A Digital-First Video Streaming Platform.
The Platform Delivers Films, Series And Live Events To Subscribers Across The Country On Mobile, Web And Connected Television. Content Catalogue And Viewing Hours Had Grown Rapidly, And With Them A Cloud Bill That Rose Every Month Without Anyone Able To Explain Why.
Engineering Optimised For Delivery Speed And Playback Quality, Which Was Correct For The Product But Left Cost Unowned. Finance Received A Single Monthly Invoice With No Breakdown By Feature, Title Or Subscriber, And Could Neither Forecast It Nor Challenge It.
Problem Statement
Despite Strong Subscriber Growth, The Platform Struggled With:
- 01
No Unit Economics
Nobody Could State What Delivering One Streaming Hour Actually Cost The Business.
- 02
Unowned Spend
Engineering Made Cost Decisions Daily Without Ever Seeing A Cost Figure.
- 03
Unpredictable Invoices
Finance Could Not Forecast Next Month's Bill Within Any Useful Margin.
Proposed Solution
The Engagement Built A FinOps Practice Around:
- Established Tagging And Allocation So Every Rupee Maps To A Service, Feature Or Environment.
- Defined Unit Cost Metrics Reporting Spend Per Streaming Hour And Per Active Subscriber.
- Introduced Showback Reporting Giving Each Engineering Team Visibility Of Its Own Consumption.
- Optimised Storage Tiers And Data Transfer Paths Carrying The Largest Share Of Delivery Cost.
- Implemented Commitment Purchasing Against Measured Baseline Rather Than Peak Or Guesswork.
- Automated Non-Production Environment Scheduling And Anomaly Alerts On Unexpected Spend Movement.
Making Cost A Number Engineers Can See
Teams Now See What Their Services Cost Alongside The Performance Metrics They Already Watch. Optimisation Decisions Happen During Design Rather Than During A Quarterly Review, And Finance Forecasts From Unit Costs Instead Of Extrapolating Last Month.
Unit Cost Metrics Connect Cloud Spend To Streaming Hours, Making Efficiency Comparable Across Features And Releases.
Showback Reporting Gives Teams Their Own Numbers Without Charging Budgets Or Creating Internal Billing Disputes.
Commitment Purchasing Against Measured Baseline Captures Discount Without Locking In Growth That Never Materialises.
Anomaly Alerts Surface Cost Movement Within Days Rather Than At The End Of A Billing Cycle.
Result :
Spend That Tracks Subscribers, Not Surprises
Cloud Cost Per Streaming Hour Now Falls As The Platform Grows Rather Than Rising With It, And Finance Forecasts Within A Narrow Margin. Engineering Teams Raise Cost Questions During Design Without Being Asked To.
Lower Cost Per Stream Hour
Forecast Accuracy
Savings From Commitments
Day Anomaly Detection
Lessons Learned
The Engagement Highlighted Three Lasting Takeaways:
-
Unit Cost Beats Total Cost
A Rising Bill Is Healthy If Cost Per Subscriber Is Falling.
-
Show Before Charging
Visibility Changed Behaviour Well Before Any Budget Was Reallocated.
-
Commit To The Baseline
Discounts On Capacity You Never Use Are Not Savings.
TECHNOLOGIES - TOOLS USED
Cost Management Runs On Allocation Tagging And Reporting Tooling Linked To Product Usage Metrics, Producing Unit Cost Views Alongside Standard Billing. Commitment Planning, Storage Lifecycle Policies, Environment Scheduling And Anomaly Detection Operate Continuously, With Showback Dashboards Published To Every Engineering Team.
- Cost Allocation & Tagging
- Unit Economics Reporting
- Showback Dashboards
- Commitment Planning
- Rightsizing Analysis
- Storage Lifecycle Policies
- Data Transfer Optimisation
- Environment Scheduling
- Anomaly Detection
- Budget & Forecast Tooling
- Usage Telemetry
CONCLUSION
Cloud Cost Is An Engineering Outcome Reported As A Finance Problem. Allocating Every Rupee, Expressing Spend Per Streaming Hour And Giving Teams Their Own Numbers Turned Cost Into Something Engineers Optimise By Default, So Growth Now Improves Margin Instead Of Quietly Eroding It.
