Optimizing Data Integrity for Moodiday.com

Project Overview

Moodiday.com is a cannabis e-commerce platform serving regulated product catalogs. They needed to improve data integrity and analytics reliability across PostgreSQL databases hosted on Azure and DigitalOcean to keep reporting accurate and operations stable.

The Challenge

Data discrepancies appeared in analytics due to a complex join between pages and widget_in_view. A LEFT JOIN on non-unique identifiers inflated row counts and flattened the dataset, which compromised reporting accuracy. Because the ingestion pipeline had already introduced inconsistencies, the team needed a forward-looking fix that would prevent the issue from recurring while improving performance.

Our Solution

We redesigned the joining strategy by introducing a unique session/page-load identifier, ensuring each widget_in_view event could be reliably tied to a specific page load. This preserved event relationships without inflating the dataset and created a foundation for dependable reporting.

Key workstreams included:

• Auditing ingestion and query logic across the data pipeline
• Designing a stable identifier to normalize event relationships
• Recommending schema and query improvements to reduce duplication
• Aligning reporting structures with compliance expectations

Outcome

While historical inconsistencies could not be fully repaired due to prior ingestion constraints, the new model established a clean forward path. Moodiday gained clarity on the root cause, a scalable approach to session-level data integrity, and a stronger foundation for analytics and compliance reporting going forward.