Data infrastructure built around media decisions
The warehouse was not the finish line. It was the foundation for apps that helped people schedule broadcasts, target advertising, optimize campaigns, and evaluate results.
The project's specifics
Case study at a glance: Azati supported a 100-month data engineering engagement, more than eight years of continuous support, for a large US media and entertainment company. The work covered media data engineering, streaming analytics, broadcast scheduling, advertising optimization, data warehousing, ETL, historical modeling, database performance, and DB2 Warehouse migration across an environment containing approximately 90 TB of data lake storage and 300+ ETL processes.
The business problem: A large US media company needed a data-driven software ecosystem supporting decisions across streaming, broadcast, advertising, and audience operations, from content scheduling and premiere planning to targeted campaigns and post-campaign analysis.
Why this is hard: The engineering challenge was not hyperscale storage. It was operational complexity. The environment combined a hybrid data lake and application database with 300+ ETL processes supporting streaming, broadcast, advertising, and audience workflows. Keeping those processes reliable over a 100-month engagement required careful orchestration, historical data management, validation, performance tuning, and controlled infrastructure changes.
A change to one part of the environment could affect downstream analytical applications, historical models, or operational workflows. The platform therefore had to evolve without breaking the dependencies that connected data to business decisions.
What was at stake: The ability to turn large volumes of viewing, audience, advertising, and broadcast data into usable business decisions across millions of streaming subscribers, multiple TV channels, and streaming platforms.
Technology stack
IBM DB2 Warehouse
Netezza
Microsoft SQL Server
SQL
PL/SQL
Python
Node.js
MS SSIS
Prefect
dbt
Flyway
SAP PowerDesigner
Cognos
Angular
AWS S3
AWS EC2
IBM Object Storage
Kubernetes
Great Expectations
The project's technical reality
The client was not working with a clean-sheet cloud data platform. The environment combined established enterprise data technologies, historical datasets, analytical applications, and evolving infrastructure.
Azati therefore modernized parts of the data estate where it made sense, improving orchestration, data quality, database performance, historical modeling, and migration processes without assuming the existing environment could simply be replaced.
The technology landscape included IBM DB2 Warehouse, Netezza, SQL Server, SSIS, Prefect, dbt, Python, AWS, IBM Object Storage, and Kubernetes.
The challenge: connect media data to decisions at scale
Challenge 01
Multiple business lines, one analytical foundation
Streaming, broadcast, advertising, promotion, and audience operations generated different datasets and required different decision cycles. The platform needed to support all of them without turning every new analytical requirement into a separate data project.
#1
Challenge 02
Making historical media data usable at scale
Media analytics depends on more than the latest viewing or campaign data. Teams often need to understand what the audience, content, or campaign looked like at a particular point in time.
The environment included 300+ ETL processes operating across streaming, broadcast, advertising, and audience workflows, alongside approximately 90 TB of data lake storage and 30 TB of application data. The engineering challenge was keeping these workflows reliable as data sources, schemas, analytical requirements, and infrastructure evolved.
#2
Challenge 03
Modernization without discarding working systems
The analytical environment had to evolve while existing applications and downstream processes continued to depend on it. Database migration, validation, performance tuning, and data-quality controls had to be treated as production engineering work, not as a one-time warehouse implementation.
#3
Why the difficult part wasn't the data volume
A 90 TB data lake is only part of the engineering problem. In a mature media environment, complexity comes from the number of workflows that depend on the data, and from keeping those workflows reliable as sources, schemas, applications, and business requirements change.
In this engagement, more than 300 ETL processes connected data across streaming, broadcast, advertising, and audience workflows. Azati's role was to keep that ecosystem usable while evolving its orchestration, historical modeling, data quality, database performance, and infrastructure.
The result was not a reporting repository operated separately from the business. It was a connected set of data products supporting programming, audience targeting, campaign optimization, and performance analysis.
300+ ETL processes without separating data engineering from domain logic
The team worked across 300+ ETL processes supporting streaming, broadcast, advertising, and audience workflows, alongside a 90 TB data lake and approximately 30 TB of application data. The work included orchestration, historical modeling, SCD Type 1–3 strategies, data-quality controls, and database optimization.
