11-07	3	7.50	22.50

The codes are stored as text. The dates are date values formatted as yyyy-mm-dd. Prices and totals are in pounds sterling, displayed to two decimal places without currency symbols.

The unit-price cells contain =Prices!B2 and =Prices!B3. The total cells contain =D2*E2 and =D3*E3. Thus 2 × £19.90 = £39.80, and 3 × £7.50 = £22.50; together they make £62.30.

Now make Orders the active sheet and use Save As → CSV (Comma delimited). With comma separators, decimal points and formula results displayed, the CSV text can look like this:

Code,Customer,Order_date,Quantity,Unit_price_GBP,Total_GBP
00127,"North, West",2026-11-06,2,19.90,39.80
00408,Elm Ltd,2026-11-07,3,7.50,22.50

Only Orders is exported; Prices and Summary require separate files. Excel’s regional and separator settings can also produce semicolons rather than commas. The .csv extension alone does not settle which separator was used. See Microsoft’s text-file import and export instructions.

What survives—and what does not

The CSV retains the headings and data fields in this example. But 39.80 is an exported result, not =D2*E2. Changing the quantity after import will not recalculate that total unless you add a formula.

The blue headers, row colours, column widths and number-format settings are absent. The Sales table and its filter settings are absent too: its rows survive, not the table object.

Precision needs checking. Excel documents CSV export as saving text and values as displayed. If a cell contains more decimal places than it shows, do not assume the extra digits survive. Formatting can influence the exported characters even though the formatting rule is lost.

There is a narrow formula exception: when Excel displays formulas instead of results, it can export the formula characters. That does not preserve the workbook’s structure or missing sheets. Microsoft’s format-loss guidance covers these export limits.

What the CSV format leaves to the application

CSV means comma-separated values. A delimiter is the character that separates fields. Under the common convention documented by RFC 4180, commas separate fields, and double quotes enclose fields containing commas, quotation marks or line breaks. An internal quotation mark is written twice.

That is why "North, West" occupies one customer field rather than two. Quotes mark field boundaries; they are not an instruction saying “store this as text.” RFC 4180 is an informational description of a common convention, not a guarantee that every application behaves identically.

CSV has no built-in column declarations equivalent to “identifier,” “date” or “currency.” It can carry the characters 00127 perfectly well. The receiving application must decide what those characters mean.

When Excel opens a CSV directly, automatic conversions may turn that code into the number 127. Long numeric identifiers can also lose digits when treated as numbers. These are import decisions, not evidence that CSV cannot contain leading zeros or long strings. Excel for Microsoft 365 provides automatic data-conversion controls under File → Options → Data.

Dates and decimal numbers need an agreed interpretation too. For example, 03/04/2026 can mean 3 April or March 4 under different regional conventions. Power Query’s locale settings control how such text is converted. Writing an unambiguous date helps, but the importer still needs the intended column type.

Check the export separately from the import

Keep the XLSX as the working original when you need formulas, several sheets or formatting. Treat the CSV as a separate exchange copy, then check two things:

1. What was written? Open the CSV in a plain-text editor without saving changes. Check the headings, field separators, codes and displayed precision against the intended export.
2. What was interpreted? In Excel, use Data → From Text/CSV to inspect the import. Check the delimiter and character encoding. In the preview, turn off automatic data-type detection so columns start as text. Then choose Transform Data to assign types deliberately, leaving identifiers as text. Microsoft’s Text/CSV connector documentation explains these controls.

Character encoding is the rule used to represent characters in the file. UTF-8 can carry accented and non-Latin text, but the receiving application must read it correctly. Microsoft’s UTF-8 opening guidance explains Excel’s handling and import alternatives.

Finally, check the imported codes, dates, row count and totals—not just whether the sheet looks tidy. Excel may create a new table during import, but that does not restore the original Sales table’s settings. Saving as XLSX preserves the new workbook; it cannot recover sheets, formulas or formatting omitted from the CSV.