Northstar Travel Group

Northstar Travel Group is the B2B media company behind Travel Weekly, Phocuswire, Business Travel News and around twenty other travel-industry brands, and I was their fractional data engineer: first preserving a decade of analytics history ahead of Google's shutdown deadline, then building the BigQuery warehouse and the event-registration attribution their reporting runs on.

Key Takeaways

  • Preserved a decade of Universal Analytics history for 21 properties ahead of Google's deletion deadline, after evaluating six migration paths and choosing the one with the most usable data structures.
  • Built the consolidated BigQuery warehouse that turned twenty-one nested GA4 exports into typed, pre-aggregated tables an analyst and Tableau can actually use.
  • Tuned the warehouse for cost at volume, with partitioning and clustering choices tested on sample datasets before committing.
  • Repaired event-registration attribution across domains, so a visitor keeps one identity from a marketing site into a registration platform.
  • Reconciled the warehouse against GA4's own interface, column by column, before asking the reporting team to trust it.
  • Documented the tracking architecture as a flowchart and put warehouse change history in version control, so onboarding a new event site became a single step.

Organization

Northstar Travel Group publishes Travel Weekly, TravelAge West, Business Travel News, Meetings & Conventions, Phocuswire, Phocuswright and roughly fifteen more travel-industry brands from Rutherford, New Jersey, at around 500 employees and $100M in revenue. Its events business runs on third-party registration platforms across separate domains.

Challenge

For a media company that sells on audience, losing pre-GA4 history means losing the year-over-year baseline. At the same time, event marketing spend could not be attributed because visitors crossing from marketing sites to registration platforms lost their identity at the domain boundary, and reporting was assembled by hand across dozens of properties.

Solution

  • Historical Universal Analytics data extracted to BigQuery for 21 properties before the cutoff.
  • A consolidated warehouse unioning every brand's GA4 export, flattening event parameters into analyst-friendly columns, refreshed nightly with status tables for freshness.
  • Six Tableau-facing views at daily and monthly grain across three levels of attribution detail.
  • Cross-domain tracking corrected across all event domains, with attribution views rebuilt to match visitors reliably.
  • One measurement architecture: a single property with multiple streams, standardized event names, and GA4 form tracking, documented visually.
  • A vendor audience-tracking implementation completed to specification and verified with the vendor's team.

Results

  • 21 properties of pre-GA4 history preserved.
  • Roughly 98% of event campaign spend previously unattributable became measurable.
  • 20 to 40 hours a month of reporting assembly recovered.
  • 50 to 90% warehouse query cost avoided through partitioning and clustering.
  • Historical attribution refreshed every 12 hours.