Client confidential · Creator media and paid media
We rebuilt a failed data migration in BigQuery, expanded platform coverage, and added automated quality monitoring.
Situation
A creator media platform publishes creator content and runs paid media campaigns through creator profiles for major national brands. Its business depends on accurate performance data for every ad platform, creator, episode, and conversion.
When the client came to MetricMaven, a prior migration had stalled at roughly 30% completion. The new models did not work, the previous consultancy had walked away, and the old connector pipelines could not support newer platforms or custom attribution windows.
We audited the failed implementation and found it needed a complete rebuild. The engagement became a long-term partnership to rebuild the client’s analytics.
The problems went beyond an unfinished migration: 100+ undocumented data models, reports arriving 2 to 3 days late, 40+ hours of manual work each week, no 1-day view-first attribution, coverage limited to Facebook and Google, and roughly 85% data accuracy in a client-facing Bubble app.
What we did
We rebuilt the warehouse in four layers, combined Maven connectors with custom API integrations, expanded from 2 to 7 ad platforms, and added automated checks that alert the team in Slack, without interrupting reporting.
We rebuilt the data in BigQuery, Google’s data warehouse, in four layers: raw data with standard names, cleaned models by subject, fact and dimension tables for analysis, and reporting tables that power the client’s Bubble app. Business logic moved into reusable, documented dbt code with automated tests.
Where off-the-shelf connectors fell short, we combined Maven’s connectors with custom Google Cloud integrations. Standard metrics came through Maven connectors; custom APIs handled advanced attribution and conversion needs.
We expanded coverage from Facebook and Google to seven channels: Facebook, Google, TikTok, Pinterest, LinkedIn, Snapchat, and Roku. Automated checks compare live numbers against validation benchmarks and alert the team in Slack before a discrepancy reaches stakeholders.
System view
A failed migration required a full warehouse rebuild, broader platform coverage, and monitoring that could protect reporting continuity.
Facebook, Google, TikTok, Pinterest, LinkedIn, Snapchat, and Roku
Standard connectors combined with Google Cloud integrations
Raw, cleaned, and reporting-ready layers with reusable business logic
Reporting tables for a client-facing application without interrupting reporting
Result
Accuracy, processing time, platform coverage, and weekly team capacity all improved.
Every engagement starts with a tracking audit: what you track today, what’s missing, and what to fix first.
