CSV: the 50-year-old format that refuses to die
No official standard, mangled by Excel, ambiguous about everything — and still the lingua franca of data. Why, plus the survival rules.
5 min read · Reviewed July 2026
CSV predates the personal computer, has no enforceable standard, and every tool interprets it slightly differently. It should have died decades ago. Instead it moves more business data daily than any format alive, because it has the one feature that beats all features: everything opens it. The fanciest data platform and a 1998 spreadsheet can exchange CSV. That universality is why 'export as CSV' will outlive us all.
The sharp edges
Commas inside values must be quoted, quotes inside quoted values must be doubled ('' style), and line breaks inside cells are legal — three rules that naive split-on-comma code ignores, which is why our converter has a real parser instead. Encodings bite too: Excel historically mangles accented characters unless the file carries a UTF-8 BOM. And the biggest one: CSV has no types. Everything is text.
That last point is behind the most famous CSV disaster genre: identifiers. ZIP code 01238 opens in Excel as 1238; a 16-digit account number gets rounded into scientific notation — silently, and saved back that way. Gene researchers literally renamed human genes (MARCH1 became MARCHF1) because Excel kept turning them into dates in published datasets. The format wasn't wrong; the type-guessing importer was.
The survival rules
Quote defensively, keep IDs as strings (import them as text columns, never 'General'), use UTF-8, and when converting from JSON — like the tool above does — let strings stay strings. If your data has nesting, CSV is the wrong container; flatten it deliberately or use JSON. Treat CSV as what it is: a brilliant, dumb pipe for tables. The dumbness is the feature.