The Challenge
The "Slow Dashboard" Chokepoint
Sicar, a leading Point of Sale (POS) provider in Latin America, offered "Sicar X," a cloud-based solution boasting actionable reports and dashboards. However, a critical flaw was undermining customer satisfaction: reports could take minutes to load, severely limiting their usability. To compensate, Sicar was forced to restrict reports to only the last six months of data, diminishing their value and sparking customer complaints. The promise of "data-as-a-service" was being choked by slow queries.
The Cause
Data Silos and NoSQL Limitations
The Solution
A Unified BigQuery Reporting Engine
We executed a focused, 4-week data overhaul to create a single, high-performance source of truth.
Fivetran Automation: We deployed Fivetran as a managed data ingestion service, automating the transfer of data from all four sources into Google BigQuery.
SQL Transformation: We replaced thousands of lines of complex MongoDB aggregation code with streamlined SQL queries in BigQuery.
Medallion Architecture: We implemented a layered data model that first landed raw data, then applied transformations to create normalized analytics tables, and finally generated optimized reporting tables mirroring Sicar X's existing dashboards.
Our Architecture
The Results
90% Faster Queries, Unlimited Insights
The new architecture delivered a dramatic performance uplift.
Query Speed: Reports that previously took 2.5 minutes to load in MongoDB now generated in under 10 seconds in BigQuery—a 90% improvement.
Data Liberation: Sicar was able to remove the six-month data cap on reports, providing customers with complete historical visibility and unlocking new analytical capabilities.
Fivetran Commitment: Impressed by the speed of implementation and performance gains, Sicar signed an annual commitment to Fivetran, abandoning plans to build a self-hosted pipeline.