Tag evaluation runs
Tags are (re)computed by a tag evaluation run — manual, scheduled, or via API. A run evaluates the active rules (or a specific subset), then creates assignments for newly matching entities and removes assignments that no longer match.
Because a run reconciles assignments (adds and removes), tags stay in sync with your
data over time — an entity that stops matching loses the tag on the next run.
Complete example
Tag inventory items projected to run out within 7 days.- A single metric condition (
days of stock ≤ 7) under anANDgroup. - Scope is just
entityType: INVENTORY_ITEM— no brand/taxonomy restriction, no propagation (defaults to entity-only). - On each evaluation run, matching inventory items receive the tag; items that recover (more than 7 days of stock) lose it.
CASCADE_DOWN propagates the tag from
each matching product to its variants.
Once entities are tagged, point alerts and
workflow triggers at those tags to monitor or
automate against them.

