How Real-Time Data Will Transform the Business of Online Video

The online video space is facing a lot of pressure right now. Television audiences are moving online, but revenue generation is lagging. In fact, online video is a lot like the mobile space a decade ago—filled with hype and promise, but difficult to monetize.

Traditional broadcasters who have both a legacy television and an online video business are suffering through the pressure. Even companies who were born and raised from streaming are finding it hard, and struggle to generate head-turning revenue from online video. Sure, Netflix may have a lot of subscribers. But the largest potential revenue opportunity isn't in a subscription model - as Hulu's financials reveal, the best model is in fully-monetizing content by using engaging and targeted advertising. Despite the greatest growth coming from audiences moving online, when it comes to numbers the revenue generated from online advertising just can't seem to compete with the 100% fill rates and high CPMs of traditional broadcast. 

Relieving that pressure and transforming online video into a high-value industry comes down to harnessing and using data. 

Data Challenges in Media Today

Content distributors must leverage technologies from a variety of different vendors in order to collect and analyze the data they need to provide a great viewer experience. Data can come from encoders, CDNs, ad-servers, audience measurement tools, and more.

And when it comes to selecting those technologies, content distributors ultimately have two options: build or buy. With a few exceptions, most content distributors believe that they can get the most value from purchasing best-of-breed products and building them into their own technology stack. There-in lies the problem. Because although each customer sees each technology as part of a greater system that should work together, vendors in the space continue to disparately develop products without a vision for the customer’s complete stack. What you get are products that perform specific functions, but lack the ability to synchronize with other systems. For the content distributor, realizing the benefit of any one system can only come from utilizing all systems in unison.

So, in what does this lack of synchronization result? It slows everything down. When systems don’t play well together, people must take extra time to integrate them. Employees may have to touch dozens of systems independently because the data in each system is siloed. And this increases the risk for error.

The question therefore is, how can we unify the different systems? Well, it must start with creating a common thread to link them together. That thread is the data.

There are four main pain-points today around collecting data about the video experience:

Integration Process, Testing, and Roll-out. Collecting new data or initiating new services that need data from the player results in updating and maintaining data collection integrations, which are highly labor intensive and present other issues as well. The video player, like any piece of software, is subject to a battery of testing to ensure compatibility across devices and browsers. Whenever it needs to be updated, the new code must be tested as well as the player itself (once the changes have been integrated) to ensure that the new code operates well with the existing integrations. This makes content distributors and broadcasters resistant to making changes to their players, despite wanting to collect more data or try different systems that will help them better monetize the video experience.

Data Interoperability. There is no data interoperability between the collection beacons used by various technology companies. For example, Nielsen’s audience measurement beacon may be collecting data in a different way, or different frequency than Nice People at Work’s QoE beacon. This results in duplicate, inefficient, and heavy 1-to-1 data collection installations on the video player—lots of beacons used by different technology companies are often collecting the same data! 

Data Relevance. Access to data from tools used today is provided minutes or hours after it has been collected. Although a problem can be identified after analysis (which may happen even later than the data is received), it’s often too late to take any corrective action. What’s more, newer programmatic technologies, like Machine Learning and Artificial Intelligence, need data in real-time.

Data Parity. The beacons installed on the video player by third-party companies collect data at different intervals, and calculate metrics and KPIs in different ways. Since the data points being looked at are different (aka a 1-minute data point coming from aggregation of 5 second heartbeats vs. a 1-minute data point coming from an aggregation of 10 second heartbeats) metrics between third party tools cannot be looked at equally. Even if the same data points are used when forming metrics, various third party tools have their own way of calculating commonly used metrics (i.e.,should Buffer Ratio include or exclude true outliers)

Getting all these systems to work the same, with respect to data, is critical. Without that, using the data they throw off takes a Herculean effort of normalization and processing…not to mention having to employ a variety of different interfacing tools. But that’s just half the problem. The other half is making that parity happen in real-time. If one system is working at five minute intervals and another is working at five second intervals, it increases the difficulty of getting them to interoperate. To truly unite the technology stack, these tools must all speak to each other in real-time because, ultimately, that is what the content distributors and broadcasters need in order to increase the opportunities to generate revenue from online video.

The Future of Media is Automation

So, if we can connect these systems with real-time data, what else can we do? With few exceptions, media has been an analytics-driven industry. Gather data, analyze, and adapt. However the truth of the matter is that few companies can afford an army of experts to not to only interpret results taken from analytics, but turn these insights into actions. And if content distributors or broadcasters don’t take action upon data being collected, there is no realized value from the analysis at all!

Ultimately, most of the challenges around creating a better video experience (and, consequently, increasing opportunities to generate revenue) aren't around trend identification, they're still in the nuts and bolts of creating an efficient system—controlling the quality of content delivery, filling ad slots with ad creatives, creating an interactive and engaging experience with end-users, etc. These are challenges that can only be addressed on a granular basis, and in real-time.

Analytics platforms can help content distributors and broadcasters identify shifts and trends in the online video industry, but because we don't have standby armies of engineers or customer services professionals addressing issues in real-time, analytics won’t help anyone tackle some of those core challenges. It will take a true unification of the technology stacks, which can only be provided by all the technologies using real-time data, to greatly improve chances for generating revenues.

Moving from Analytics to Automation-Driven Video Strategies

Connecting the disparate systems of a video delivery technology stack together through real-time data, enables media companies to leverage new and powerful programmatic technologies like Machine Learning and Artificial Intelligence.

Instead of relying upon people to turn data into actions, media companies can build systems that trigger actions directly from data, without human interaction. Incremental and systematic changes using real-time data means that they can influence an end-user’s experience in real-time. For example, server-side ad insertion (SSAI) is growing in popularity because of its ability to avoid ad-blockers, but it removes the possibility of personalizing the ads through programmatic delivery, because it lacks real-time insight/information from the player. Imagine using real-time data to enable ads to be loaded on a 1-1 basis to maximize revenues for SSAI.

But automation built on irrelevant data will create no benefits – to move the revenue needle, automation will need to be built on a solid foundation of information, and the building blocks of this of this foundation are made of real-time data.

Real-time data will allow media companies to ask themselves critical questions such as, “Can we ensure a perfect match form the first second between image resolution, screen size and connection speed?” “Could we create video distribution with 100% ad-fill rates?” and “Could we build truly interactive environments, enabling end-users to engage with enriched content and personalized video experiences?” And they can answer these questions through automation.

Enabling the Real-Time Data Revolution

datazoom is creating a Data-as-a-Service (DaaS) network to deliver and integrate data into your online video technology solution stack in sub-second real-time. If you'd like to discuss your data strategy, and how you might be able to increase monetization opportunities through real-time data, let’s set a time to meet at NAB.

[email protected]