Migration required engineering, not just a vendor tool
When the IBM DB2 Warehouse environment moved from Gen2 to Gen3, the team developed its own migration and validation mechanisms using Prefect and SSIS rather than relying solely on ready-made migration tooling.
Planning a streaming, broadcast, or advertising data-platform initiative?
Talk with Azati about the data architecture, engineering capacity, and modernization path your business actually needs.
Discuss your media data platform roadmapHow can a media company turn high-volume streaming and broadcast data into better content and advertising decisions?
A media company can do this by connecting audience, content, broadcast, and campaign data to analytical applications that support programming, targeting, inventory optimization, and performance analysis. The data platform must preserve historical context, process growing volumes reliably, and deliver outputs to the teams making operational decisions.
How Azati contributed to the media analytics platform
Keeping 300+ ETL processes operational over a long-running engagement required orchestration, data-quality controls, historical modeling, validation, and performance engineering as the surrounding data environment evolved.
Azati worked inside the client’s broader engineering organization, using the team’s existing processes while carrying long-term context across data, applications, and media workflows.
01
Streaming analytics platform and content performance optimization
Streaming platforms generate detailed behavioral data that can be used to understand audience engagement, measure content performance, and inform programming decisions.
Azati developed data and functional capabilities for an analytical application that brought together statistical data to help teams evaluate streaming performance and assess how programming decisions aligned with audience behavior.
Key capabilities:
- Streaming performance analytics
- Audience and viewing analysis
- Metric calculation
- Analytical dashboards
- Content-placement analysis
- Premiere optimization
- Targeted advertising analysis
02
Advertising analytics and ad inventory optimization
Advertising decisions become more effective when audience analytics, historical performance data, and campaign results can inform targeting and allocation decisions rather than relying only on broad demographic assumptions.
Azati engineered data and analytical capabilities for the client's targeted advertising application. The client reported a 17% increase in advertising revenue during the relevant period. While advertising-time volume remained steady, Azati presents this as an associated client outcome rather than a result attributed solely to the engineering work, as other commercial factors were not isolated.
Key capabilities:
- Audience-based campaign planning
- Targeted advertising
- Statistical campaign analysis
- Campaign optimization
- Historical-performance analysis
03
Broadcast scheduling and content placement
For television businesses, programming decisions have to balance expected audience performance with the constraints of the available broadcast schedule.
Azati developed and maintained data-driven capabilities for broadcast analytics. These capabilities gave programming teams a structured basis for evaluating schedules and anticipated audience performance.
Key capabilities:
- Broadcast schedule construction
- Content-placement optimization
- Potential-viewership forecasting
- Schedule-performance analysis
- Comparison of planned versus expected results
04
Targeted promotion across owned channels
Media companies also need to optimize how they promote their own programming and services. To help the client plan and optimize targeted promotional campaigns across its television channels, the Azati team engineered core application capabilities.
The application used audience and performance data to support decisions about which audiences to target and how internal promotional activity should be allocated. Together with the advertising capabilities, this extended the use of audience data beyond external advertising into the company's own content-promotion activities.
Key capabilities:
- Targeted promotional campaigns
- Audience analysis
- Campaign planning
- Promotion optimization
- Performance-based decision support
05
Cross-platform digital campaign performance analytics
Campaign planning is only one side of the process. Media organizations also need to understand what happened after campaigns were launched. The analysis supported campaign performance measurement across digital channels, television properties, and streaming platforms.
Azati helped build an analytical application for evaluating digital advertising campaigns across television and streaming properties. The application supported analysis of campaign performance across multiple digital channels, giving teams data for evaluating marketing activity and informing subsequent decisions.
Key capabilities:
- Digital campaign analysis
- Cross-platform performance measurement
- Campaign reporting
- Audience and engagement analysis
- Historical campaign analysis
06
Media data warehousing, ETL pipelines, and historical modeling
In media analytics, the question is often not only what happened, but what the audience, content, or campaign looked like when the event occurred. Using different SCD strategies allowed the platform to preserve changing attributes of entities such as subscribers and content when historical context mattered for analysis.
Azati engineered and maintained components of a hybrid analytical environment comprising a 90 TB data lake and approximately 30 TB of application data, with more than 300 ETL pipelines. The environment also grew by approximately 20 GB of data per day.
