Data Dictionary

Short answer: A data dictionary documents data about data, often called metadata. For each field, such as customer status or order date, it records what the field means and its format. It also lists the values the field may contain, which system it comes from and who is responsible for it. Database software also keeps its own internal technical catalog, sometimes called a data dictionary. In business use, though, the term usually means documentation that people can read.

A business glossary is the closest related tool, and people often confuse the two. A glossary defines business terms, such as active customer, in language the whole company uses. A data dictionary sits closer to the fields themselves. It explains where each system stores that term, in what format and with which allowed values. A dictionary helps new analysts, report builders and migration and integration teams. It also helps AI tools that need to read company data correctly.

A practical approach starts with the tables and reports people use most often, not every field in every system. Software can usually pull technical details, such as field names and types, straight from the databases. The people who use the data add its business meaning, allowed values and owner. Entries link to glossary definitions where they exist. The team stores the dictionary where people can find it and keeps a dated record of each change. It updates the dictionary as part of the change process whenever someone adds or redefines a field.

An operations dashboard and a sales dashboard both show a field called customer count, at 3,400 and 2,100. While writing the dictionary entries, the team finds that the first figure counts every account with an account code. The sales figure counts customers who ordered in the last 12 months. That means 3,400 minus 2,100, or 1,300 accounts, have not ordered in a year. The team renames the fields "registered accounts" and "active customers", and the weekly operations dashboard uses those labels.

Shared meaning reduces reporting errors and the time spent explaining numbers. A dictionary speeds up onboarding and vendor projects. It supports data governance by recording owners and rules. It is also a practical step in preparing data for AI, since tools and people both need to know what each field means.

Dictionaries go out of date quickly unless updates are part of how the team makes changes. Very detailed dictionaries that nobody reads add little value. A purely technical dictionary does not help business users, so business definitions matter as much as formats. A smaller company does not need specialized software at the start. A well-kept shared spreadsheet can work.