Context
Pittol initiated a web analytics restructuring project following its migration from VTEX CMS to VTEX IO. While the transition brought infrastructure improvements, it also exposed critical gaps in data collection and organization.
With inconsistencies in GA4, attribution errors, and issues across media integrations, the team struggled to accurately measure performance and confidently optimize campaigns.
Challenge
The main issue was data reliability. A misconfigured GA4 Measurement ID compromised data collection, leading to inconsistent metrics.
Before the fixes, the internal dashboard reported 16.91% more transactions and 11.30% more revenue than GA4, while unattributed traffic reached 12.1%—making it difficult to identify the true source of conversions.
The situation was further impacted by infrastructure issues, including UTM loss in email marketing campaigns, duplicated transactions in GA4, outdated conversion tracking in Google Ads, and an overloaded GTM container filled with obsolete tags.
Strategy
Metrics Stage led a full restructuring of the web analytics foundation to restore data accuracy.
The project included fixing GA4 configurations, implementing server-side GTM to mitigate browser tracking restrictions, and cleaning up the GTM container—removing outdated tags and reorganizing the tracking structure.
Additionally, Enhanced Conversions and dynamic remarketing were activated in Google Ads, along with the implementation of transaction ID controls to prevent duplicate purchases in GA4.
Results
The new architecture restored predictability and trust in marketing data.
Key Impacts:
- Revenue discrepancy reduced to 3.84%
- Unattributed traffic decreased from 12.1% to 3.2%
- Elimination of most duplicate transactions
- 40% increase in Google Ads ROAS
Conclusion
With a more structured and reliable data infrastructure, Pittol can now operate marketing campaigns with greater confidence and efficiency.
This project highlights how tracking quality directly impacts marketing performance and the ability to make data-driven decisions.