The data engineering work included:
- ETL development and optimization
- Analytical data marts
- Historical data processing
- Data transformation
- Database performance optimization
- SCD Type 1, 2, and 3 modeling
- Automated data-quality controls
- Integration across analytical data sources
07
Data infrastructure migration and performance engineering
The analytical infrastructure evolved during the engagement, including a migration from IBM DB2 Warehouse Gen2 to Gen3. Because the migration involved critical analytical infrastructure and there was no ready-made path that met the project's requirements, the team developed custom mechanisms for data transfer and validation using Prefect and SSIS.
Azati also worked on database and PL/SQL optimization for high-volume analytical workloads. This allowed the underlying analytical environment to evolve without treating the existing data estate as disposable.
Key capabilities:
- DB2 Warehouse Gen2 to Gen3 migration
- Custom migration workflows
- Data validation
- Prefect orchestration
- SSIS-based processing
- PL/SQL optimization
- Database performance tuning
- Historical-data preservation
Delivery & outcomes
| Area | What Azati delivered | Business / operational outcome |
|---|---|---|
| Data platform & ETL | 300+ ETL processes across a 90 TB data lake and ~30 TB application environment | Supported interconnected streaming, broadcast, advertising, and audience workflows within one data environment |
| ETL reliability & data quality | ETL optimization, orchestration, historical modeling, SCD Type 1–3, automated quality controls | Kept historical and analytical data usable as sources and business requirements evolved |
| Infrastructure migration | Custom DB2 Warehouse Gen2 to Gen3 migration and validation | Allowed critical analytical infrastructure to evolve without relying solely on standard migration tooling |
| Streaming analytics | Metrics, dashboards, content performance and premiere analysis | Programming teams could use audience data to evaluate content decisions |
| Advertising optimization | Targeted campaign planning, audience analysis and performance optimization | Client-reported 17% advertising revenue increase at comparable advertising-time volume |
| Broadcast & content | Schedule construction, content placement and viewership forecasting | Programming decisions gained a structured analytical basis |
| Historical analytics | SCD Type 1–3, historical models and analytical data marts | Preserved the context needed to compare audiences, content and campaigns over time |
| Data applications | Advertising, promotion, broadcast and digital campaign analytics | Connected the data platform to operational media decisions |
Bring us a data-platform challenge
If you're struggling through fragmented audience data, aging ETL, warehouse migration, or analytics that do not reach operational teams, Azati will help identify the engineering path before recommending a delivery model.
Assess your media data platformBusiness impact
Advertising decisions became measurable against audience and campaign data
The campaign-optimization application supported targeted advertising planning and performance analysis. The client reported a 17% increase in advertising revenue during the relevant period at comparable advertising-time volume. The available information does not isolate the application's contribution from other commercial factors.
One data foundation supported multiple media decisions
Streaming analytics, broadcast scheduling, content placement, audience targeting, promotional planning, and digital campaign analysis were supported by the same broader data environment rather than isolated reporting workflows.
Historical media data remained usable as the platform evolved
Historical modeling, SCD strategies, and analytical data marts preserved the context needed to compare audience, content, and campaign states over time, important when today's business decision depends on what was true months or years earlier.
The data estate could evolve without starting over
Custom migration and validation mechanisms supported the DB2 Warehouse Gen2-to-Gen3 transition while existing analytical dependencies remained part of the engineering picture.
What this engagement shows
For large media companies, data-platform modernization becomes more valuable when it is designed around the decisions the business actually needs to make.
In this engagement, the data environment was not treated as a warehouse project separate from the business. Streaming analytics, advertising optimization, broadcast scheduling, and campaign analysis gave the engineering work a direct operational purpose.
Who this engagement model is best suited for
This approach is particularly relevant to media and entertainment organizations that:
- Operate streaming, OTT, broadcast, or hybrid media businesses.
- Have fragmented or aging data infrastructure supporting multiple media products.
- Manage large volumes of historical viewing, audience, content, or advertising data.
- Need analytics to inform content placement, premiere planning, audience targeting, or advertising optimization.
- Are modernizing a legacy data warehouse or complex ETL environment.
