CSV and Data Glossary

Plain-English definitions for the terms that come up constantly when working with CSV and tabular data.

CSV (Comma-Separated Values)

A plain-text format for tabular data, one row per line, values separated by commas. The most widely supported format for exchanging data between unrelated systems.

Delimiter

The character that separates values within a row, a comma in most CSVs, but semicolons and tabs are common too, especially from European-locale exports.

Header Row

The first row of a CSV, containing column names rather than data. Not every CSV has one, always check before assuming row 1 is data.

Character Encoding

The scheme that maps bytes to characters. A mismatch between the encoding a file was written in and the one it's read as produces garbled text (mojibake).

UTF-8

The modern, near-universal character encoding standard. The safest default for saving and reading CSV files, correctly handles accented characters, symbols, and emoji.

BOM (Byte Order Mark)

A few invisible bytes some tools add at the start of a UTF-8 file to signal its encoding. Excel wants it to detect UTF-8 correctly; some other tools misread it as a stray character.

Quoting

Wrapping a value in double quotes so a delimiter character inside it (like a comma in "Chicago, IL") isn't mistaken for a column separator.

Escaping

Representing a special character (like a literal quote mark) inside a quoted value, typically by doubling it ("" for a single ").

Row

One horizontal line of data in a table, one record. In a customer CSV, each row is usually one customer.

Column

One vertical field across all rows, sharing a single data type and meaning. Also called a field or attribute.

Schema

The structure of a dataset, which columns exist, in what order, and what type of data each holds.

Data Type

The kind of value a column holds, text, integer, decimal, date, boolean. CSV itself has no data types, everything is text until a tool parses it.

Null Value

The absence of a value, distinct from an empty string or zero. In CSV, best represented as a truly empty cell rather than text like "N/A" or "null".

Whitespace

Spaces, tabs, or invisible characters around a value. Untrimmed whitespace ("New York " vs "New York") causes values that look identical to be treated as different.

Deduplication

Removing rows that represent the same real-world record, either exact duplicates or near-duplicates caught via fuzzy matching.

Fuzzy Matching

Comparing values by similarity rather than exact equality, using a similarity score and threshold, so "Jon Smith" and "John Smith" can be recognized as the same entity.

Join

Combining two datasets by matching rows on a shared key column, rather than simply stacking rows together.

Inner Join

A join that keeps only rows with a match in both datasets. Rows without a match on either side are dropped.

Left Join

A join that keeps every row from the first (left) dataset, filling in blanks where there's no match in the second.

Outer Join

A join that keeps every row from both datasets, matched where possible, blank where not, sometimes called a full join.

Pivot Table

A summary table that cross-tabulates data across two dimensions, rows and columns, computing an aggregate (sum, count, average) for each combination.

Group By

Summarizing data into one result per group, a simpler, one-dimensional version of a pivot table.

Aggregation

A function that reduces many values into one summary number, sum, count, average, min, or max.

Flat File

A file storing data in a single table with no relationships to other files, CSV is the most common flat-file format.

Nested Data

Data with hierarchical structure, values inside values, like an order with a list of line items. Not naturally representable in flat formats like CSV.

JSON (JavaScript Object Notation)

A lightweight, human-readable format for nested/hierarchical data, the standard for web APIs and configuration files.

XLSX

The modern Excel spreadsheet file format, supports multiple sheets, formulas, and formatting, unlike plain CSV.

TSV (Tab-Separated Values)

Like CSV, but using tabs instead of commas as the delimiter, useful when the data itself contains commas.

Parquet

A columnar storage format optimized for large-scale analytics, much smaller and faster to query than CSV at scale, but not human-readable.

ETL (Extract, Transform, Load)

The process of pulling data from a source, cleaning/reshaping it, and loading it into a destination system, the general pattern most CSV workflows follow.

Want to see these concepts in action?

Browse the full guides library for step-by-step walkthroughs, or jump straight into your data.

Browse All Tools

Turn this into a saved workflow

Create a free account to save the steps from this guide as a reusable workflow and re-run it on any file, from any device.

Sign in for free