A shared glossary can prevent surprisingly expensive AI mistakes
Define ambiguous company terms so AI assistants and teammates interpret metrics, customer stages and product language consistently.
One team says “active customer” and means a paying account. Another means an account with recent usage. An AI assistant summarising both teams can produce a confident comparison between different things.
A shared glossary makes those differences explicit.
Define the terms that affect decisions
Start with words that appear in reports, customer promises or project acceptance criteria. For each term, record its meaning, scope, owner and a concrete example. Add exclusions where confusion is likely.
For a metric, include the calculation and the reporting period. “Active” without a time window is not a usable definition. When two teams legitimately use different definitions, name both instead of forcing them into one vague compromise.
Keep local meanings visible
An assistant may know the ordinary meaning of a word and still miss your company's meaning. Attach the relevant glossary entries to analysis tasks and ask the assistant to flag undefined terms that affect the result.
Do not treat the glossary as proof that a number is correct. It defines what should be counted; the underlying data and calculation still need verification.
Maintain definitions with their uses
When a definition changes, record its effective date and identify the reports or workflows affected. Avoid silently comparing a new definition with historical figures produced under an older one.
Begin with ten disputed or frequently explained terms. Ask people from two different teams to apply the definitions to the same examples. If they get different answers, revise the entry until the ambiguity is visible and resolved.
A useful glossary earns its place by preventing misunderstandings in real work, not by documenting every acronym the company has ever used.