Basket facts

Basket facts hold the observation-level session and conversion events for a retail snapshot. A fact can carry an observation date, source reference, session count, shopper count, conversion count, target completions, store exposure and consent basis. The Retail Analyst uses those values to understand what contributed to a snapshot’s reported outcome. The source reference helps distinguish an imported observation from a second copy of the same event.

The collection workflow operates on one snapshot at a time. It records an operation key, a run and numbered steps, then leaves a state history entry for the transition. This matters when someone repeats a collection attempt after an interruption: the team can identify the original operation instead of treating every click as new evidence.

• Scope. Every fact belongs to a parent snapshot, so dates and store context remain together.
• Outcome. Pending, accepted and rejected observations can be reviewed separately.
• Provenance. Source reference, consent basis and operation history explain where a fact came from.

Agree what counts as one session, one shopper and one conversion in the source system. The team should use the same interpretation when a fact is inspected and when a metric definition is approved. Document how a source reference is produced and who can change the outcome of an observation. The consent basis should be available for the reviewer without exposing more shopper detail than the measurement needs.

Agree the handling of a repeated source event. An analyst should be able to show the earlier operation key and its result. A corrected observation should retain the reason for the change in the surrounding review and state history. The purpose is to keep a conversion rate explainable after the reporting period closes.

For a dated store snapshot, collect a batch of consented facts. Compare the source references, observed dates and session counts with the source window. Mark a disputed event rejected and keep it visible to the reviewer. The reviewer examines the accepted scope before approving the snapshot revision. If the same collection request is sent again, confirm that the operation key points back to the recorded run and that the accepted facts are not duplicated.

Next compare a fact’s session and conversion counts with the snapshot-level fields. A discrepancy should lead to a specific source reference and observation, not an unexplained dashboard adjustment. That review is the bridge between the underlying events and the result people act on.

• Basket facts awaiting review — Inspect pending observations with their parent snapshot and consent context.
• Accepted basket facts — Trace the events used in a reviewed measurement.

Follow collection and review and inspect result calculation.