Common CSV Problems: Delimiters, Quoting, and Encoding Explained

Everything Bunched Into One Column

If an entire row appears to load into a single column instead of splitting properly, the file most likely uses a different delimiter than the comma a standard viewer expects — semicolons are especially common, since some regional versions of Excel export CSV files with semicolons by default (particularly in locales where a comma is already used as the decimal separator). Checking the raw file in a plain text editor for semicolons or tabs usually confirms this quickly.

Rows With the Wrong Number of Columns

This usually means a field containing a comma wasn't properly wrapped in quotes when the file was generated — the parser splits that field at the internal comma, and the row ends up with one extra column it shouldn't have. This is a problem with how the source file was written, not something a viewer can reliably guess around after the fact.

Garbled or Mismatched Characters

Accented letters or non-Latin characters rendering as strange symbols is almost always an encoding mismatch — the file was saved in one character encoding, but is being read assuming a different one. UTF-8 is the safest modern default; if a file displays incorrectly, trying to re-open or re-save it explicitly as UTF-8 often resolves it.

An Extra Character at the Very Start of the File

A stray-looking character at the beginning of the first cell is often a byte-order mark (BOM) — a few invisible bytes some tools (Excel especially) add to signal UTF-8 encoding. It's harmless for Excel, which expects it, but can appear as unwanted extra characters in tools that don't specifically account for it.

A Quoted Field That Won't Close

If a value contains an unescaped double quote — a straight quote that wasn't doubled as RFC 4180 requires — a naive parser can misinterpret it as the end of the field, then treat everything afterward, potentially including subsequent commas and rows, as part of one giant runaway field. This usually traces back to how the source file was generated rather than something fixable while just viewing it.