Skip to content
Docs

Concept layers

A term can mean one thing company-wide and something more specific to one team. The glossary handles that with three tiers and a clear precedence, so every workspace gets the most specific definition that applies to it without anyone having to duplicate the rest.

The three tiers

  • Standard (global): A standard set of marketing concepts maintained by Adverity. Shared across every tenant; you never edit these directly

  • Organization (tenant-wide): Your company’s shared vocabulary, applied across all your workspaces. This is where company-agreed metrics live

  • Workspace: Definitions scoped to a single workspace, for when one team’s usage differs from the company norm

Precedence

When the same term exists at more than one tier, Adverity Atlas resolves it most-specific-first:

  1. Workspace: If the current workspace defines it, that wins

  2. Organization: Otherwise the tenant-wide definition

  3. Standard: Otherwise the Adverity-maintained baseline

A workspace only ever sees one effective definition per term, the winner of that resolution. Tiers it does not override are inherited as-is, so you define only what genuinely differs.

Note

Example: “Conversion” is defined as a standard concept. Your organization overrides it to mean a qualified lead. One workspace, running e-commerce, overrides it again to mean a purchase. Analysts in that workspace get “purchase”; everyone else in the tenant gets “qualified lead”; a brand-new tenant with neither override gets the standard.

Concepts follow the workspace’s data

A workspace only shows the concepts its Chat can use, at every tier. A concept is shown in a workspace when at least one of these is true:

  • Someone created it by hand: with Add concept, as a workspace override, or by promoting a workspace concept to the organization tier

  • A column classified to the concept, or to a standard or organization concept of the same name, sits in a data source actively bound to the workspace, on a column, table, and collection the binding does not hide

  • An active measure for the concept runs on a table in a data source actively bound to the workspace, and the binding does not hide that table or its collection

  • For a calculated concept, such as CTR, a derived measure has been proposed on such a table, even while it is still a draft

A concept that meets none of these is left out of the workspace’s glossary, glossary search, and the terms Chat resolves.

Note

Support is matched by name across the standard and organization tiers, so an organization override of a standard concept stays visible wherever columns are classified to the standard concept it overrides. A workspace concept counts its own columns, but another workspace’s concept of the same name does not count.

Promoting across tiers

Concepts move up the tiers deliberately, not by accident:

  • Promotion: A useful workspace concept can be promoted to the organization tier so every team gets it

  • Review before rollout: Promotion files an approval request rather than applying immediately, so shared vocabulary is reviewed before it becomes company-wide

Activity and revisions

Glossary changes are tracked, who changed what, and when, and concepts keep a revision history. A definition that drifted can be compared against an earlier version field by field, so you can see exactly how it changed over time.

A concept’s History card and the Glossary Activity page show only what the current workspace can see:

  • Classification changes appear only for tables the workspace can query: the data source is bound to the workspace, and neither the table nor its collection is hidden

  • Changes to a concept’s column mappings appear when they apply tenant-wide or to this workspace, not when another workspace made them

  • Changes to the concept itself, such as its definition or status, always appear

  • Older classification entries that did not record their table are left out

On Glossary Activity, the change-volume heatmap for the last six months, the feed of changes, and the panel of most changed concepts over the last 30 days all apply this filter.

Open an entry from a concept’s History card, or from Glossary Activity, to see its detail. For a classification change, the detail lists the affected Columns, grouped by table:

  • Each table shows its data source, its collection, and the table name. A data source that is not bound to the current workspace is marked Not in this workspace

  • Click a column to open its field profile. Columns that no longer exist, that the workspace hides, or whose data source is not bound to the workspace, are listed without a link

The detail also names what triggered the classification change:

Trigger

Meaning

Classify run

A classification run assigned or changed the columns

Profiling completed

Classification followed a completed profiling run

Glossary reclassification

A glossary change caused columns to be reclassified

Approved manually

Someone approved the classification

Rejected manually

Someone rejected the classification

What’s next?

Was this page helpful?