An Excel File Is Not What It Looks Like. Here Is What CSV Export Strips Away.
An Excel file is not what it looks like. Open a well-formatted spreadsheet and you see a clean grid of columns and values. Look at the underlying structure and you find a ZIP archive containing multiple XML files: one for the worksheet data, one for shared strings, one for styles, one for relationships, and several others depending on what the workbook contains.
CSV is the opposite. Open a CSV file in a text editor and you see exactly what is in it: values separated by commas, rows separated by line breaks. Nothing is hidden.
This transparency is why CSV became the universal exchange format for tabular data. When systems need to move data between them, CSV is almost always the format they agree on. Not because it is richer than Excel, it is not, but because it makes no assumptions about what the reader already has installed.
From VisiCalc to OOXML: The Evolution of Spreadsheet Files
The spreadsheet predates Excel by more than a decade. VisiCalc, developed by Dan Bricklin and Bob Frankston and released in October 1979, was the first commercial electronic spreadsheet. It ran on the Apple II and was widely credited with driving the adoption of personal computers for business use. VisiCalc stored its data in a simple proprietary format.
Lotus 1-2-3, released in January 1983, displaced VisiCalc and became the dominant business software of the mid-1980s. Microsoft released Excel 1.0 for the Macintosh in September 1985 and Excel 2.0 for Windows in 1987. By the early 1990s, Excel had overtaken Lotus 1-2-3, and the Excel file format (.xls) had become a de facto standard for spreadsheet data exchange, even though it was a proprietary binary format undocumented publicly by Microsoft.
Microsoft introduced the Office Open XML format, which produces.xlsx files, with Office 2007. The format was standardized as ECMA-376 in December 2006 and as ISO/IEC 29500 in November 2008. An xlsx file is a ZIP archive containing XML documents following the OOXML specification, which runs to thousands of pages across multiple parts. The relationship between the components, the worksheet data, shared strings, styles, and relationships, reflects a design optimized for full-featured office software, not for data interchange. That complexity is what makes the format incompatible with simple data pipelines.
Why Systems Prefer CSV Over Excel
Every database that exports query results, every analytics platform that shares reports, and every API that returns tabular data uses CSV as a default. The format traces to the early 1970s, when IBM's Fortran compiler used comma-separated values for list-directed input and output on OS/360 systems. That simplicity survived every evolution in computing because it makes no demands on the recipient.
RFC 4180, published by the IETF in October 2005, formalized the CSV format for the first time, defining the comma as the delimiter, specifying how to quote fields containing commas or line breaks, and establishing CRLF as the line ending. The RFC acknowledged that practice varied and sought to establish a baseline compatible with existing implementations. It is not a strict standard in the sense that all parsers comply with it; many do not. But it gave tooling a common reference point.
An Excel file requires software capable of reading the OOXML specification. A CSV file will open in any context that can read plain text, which is every computing context that exists. That difference is the practical reason CSV remains the preferred format for data transfer between systems.
What Survives and What Does Not
The values survive. Dates, numbers, and text transfer intact, with one important caveat: how dates and numbers are formatted in Excel determines how they appear in CSV. A date displayed as "July 15, 2026" in one locale might export as "15/07/2026" or "2026-07-15" depending on system settings. A number stored as currency with two decimal places might export as "1234.56" or "1,234.56" depending on regional formatting. The underlying value is unchanged; its plain text representation depends on context.
Formulas do not survive as formulas. They export as their most recently calculated result. A cell showing =SUM(B2:B10) becomes the number that formula produced. This is often exactly what the receiving system needs, since a pipeline wants the answer, not the equation, but it means the exported CSV has no connection to the source logic.
Merged cells present a specific problem. When cells are merged in Excel, only the first cell in the merge holds the value. On export, the other cells in the merge appear empty. For data pipelines that assume one value per cell per row, this creates gaps that are easy to miss.
The Gene Name Problem and Auto-Conversion Risks
One well-documented hazard of Excel-to-CSV conversion is Excel's automatic type inference, which converts certain strings to dates or numbers before you export them. A gene name like MARCH1 (Membrane Associated Ring-CH-Type Finger 1) is converted to the date March 1st when entered into a cell without explicit text formatting. The gene name DEC1 becomes December 1st. The gene SEPT2 becomes September 2nd.
A 2016 study published in Genome Biology found that roughly 20 percent of genomic research papers that included supplementary gene lists in Excel files had gene names incorrectly converted to dates. The problem was systematic enough that the HUGO Gene Nomenclature Committee renamed several genes in 2020 specifically to avoid Excel auto-conversion: MARCH1 became MARCHF1, and SEPT1 became SEPTIN1, among others. This is a case where auto-conversion in a spreadsheet application directly influenced the naming conventions of an entire scientific domain.
The same risk applies to any data containing values that Excel might interpret as dates, scientific notation, or leading-zero numbers. A product SKU of "007" becomes "7." An entry of "1E5" becomes 100,000. These conversions happen silently in Excel and persist into the CSV export if the cells were not pre-formatted as text before data entry.
Character Encoding and Delimiter Variants
CSV files do not have a standard encoding declaration. Most tools produce UTF-8, but Excel traditionally saved CSV files with a Windows-1252 encoding for certain regional characters, which causes display problems when the file is opened in systems expecting UTF-8. Excel 365 improved this behavior, but older versions and some regional configurations still produce encoding surprises.
European locales add a complication: in countries where a comma is used as the decimal separator, Excel uses semicolons as the field delimiter instead of commas. A CSV exported from a German-locale Excel installation is technically semicolon-separated, and standard CSV parsers expecting commas will misread the structure. The name "CSV" conceals the fact that the actual delimiter depends on the regional settings of the exporting system.
Conclusion
The conversion from Excel to CSV separates a document designed for human reading from data designed for system processing. The conversion strips away everything Excel added on top of the raw information and returns what was there before the formatting began. That stripped form is often exactly what the next step in the workflow requires.
ToolHQ's Excel to CSV converter handles the conversion when the destination is a system, a script, or a pipeline rather than another spreadsheet user.
Frequently Asked Questions
What happens to Excel formulas when you export to CSV?
Formulas become their last calculated result. The CSV contains the number or text that the formula produced, not the formula itself. The receiving system gets the value but has no access to the underlying logic.
Can you export multiple Excel sheets to a single CSV?
No. CSV is a single flat table. Each sheet must be exported to its own CSV file separately. This is a structural limitation of the format, not a software restriction.
Why does my CSV look different in different programs?
Date and number formatting is locale-dependent. The same Excel value can export as different text representations depending on system regional settings. Always check date formats after conversion.