I define Internet TV as professionally- or prosumer-created video delivered over the internet, often thru a web browser. That definition doesn't quite cover all of YouTube, since so much of their content is random (to use as kind a word as I can find), but certainly YouTube in toto is part of Internet TV, too.
I've gained a tremendous amount of experience and exposure to the processes involved in delivering Internet TV content, and its much more complicated than most people imagine. When I go to tech conferences and sit in on sessions titled "How to publish video online", I'm flabbergasted that these presentations often revolve strictly around the presentation layer inside HTML, and ignore the complexities of publishing for-profit video content online.
The value chain for publishing Internet TV involves the following components:
- Encoding
- Video Content Management
- Advertising
- Protection
- Delivery (CDN)
- Playback
- Measurement
Here are some additional thoughts on each of the value chain components.
Encoding
Encoding means transforming video sources into digital bits suitable for transportation over the Internet. It used to be that encoding was integral to Delivery and Playback, but as the entire video industry moves to standards like H.264, encoding can become independent of these downstream value chain components.
That doesn't necessarily make encoding any easier. Even with an emerging standard, there are still many considerations for encoding, starting with the nature of the video source. Libraries of existing video content can be processed in batches asynchronously. Live video feeds must be processed in real-time. Ideally you must also encode at multiple dimensions and quality levels (translating into an array of different bit rates). This means additional processing power for live events, and additional storage for libraries. And of course, you should make all your live events available afterwards on-demand.
Video Content Management
Like any other kind of content being published on a regular basis, video content needs to be managed. Fundamentally, this means keeping track of which content is ready for publishing, and which content is already published. From there, you can imaging all sorts of additional business rules and related metadata that you'd like to manage for your content, which is often encoded as XML and stored in databases on robust back-end systems.
Often times, video content management encompasses many of the value chain components in aggregate, such as encoding files that are uploaded, storing them on servers, publishing them to websites, communicating with advertising systems, and so on. And sometimes, they don't, which means you'll have to integrate these other components yourself.
Advertising
There are two aspects to advertising that you must consider when assembling your solution: where to put ads in and around your video content, and what ads to display.
The "where" question is often defined using a "playlist" concept, which may be component managed by your video content management system. With a mix of the old (TV) and the new (Web), the playlist may include a sequence of content segments, interspersed with in-roll (pre-roll, mid-roll, post-roll) video ads, along with parallel companion banner ads that appear next to the content and overlay ads that appear on top of the content.
For short video segments, the playlist may want to have complex business rules that dictate in-roll ads only after N plays or only after N minutes of playback. For longer video segments, the challenge is in identifying the natural breaks within the video content suitable for traditional mid-roll advertising. For professional content, these breaks are already built into the content for traditional broadcasts by the editor, and the industry must move towards standards that allow editors to pass this metadata along thru the encoding and publishing steps with as little additional effort as possible.
Once the "when" question is addressed, the next challenge for advertising systems is the "what", to fill the ad spots at playback that are most suitable for the viewer of the content. Ultimately, the value of each ad will be measured by the relevance of the ad to each viewer. Some advertising solutions focus on matching ads to the context of the content, while others focus on matching ads with the behavior of the viewer.
Each have their strengths and limitations, IMHO. Educational or informational programming, like cooking shows, will benefit most directly from contextual advertising, because viewers are specifically in the mindset of knowledge acquisition.
Protection and Delivery
Good fences make good neighbors. Good content protection makes your content available most conveniently to your viewers, while preserving the business model you need to remain viable. The corollary is that castles aren't very inviting: make your content too difficult to use, and no one will try.
For Internet TV content, protection is often closely tied to the form of delivery, either streaming or download.
Streaming involves delivering bits in real time to a viewer in an encrypted fashion, such that the bits are displayed and then immediately discarded. The pros for content owners is that the content remains centrally managed, and the content is available on-demand for viewers. The cons are that any disruption in that real-time delivery can translate into a disruption of the playback experience, and that users must be online in order to watch the content.
Downloading involves delivering bits in asynchronously, in an encrypted fashion on a per-viewer basis. The pros for viewers is that they can watch the content regardless of the state and quality of their online connection. The cons are that viewers will opt to download higher bitrate versions of content, increasing the delivery costs.
Speaking of delivery costs, the CDN cost difference between streaming and download have dropped significantly in recent years, and streaming has incorporated protection, while protecting downloadable content is still a separate step and cost at this point.
Playback
The emergence of standards like H.264 take video quality almost completely out of the equation when evaluating playback environments. But the other aspects are still tremendously important, including protected delivery and advertising.
And of course, Internet TV should eventually transcend traditional TV, particularly in terms of interactivity. What that actually looks like in 50 years, I don't think anyone really knows. But playback environments need the developer base, tool set, and device support to enable exploration and experimentation.
Measurement
After all that, we come to the most important, and least mature, component of the entire Internet TV value chain. You can't manage what you don't measure, so make sure you put measurement in place *first*, and the rest of the components as you see fit. Fundamentally, this means figuring out what questions you want answered, then figuring out how best to answer them.
The challenge with measurement is defining exactly what you want to measure, and understanding what metrics the various other components provide. Server-side components may track hits, and client-side components, rolled into your playback environment, may track starts and stops. The industry is starting to tackle the challenges around definitions of measurement, and how to translate low-level metrics like these into higher-level metrics like audience size and engagement time.

Recent Comments