How to convert Excel to CSV without breaking your data

Define what must survive

Converting a workbook to CSV is not the same as changing its extension. XLSX is an Office Open XML package that can hold many worksheets, formulas, styles, charts, merged cells, validation, and relationships. CSV is a text representation of one flat table.

Create a preservation checklist before exporting:

Field or feature	Example	CSV decision
Identifier	001247	Preserve as text
Long reference	9876543210123456	Preserve as text, never floating-point
Date and time	2026-07-25T14:30:00+05:30	Declare format and timezone
Decimal	1234.50	Agree decimal and field delimiter
Formula	=SUM(B2:B10)	Export stored result or retain XLSX
Multiple sheets	Orders and customers	Export separate files or retain XLSX
Display format	12.5%	Decide whether receiver needs 0.125 or 12.5%

The receiving system’s import contract is the authority. Ask for its delimiter, headers, encoding, date syntax, decimal convention, null representation, and maximum rows. A valid CSV can still be wrong for a particular importer.

Keep the XLSX master

Copy the workbook and leave the original unchanged. CSV cannot become a faithful backup of an XLSX workbook because it cannot reconstruct formulas, charts, formats, comments, hidden sheets, or workbook relationships.

Microsoft’s current documentation states that CSV saves only the active worksheet and loses formatting. FileGizmo makes that boundary explicit by asking you to select one sheet and by exporting stored cell values rather than presenting CSV as a workbook replacement.

If formulas must be current, open the workbook in its trusted spreadsheet application, allow expected calculations to finish, inspect external links, and save it before conversion. FileGizmo does not contain Excel’s calculation engine. A workbook saved with stale cached values can produce stale CSV values.

Macro-enabled XLSM and legacy XLS files are outside the current FileGizmo tool. Convert them to a reviewed XLSX copy in trusted desktop software when appropriate, and do not assume macros or legacy features belong in the plain-data export.

Protect identifiers before export

Postal codes, phone numbers, tracking IDs, membership numbers, account references, and product codes can contain only digits while still being text.

Excel can infer numbers when a CSV is opened directly. This may:

• remove leading zeros;
• display long values in scientific notation;
• round digits beyond normal numeric precision;
• convert strings that resemble dates;
• interpret formula-triggering characters.

Format identifier columns as Text in the workbook before preparing the export. After conversion, inspect the CSV in a plain-text editor. When reopening it in Excel, use Data > From Text/CSV and assign Text to those columns. Microsoft’s import guidance recommends this route for preserving leading zeros.

Do not add an apostrophe to the actual identifier unless the receiving system expects it. An apostrophe can be a spreadsheet-entry convenience but may become unwanted source data in another importer.

Choose one clean worksheet

Each CSV should contain one logical table. Use one header row with unique, non-empty column names. Remove title blocks, decorative blank rows, repeated headers, subtotals, and footer notes from the export copy unless the receiver explicitly needs them.

Merged cells do not describe a stable rectangular dataset. Unmerge them and repeat required values into the appropriate rows. Avoid relying on fill colour, bold text, indentation, or comments to convey meaning because those cues disappear.

If the workbook contains several related tables, export them separately with clear names:

• customers.csv
• orders.csv
• order_items.csv

Preserve stable keys such as customer_id and order_id so the receiver can rebuild relationships. Do not paste three tables beneath each other in one file merely to keep a single download.

Choose delimiter and encoding deliberately

RFC 4180 describes comma-separated records and double-quote escaping for fields that contain commas, quotes, or line breaks. Real import systems also use semicolon and tab delimiters.

Excel to CSV offers comma, semicolon, and tab output and writes UTF-8. Choose the delimiter documented by the receiver:

• Comma for a conventional CSV contract.
• Semicolon when the receiver specifies it, often in workflows where comma is the decimal separator.
• Tab for a tabular text contract or data with many commas.

Do not manually replace commas after export. A comma inside "Mumbai, Maharashtra" is data, not a separator. A conforming writer quotes it. Manual search-and-replace can corrupt quoted fields and multiline descriptions.

UTF-8 supports multilingual names and symbols, but a legacy importer can still expect another encoding. Test Hindi, accented Latin text, currency symbols, and emoji when they are valid source data. If the receiver requires a different encoding, perform that conversion in a controlled tool and recheck the result.

Convert locally and inspect the result

1. Open Excel to CSV.
2. Choose a reviewed modern XLSX copy.
3. Select the exact worksheet.
4. Pick the receiver’s delimiter.
5. Convert in the browser worker.
6. Download the UTF-8 CSV.
7. Open it in a plain-text editor before importing it elsewhere.

FileGizmo exports stored values and escapes potential spreadsheet-formula prefixes in its CSV output. OWASP documents the risk of untrusted input becoming active formulas when opened in spreadsheet software. Sanitization should happen at the export boundary without destroying the governed source.

The local processing boundary keeps the selected workbook on the device, but it does not make an unreviewed workbook safe. Treat unexpected formulas, links, and macros cautiously in the application that opens them.

Run a reconciliation check

Compare the source worksheet and CSV using measurable checks:

1. Count data rows, excluding the one header row.
2. Count columns in the header and representative records.
3. Compare the first five and last five identifiers.
4. Search for blanks in required columns.
5. Check duplicates in the primary key.
6. Inspect the longest identifier and largest number.
7. Verify one date around midnight and one with a timezone.
8. Check fields containing commas, quotation marks, and line breaks.
9. Confirm non-ASCII text.
10. Import the file into a test or staging destination before production.

A file opening successfully is not proof that the data survived. Row counts can match while identifiers were rounded or dates shifted.

For the broader choice among tabular, structured, and workbook formats, read CSV vs JSON vs Excel. If the destination needs JSON objects instead of a table, use CSV to JSON after cleaning headings and deciding how types should map. The TinyWow comparison explains the privacy and workflow differences between a broad server-backed converter catalogue and FileGizmo’s focused browser-local tools.