beginner 3 min answer

A 60-person company is told it needs a business glossary. Twelve people work with data and there are about 25 terms that anyone argues about. A glossary tool costs a subscription and a week of setup. What does the tool buy over one spreadsheet with a named owner, and when does the spreadsheet win?

business-glossarypragmatismlinkagedrifttooling
Show the full answer Hide the answer

What is gained

A glossary tool buys exactly two mechanisms that a spreadsheet cannot provide, and everything else it offers is presentation.

Linkage. The term "completed order" points at the specific tables, columns and view that implement it. That turns a definition into something a reader can verify in one click, and it makes the question "has this drifted?" answerable by machine rather than by interview.

Change notification. When the implementing SQL changes, or when the definition changes, the tool can tell the people who depend on it. A spreadsheet changes silently, and nobody rereads a spreadsheet.

What is paid

A subscription, a week of setup, and a quieter cost: the harvester will create several hundred stubs from column comments and table names, coverage will read as 90-something percent, and the 25 terms that actually matter will be as wrong as they were before. The number on the dashboard goes up while the problem is untouched. Teams then spend a quarter curating stubs nobody asked about, because the tool makes incompleteness visible and prioritisation invisible.

The spreadsheet's hidden advantage is the mirror image: it has no illusion of coverage. Twenty-five rows look like twenty-five rows.

When the cost becomes visible

Not at 25 terms. At 25 terms, maintenance is about 20 minutes of review each per quarter — roughly 8 hours a quarter — and that bill is identical in either tool. Definitions drift because nobody reads them again, not because of where they are stored; expect somewhere between a third and half of unreviewed definitions to no longer match their SQL after a year, in a spreadsheet or in a platform.

The tool starts earning its subscription at three thresholds. Past roughly 60 to 100 terms, where no one person can hold them and search matters. When more than one team owns definitions, so a disagreement needs a recorded owner and an escalation route rather than a conversation. And when you want the linkage to be binding — a schema change to a column behind a governed term failing someone's build, which is a genuine control and the only version of a glossary that changes behaviour rather than describing it.

How to keep the option to reverse

Keep the 25 definitions in a plain text or CSV file in the repository next to the SQL that implements them, with one named owner in the header. It is diffable, reviewable in a pull request, costs nothing, and imports into any tool later. The decision you are deferring is the tool, not the discipline.

And build the one enforcement point now regardless: a single shared view per governed term that every dashboard reads. A definition that no tool queries is a document whichever product it lives in — the spreadsheet is not the weakness, the absence of an enforcement point is.

When this is the wrong answer

If the company is in a regulated activity where a supervisor can ask for the lineage of a reported figure, buy the tool on day one. The deliverable there is the evidence trail, not the definition, and reconstructing it from spreadsheet history after the request arrives costs far more than the subscription.