Customer Exports

Customer exports are scheduled file drops from the BI analytics model into customer-specific S3 buckets. They are intended for customers who need a daily data dump rather than direct Snowflake access.

Customer exports are configured per engagement, not self-service. The brands, tables, format, schedule, retention, and delivery method are agreed upon during onboarding.

Delivery Contract

Contract What to Expect
Cadence Daily file drops.
Freshness Delivered within 24 hours of the scheduled run.
Source Curated tables from the ANALYTICS schema.
Scope Filtered to the customer’s configured brands and locations.
Format Customer-specific; Parquet is the standard format for large exports.
Destination A dedicated customer S3 bucket under the analytics-export/ prefix.
Retention Customer-specific; the standard retention window is 30 days.

Each export is a point-in-time dump for that run. It is not a streaming feed, and it should not be treated as real-time data.

Table Coverage

Customer exports use the same public analytics tables documented in the Data Dictionary. A customer export may include all tables or a configured subset.

Commonly exported tables include:

Table Typical Use
DIM_DATES Shared calendar for date-based reporting.
DIM_LOCATIONS Location metadata and filtering attributes.
DIM_EMPLOYEES Employee attributes for recognition reporting.
FACT_REVIEWS Latest review state.
FACT_REVIEW_VERSIONS Review edit history.
FACT_EMPLOYEE_MENTIONS Employee mentions and manual review status.
FACT_CUSTOMERS_SAVED Saved customer revenue and visit outcomes.
FACT_INTERACTIONS Guest-business interactions and response metrics.
FACT_MESSAGES Message-level interaction detail.
FACT_TRANSACTIONS Transaction-level POS data.
FACT_TRANSACTION_ITEMS Transaction line items.
FACT_LOCATION_DAILY_STATS Pre-aggregated location/day performance metrics.

Consumption Notes

  • Treat each file drop as the delivered dataset for that run.
  • Use table primary keys from the Data Dictionary for deduplication or upserts.
  • Date and timestamp fields are in UTC.
  • Daily rollups use date_id as the reporting date.
  • Customer identifiers are opaque IDs, not customer names or contact details. The same person is not deduplicated across different POS or platform sources.

Escalation

Escalate if an expected export is missing more than 24 hours after its scheduled run, or if a file is present but cannot be read. Include the customer name, export date, table name, and file path when reporting an issue.