No Buffering: How Comcast's Data Team Eliminated Financial Reporting Delays

At Comcast, a three-person Business Intelligence (BI) team was responsible for delivering critical financial data to over 1,300 users—but their legacy Oracle Business Intelligence (OBI) system struggled to keep up. Data refreshes happened only once per day, leaving analysts with stale information. Worse, even simple changes to reports took developers up to two weeks to implement due to complex ETL workflows and multi-layered dependencies.

Prabha Kuppusamy, Senior BI Developer at Comcast, recalls:

*"Our overnight process took 8–9 hours just to deliver day-old data. Real-time reports slowed to a crawl during peak hours, and users resorted to extracting data just to pivot it in Excel."*

The team faced two major hurdles:

1. Delayed Data: A 24-hour lag in reporting made it impossible for accounting teams to reconcile finances during the critical "soft close" period (22nd–month end).
2. Slow Development: Building new reports required coordination across Informatica, DAC, data warehouses, and OBI, dragging out even minor updates.

The business demanded a solution: refresh key data every 15 minutes while merging two disparate datasets (general ledger and subledger journals) into a single operational view—without sacrificing performance on 400+ million rows of data.

Comcast turned to Incorta, a modern analytics platform, to reengineer its data pipeline. Key innovations included:

• Eliminating Heavy Transformations: Instead of preprocessing data, Incorta ingested raw tables directly, reducing ETL workflows from 150 to just 10 base tables.
• Smart Incremental Loading: Conditional polling refreshed subledger data only when new transactions appeared, while general ledger data updated every 15 minutes.
• Optimized Query Performance: By shifting complex joins to Spark-powered materialized views and tuning extraction filters, the team cut load times dramatically.

Prabha highlights a breakthrough:

"Moving from ‘max refresh time’ queries to ‘greater than’ filters reduced incremental load times by

‍

The impact was transformative:

• 15-minute data refreshes achieved for general ledger reporting, down from 24 hours.
• Development time dropped 88%—from 8.5 days per project to just 1 day.
• Report rendering accelerated from 22+ minutes to seconds, even for 80+ million-row datasets.
• Fixed-asset module loads shortened from 17 hours to 80 minutes, enabling multiple daily updates.

User feedback was resoundingly positive:

"A survey showed most employees saved 2–8 hours monthly—no more waiting for reports or manual reconciliations."

By reimagining its data architecture, Comcast’s small BI team didn’t just meet SLAs—they unlocked real-time decision-making for finance, proving that agility and speed are possible even at enterprise scale.