Case Study
01
A Closer Look At A Streaming Company's Use of Datazoom for Data Pipeline Improvement
Date published: March 15th, 2021
Company Details
ABS-CBN Corporation is a Filipino media and entertainment group based in Quezon City, Metro Manila, Philippines. It is the Philippines’ largest entertainment and media conglomerate, producing, and streaming online news, and entertainment video content.
Highlights
Replaced the need to post-process data (table relations) with automated standardization within the Datazoom Data Pipe.
Reduced overall cost by removing bespoke solution and replacing with Datazoom.
Significantly increased the granularity of available metrics by using Datazoom.
Challenges
A major business objective of ABS-CBN is to understand how our users are consuming content across a variety of products and services. When we understand how users are behaving and why, we can use this information to drive strategic initiatives and measure the outcome.
Similar to other media platforms, we faced a lot of challenges in doing that. The first was just collecting and sorting the data. Because there wasn’t a turnkey solution in the market to do what we needed, we were collecting player events manually. We created a bespoke system to pipe some key events from the player to our database. This was a costly process in both time and money as we needed to define, implement, and validate every event ourselves. On top of that, we needed to implement and support the infrastructure. There is a lot of room for error in each of those steps. Any change we made to the events had to wait for a code release. Which was the second challenge we faced: time-to-market. It’s very difficult to understand problems and resolve them without the needed data. So if we wanted to pull more datapoints from the player, or redefine how we were doing that, it required an extensive process involving operations and development that could result in weeks or even months until implementation. Of course, we looked at those other turnkey solutions, like Conviva and NPAW Youbora, but they are very black-box. The lack of data transparency made it hard to find those actionable insights, to debug, improve, and grow the product. Which is really our third challenge. We needed access to the raw data. I needed to merge it with our other data sets to truly understand the user behavior and other influencing factors. This left me with either building myself, or Datazoom.
Key Challenges
- Collecting and sorting the data in an automated fashion.
- Easily and quickly make changes in the types of datapoints collected and how they are analyzed.
- Access to the raw data so it could be combined with other data sources to truly understand user behavior.
Implementing a Solution
Because we needed that data transparency, quicker time-to-market, and the ability to collect and sort data in an automated, scalable fashion, we chose Datazoom. The Datazoom platform enabled us to perform a number of analyses including user growth and engagement, video quality of experience, content ROI, content recommendations, ad ops, and marketing ROI. Datazoom is used by many teams at ABS-CBN; data, engineering, incident management, customer support, internal and external marketing, product, ad sales, ad ops, content production and curation, and executive. But what’s exciting to us about Datazoom is that it so easily integrates within our existing data architecture. Where other solutions, like Conviva and NPAW Youbora, are siloed platforms, I can collected data through Datazooom and connect it with several of our other systems. For example, we are leveraging Datazoom with Amplitude, as our main analytics platform, and Google Ad Manager as our ad platform, and a variety of video player vendors. Datazoom integrates seamlessly with the video players, collecting the data, enriching it with ad platform data in the pipeline, and then sending into Amplitude for analysis. We are also using Branch tangentially to attribute video and ad consumption to marketing initiatives.
What made Datazoom such a great choice was the ease of implementation. The ABS-CBN engineering team has integrated Datazoom on many different players and platforms now. The initial setup is simply plug and play, no real implementation. We have added our own custom metadata into the Datazoom integration, and this took the engineering team a few days for each player. Of course, there is always ramp-up time with a new technology. So our first integration required a two week period of testing and validating internally, to ensure we were getting accurate data. As we are now on our 4th implementation, we have cut this down significantly. Operationalization of Datazoom data is now as fast as the technical integration. And while we have integrated Datazoom into some bespoke players in the past, which requires some changes from the Datazoom team to support, the Datazoom team has been very helpful and accommodating.
The Results
One of my main goals has been to get user level viewing behavior. This means that all of our video data needs to be associated with an individual user. Datazoom supported this goal by catering to my identification method both in the data collection and when connecting to our database. For our database to index correctly on the user level, I don’t need to perform any transformations to the data myself, it is all handled in the Datazoom pipeline.
We had a bespoke player collection mechanism, connected to our own data pipeline. We removed all of this infrastructure and associated cost, and replaced it with Datazoom. The outcome is a more consistent data collection, for a lower cost.
Literally hundreds of metrics are at my fingertips since implementing Datazoom. From time to first frame, to user stickiness, and more actionable insights like content recommendations, and churn propensity.
A major business objective of ABS-CBN is to understand how our users are consuming content across a variety of products and services. When we understand how users are behaving and why, we can use this information to drive strategic initiatives and measure the outcome.
Similar to other media platforms, we faced a lot of challenges in doing that. The first was just collecting and sorting the data. Because there wasn’t a turnkey solution in the market to do what we needed, we were collecting player events manually. We created a bespoke system to pipe some key events from the player to our database. This was a costly process in both time and money as we needed to define, implement, and validate every event ourselves. On top of that, we needed to implement and support the infrastructure. There is a lot of room for error in each of those steps. Any change we made to the events had to wait for a code release. Which was the second challenge we faced: time-to-market. It’s very difficult to understand problems and resolve them without the needed data. So if we wanted to pull more datapoints from the player, or redefine how we were doing that, it required an extensive process involving operations and development that could result in weeks or even months until implementation. Of course, we looked at those other turnkey solutions, like Conviva and NPAW Youbora, but they are very black-box. The lack of data transparency made it hard to find those actionable insights, to debug, improve, and grow the product. Which is really our third challenge. We needed access to the raw data. I needed to merge it with our other data sets to truly understand the user behavior and other influencing factors. This left me with either building myself, or Datazoom.
Key Results
- More consistent data collection at a lower cost.
- Able to get individual user-level viewing behavior.
- Streamline data collection architecture.
Why Datazoom?
Datazoom has a growing reputation in the industry for delivering value as a vendor. Their level of support for their product is second to none. While their solution is generally plug and play, the engineers go above and beyond to support the integration and to deliver improvements to the product.
