The customer here is the content strategy team inside broadcasting networks. Understanding the performance of content, after it is aired, becomes vital for the creative teams. Applying these insights, the team can then work on strategies which improve the show performance. These insights are also relayed to the marketing teams for effective targeting and promotions.
Convincing and communicating content strategy teams the possibilities of using technology for creative decision making is not easy.
The content insights must be reliable and statistically significant to aid the story-telling process.
They should be easily interpretable through a simplified framework of metrics.
Computer Vision to identify actors and actresses decked under makeup and costumes.
Correlate content with viewership
A simplified metric which correlates screen share of content parameters with viewership.
Scalable for additional content hours.
A scalable solution that is not dependent on the linear-broadcast copy.
A framework to capture high utility metrics from the vast repository of metrics.
The proposed AI solution must understand all aspects of the GEC content and then can be used to analyze the impact of different content aspects on the Viewership of the show.
AthenasOwl had worked with a major broadcaster on a project to compare popular shows with its competitors. One of its findings was how the storyline, the setting, choice of cast, all of these play an essential role in determining success metrics of different episodes. This engagement laid the foundation for the BioGEC product- powered by Athenasowl, explicitly targeted towards Broadcasters, Production Houses and Content Creators.
We partnered with BARC – the world’s largest television measurement body to create a module that leverages our at-scale content tagging abilities with the industry-standard analysis and insights framework of BARC. The tool then puts these insights on easy-to-understand dashboards.
BARC and AthenasOwl collaborated to create the first version of the product tailored towards the daily soaps under the GEC genre. The product makes use of various AWS components to deliver high-quality service to the clients.
The approach here was a two-step solution to crack the problem.
As a first part, AthenasOwl platform processes the video to generate metadata that helps uniquely identify each part of the video based on its attributes.
Secondly, this data is merged with the viewership data. This merge then helps derive actionable insights based on video properties and its impact on the Viewership.
A niche taxonomy around Characters, Plots, Emotions and other attributes for Indian GEC content
Reusable AI Elements building up a customized metadata generation workflow
Automated Data Pipelines to cleanse, transform and fuse viewership data and metadata
Dashboards to analyze the metadata after combining it with the viewership data
Amazon S3: The raw video data, as well as viewership data, is stored using the Amazon S3 buckets
Amazon EC2: Web UI hosting is done on the Amazon EC2 instances
Amazon Lambda: When new viewership files arrive on the S3 bucket, AWS Lambda function gets triggered to process the file. Also, data fetch for the UI is done via AWS Lambda calls
Amazon RDS: Processed data that gets rendered to the UI is stored on Amazon RDS
Amazon API Gateway: The communication between the UI and RDS to fetch data happens via calls made to Amazon Lambda via API Gateway