Deeper Dive—What is the Datatecture and how can it improve streaming?

The Datatecture was assembled with the purpose of “mapping out the ‘data architecture' of video streaming so that content providers and vendors alike can build businesses and supporting technology stacks that are truly driven by data.” (Datazoom)

For the end user, streaming video is simple. It should work, look good and be available on multiple devices. Behind that, though, is a frustratingly complex system.

Datazoom, an enterprise video platform, has gazed firsthand upon the byzantine streaming video infrastructure and has decided to help the industry better understand the system and how data flows through it. The company has assembled a picture and inventory of the streaming video technology ecosystem, broken down into core areas and sub-groups, and called it the Datatecture.

The company is doing this with the stated mission of “mapping out the ‘data architecture' of video streaming so that content providers and vendors alike can build businesses and supporting technology stacks that are truly driven by data.”

Still confused? We are, too. So, we asked Datazoom CEO Diane Strutner to give us a more comprehensive explanation of what the Datatecture is and how it can make streaming video better.

The following has been edited for length and clarity.

Fierce Video: What exactly is the Datatecture?

Diane Strutner: There are so many different products and companies involved that make streaming happen and it can be challenging to know what all these moving parts are, who provides these moving parts and what technologies are involved.

I think the Datatecture can help with how we understand the landscape involved in video streaming. That’s important because at the end of the day we want to understand the state of our systems. There’s an industry buzzword that began to pick up last year and that’s observability. Do I know the state of my system? At the end of the day, video delivery and end-to-end workflow for video is kind of like a supply chain. So, knowing what parts to be looking at and who is providing them is a big part of this.

The second layer, for when we can understand the state of the system, is that we eventually want to optimize it. Optimization can be for improvement of performance, cost reduction or revenue increase, but how are we supposed to go about optimization if we don’t have a better understanding of this ecosystem and how they work together?

As the industry evolved, there was Netflix and YouTube and then there was everyone else. Netflix and YouTube built their own infrastructure and they built their own components of this end-to-end system. If you’re anyone else, including the Disneys of today, you’re using services from all these other companies. So, efficiencies, observabilities and optimizations that Netflix and YouTube have been able to do since they controlled all the technologies in the stack, have been untenable for all these other companies.

The Datatecture part of all this is that data is the thing that connects all these disparate services. I definitely see this as a project that will continue to grow and expand. Understanding more about what these core services do…is important along with the data that it creates. Also, we can say, how can data from here be relevant to this other process? For instance, how can data about end user bitrates that are being streamed be used to influence the transcoding ladder in real-time? Those are the levels of optimizations that Netflix and YouTube have had because they built all their own instrumentation, they created data standards, they created centralized systems that can take data in from all these different sources, and then automated for efficiencies purposes. We need the ecosystem but also the data ecosystem.  

Fierce Video: How specifically can it improve streaming video?

Strutner: For the end user, the results will be better streaming experiences across more devices and more locations; better content given to them through better recommendations and more intelligent content licensing decisions; advertising that can be more interesting and relevant to them as more data can be exchanged; and optimization around tolerances for certain settings like, I tend to drop off after seeing three ads in a row but my mom will watch seven. So, there’s a lot of customization and personalization of the experience that we can begin to have when we have more data at our fingertips.

For a content provider, it’s about running a more efficient business. Am I using the right services? Am I paying the right price for those services in comparison to the results that I’m getting? Am I able to use data to decide between the different services that I’m using? Am I suggesting the right content to users? Am I licensing the right types of content to support the user experience?

We have a customer that is using data from video playback, overlayed with data from their ad server, overlayed with data from content licensing to determine how much money they are earning from a specific title in comparison to how much they pay for it on the content licensing side. So, they can build dashboards around profitability of content, which is awesome. I’m sure Netflix has that but I don’t how many other companies have something like that.

Fierce Video: So, what does the Datatecture look like in action? Is it something where vendors and content providers are opting in to make it work?

Strutner: It is an image and an accompanying microsite. I see Datazoom as a solution to capture data from or send data to any and all of these different companies, products and services across this inventory because everything we do can either be improved with consumer data or creates data through its functionality. For myself, I was trying to understand what are the different companies out there, what do they do and what data are they creating? I started putting all these logos together and started showing it to people and they were like, “Whoa, this is cool. Can you send this to me?”

I really want this to continue to be a resource because I strongly believe that the industry will be improved if we can continue to expand upon this. We’re not asking for monetary participation in this. We’re trying to provide a resource for the industry to better understand ourselves.

We’re doing our own research but we’re also asking the ecosystem to help with creating this definitive resource.