Tableau. Published data source. One certified orders model for every dashboard.

What this means for your store

A published data source on Tableau Cloud or Server is a governed connection plus calculated fields, hierarchies, and default filters - built once, consumed by many workbooks. When US merchandising defines net revenue differently from the agency ROAS sheet, leadership gets two truths. Centralising orders, refunds, and campaign mapping in one published source locks shared logic for Net Revenue, New vs Returning, and Contribution Margin.

Scenario on a real storefront

Allbirds publishes Ecommerce Core from a BigQuery semantic view. Marketing, merchandising, and finance connect live instead of embedding their own order SQL:

Tableau Desktop → Data → Publish to Server
  Name: Ecommerce Core (certified)
  Project: Marketing Analytics
  Extract: nightly Hyper refresh at 05:00 UTC
  Included:
    - orders_fact (grain: order_id)
    - campaign_dim (GA4 + Ads mapping)
    - Calculated: Net Revenue, CAC payback flag

Workbook authors: Connect → Server → Ecommerce Core
  Do not embed a second copy of orders SQL

// Server → Data Sources → Set certification + owner
// Schedule: fail alert to #data-alerts if refresh < 95% rows

What to do next

• Certify only after finance signs off on revenue, refund, and margin rules.
• Restrict “download workbook data source” for casual users if they keep forking definitions.
• Pair with Data Quality Warnings when the GA4 export lands late - merchandising should see the delay before pacing spend.

Bottom line

Published data sources cut dashboard sprawl and metric debates. Publish one certified ecommerce model, refresh on a schedule, and point every marketing workbook at it.