- Need to connect audience and campaign data to operational media workflows rather than reporting alone.
- Need data engineering expertise that combines media-domain knowledge with data-platform modernization.
Frequently asked questions
A media data platform combines data from streaming, broadcast, advertising, audience, and content systems and makes it usable for operational and commercial decisions. It typically includes data ingestion, ETL pipelines, data warehousing, historical data modeling, data-quality controls, analytical applications, and reporting or decision-support interfaces.
In this engagement, the data environment supported applications for analytics, targeted advertising, broadcast scheduling, promotional campaigns, and digital campaign performance.
Streaming organizations can use historical viewing behavior, audience attributes, content performance, and engagement metrics to evaluate where and when content should be promoted or premiered. Historical modeling is particularly important because audience and content attributes change over time.
The streaming analytics application in this engagement used statistical data and calculated metrics to support content placement, premiere planning, and analytical decision-making across streaming platforms.
Targeted advertising uses audience characteristics, historical behavior, and campaign-performance data to identify relevant audiences and optimize campaign allocation. The value comes from connecting audience data with actual advertising outcomes rather than treating targeting as a standalone segmentation exercise.
The advertising campaign planning and optimization app's results were associated with a 17% revenue increase in a year, at comparable advertising-time volume.
Media data platform development can include data architecture, ingestion and ETL pipelines, data warehousing, historical modeling, data-quality automation, analytical data marts, database optimization, migration, and applications that turn media data into operational decisions. The appropriate scope depends on whether the organization is building a new platform, modernizing a legacy warehouse, integrating data sources, or extending an existing analytics environment.
In this engagement, Azati supported data warehousing, ETL, historical modeling, analytical applications, data quality, database optimization, and DB2 Warehouse migration across streaming, broadcast, advertising, and audience use cases.
Media companies can modernize incrementally by mapping downstream dependencies, preserving historical context, validating migrated data, improving orchestration and quality controls, and introducing newer infrastructure where it reduces operational risk. A phased approach is often safer than replacing a working data estate in a single migration.
During this engagement, Azati supported the migration from IBM DB2 Warehouse Gen2 to Gen3 and developed custom data-transfer and validation mechanisms using Prefect and SSIS.
A shared media data environment needs to handle different data structures, historical dimensions, processing volumes, and business rules across streaming and broadcast operations. It also needs data models that preserve historical context so that past audience and content states can be analyzed accurately.
The environment in this engagement combined a 90 TB data lake, 30 TB application database, and 300+ ETL processes, supporting applications for both streaming analytics and broadcast scheduling.
Slowly changing dimension strategies help data warehouses represent how attributes change over time. This matters in media analytics because subscribers, content metadata, audience characteristics, and other business entities can change, while historical analysis often needs to preserve previous states.
The project team applied SCD Type 1–3 approaches to changing attributes within the client's media data environment and used historical data to build analytical models for content and audience scenarios.
Media companies typically need orchestration, dependency management, data-quality controls, monitoring, validation, and performance optimization when hundreds of pipelines support interconnected analytical workflows. The challenge is not simply processing more data; it is keeping individual workflows reliable while changes to sources, models, and downstream applications propagate through the environment.
In this engagement, Azati supported 300+ ETL processes across streaming, broadcast, advertising, and audience workflows, using SSIS, Prefect, SQL/PLSQL, Great Expectations, and multiple database platforms.
Critical data infrastructure can be migrated incrementally by creating controlled extraction, transformation, validation, and cutover mechanisms rather than relying solely on a vendor's standard migration path. Validation is particularly important when downstream analytical applications depend on historical data.
During the engagement, Azati supported migration from IBM DB2 Warehouse Gen2 to Gen3 and developed custom mechanisms for data movement and validation using Prefect and SSIS where ready-made migration tooling was insufficient.
The decision depends on whether the organization's data workflows are generic or tied to proprietary programming, audience, advertising, and operational processes. Off-the-shelf platforms can provide infrastructure and standard analytics, while custom data products are more appropriate when the company's competitive advantage depends on specialized data models, historical context, or business-specific workflows. A hybrid approach is often practical: use managed infrastructure where possible and build the decision-support applications and data models that differentiate the business.
