Sponsored Content Datazoom, Inc.
Solving the Data Standardization Problem in Streaming Video
Video distributors understand that timely access to data is at
the root of solving quality- and service-related issues. And so,
they have invested heavily in video player software components
to capture a multitude of data elements about the overall video
experience. But that has created another problem—
fragmentation. Streaming video player technology stacks have become a
hodge-podge of custom software and vendor technologies, with
each component collecting different data about the viewing experience and delivery performance. Unfortunately, there is no
standardization amongst vendors about what data to collect,
when to collect, or how to use it, let alone how to provide standardized calculations.
What’s more, the data is usually delivered to proprietary visualization tools or storage for later analysis, and therefore unavailable for immediate application. The result? Operations personnel
must navigate to different data silos and manually piece together
a picture about quality or service issues before even thinking
about solving them, all the while hoping that their viewers haven’t already given up. What’s really embarrassing is that in many
cases Tweets verify user experience. Ideally you know a problem
before the world does.
To mitigate the potential for abandonment and churn as quality and service issues remain unresolved, distributors must address this data fragmentation problem.
What is Data Fragmentation?
The video player is a complicated element of the overall
video stack. Embedded within the code that enables decoding and playback of a video stream or file is a host of third-party and proprietary technology components designed to provide enhanced functionality and data collection about the
But, in almost every instance where one of these player components collects data about the video experience, that data is
sent back to the component owner rather than the customer.
That’s because most of those companies have business models
revolving around monetizing data visualization, which can also
include proprietary calculations.
And because the focus is on reporting, rather than action,
there is no hard and fast rule about how quickly the data should
be acquired. In some cases, it may be minutes or even hours old
by the time it hits the component provider’s tool making it very
difficult to take action and quickly resolve end-user issues.
A fragmented video player can severely impact a video distributor’s ability to provide a top-quality video experience in several ways:
• Redundant data collection—because the technology com
panies providing components for the video player aren’t
talking with each other, they are often collecting many of
the same data elements. This redundancy can eat up valuable end-user bandwidth and compute resources which
could be used to deliver higher bitrate video.
• Questionable data veracity—when so many different player components are capturing the same data element, and
then using it in proprietary calculations, who should the video distributor trust has the most accurate representation of
the raw data?
• Lack of control—each component within the player is managed by the vendor, not the video distributor, which makes it
difficult to enable specific requirements. Video distributors
are forced into a standard method for how their software
interacts with a component, instead of vice versa.
• Increasing player footprint—as new data elements and
tools are needed to monitor the video experience, more
components must be added to the video player, exacerbating the fragmentation and causing the video player size to
grow (which can impact video playback).
Session_ID and the Woes
of Data Fragmentation
When a user reports an issue, it’s critical for the video distributor to create a complete picture of the user’s video
experience. This way, the distributor can understand
where in the workflow the issue happened and resolve it.
Fragmentation makes this nearly impossible, as each
technology component registers its own Session_ID for
the user. This requires operational personnel to manually
piece together that user profile using data elements such
as their IP address, date and time, stream title, and more,
ultimately slowing down many processes, including the
ability to provide immediate support.