How a Leading US Telecom Provider Cut Big Query Costs by 52% Through AI-Driven FinOps Transformation

Problem Statement

A leading US-based broadband and internet services provider scaled rapidly on Google Cloud Platform's Big Query — but without the governance, design standards, or cost controls to match. The result was a fragile analytics environment plagued by un checked cloud spend, fragmented data infrastructure, and recurring inefficiencies. Polestar Analytics was engaged to transform the Big Query ecosystem through large-scale optimization, AI-driven automation, real-time observability, and a sustained FinOps discipline.

Key Challenges

• Table & Query Inefficiencies No partitioning, clustering, or search indexing; inefficient views and query patterns; underuse of materialized views.


• Cloud Cost & Financial Leakage Uncontrolled byte scans and slot consumption; large volumes of stale, unused datasets; no automated alerts or guardrails.


• Fragmented Analytics Consumption Dashboards running with no active viewers; no prioritization by business criticality; historical data stuck in staging tables.


• Governance & FinOps Maturity Gaps Knowledge gaps in cost-efficient practices; manual processes and delayed approvals; weak governance controls.

Solution Implemented

• Optimize — Audited the most expensive tables, views, and queries. Applied partitioning, clustering, search indexing, and ETL refactoring to cut byte scans and slot consumption at the source.


• Automate — Built a Bulk Migration Framework that reads optimization rules from a central config table and deploys changes across all projects via a single stored procedure execution — replacing hours of manual DDL work.


• Intelligentize — Deployed a Vertex AI model (Gemini 2.5 Flash) that reads from Inventory Tables, analyzes expensive queries, and generates recommendations for partitioning, clustering, and search indexing — feeding them directly back into the Bulk Migration Framework to create a closed-loop optimization cycle.
• Monitor — Delivered a centralized observability dashboard giving leadership and engineering teams a live view of cost, usage, performance, and optimization health —with spend breakdowns by project, user, and workload.