Q: “Data” is one of the biggest buzzwords in streaming video
right now. What are the kinds of data points that streaming
video publishers and OTT services need to be looking at?
Let’s start by defining “data.” A data point is non-negotiable, raw
information that can act as a point of truth. Unfortunately, I believe
our industry suffers by conflating “data” and “metrics.” Metrics
are formed by bringing many raw data points together. Although
there’s been movement toward standardizing these calculations,
most vendors calculate metrics differently, and thus we cannot
use cross-vendor metrics for comparison purposes.
I believe the industry has been sold on a world of metrics creating
vs. data gathering, and so the overwhelming majority of content
distributors don’t capture any of their own data, nor do they have
access to it or control who does. Post-processed data or metric
measurements are stored in silos by third-party tools, and are only
made available minutes, hours, or days later.
Q: How does Datazoom approach data and analytics differently?
To start, Datazoom views the functions of capturing data and
analytics as separate things.
Data falls into two categories. We can call the first Functional—
data used to power critical services—and the second Feedback—
data that lets us know how things are performing. The VAST
events exchanged between a player and an ad server are
Functional data, whereas throughput or bitrate measurements
today are Feedback data, as they aren’t critical to performance.
Datazoom is building the first data capture library of every data
point, produced by any service, that can be captured on any
player, platform, or device in video. Since every platform is
different, we have many data libraries. Datazoom is focused on
data capture, management, and routing, enabling customers to
not only capture any and all data (without duplication), but also
send it wherever it needs to go. Our ecosystem of “Connectors”
currently includes analytics (both video-specific and generic tools)
and data warehouses, and will later include ad servers, CDNs,
and other services that rely on Functional data. We want to move
the industry away from using many third-party SDKs, which bloat
the video player, over to a single SDK for improved collection and
control of data.
Q: With “zoom” in the name, presumably speed
is central to what Datazoom does? How do you
achieve data reporting and analytics at high speeds?
Since the world of video streaming operates on a second-to-second
basis, data should be held to the same standard. We guarantee
sub-second latency on data retrieval using our patent-pending
platform and Data Delivery Network over the public cloud. Speed
is (literally) built into our platform.
Q: How easy is it for publishers to integrate
the Datazoom solution into their workflows?
Our demo will take you through the integration process. It takes five
minutes. We have a one-time, one-line-of-code, copy-and-paste
integration that collects video player data, called a“Collector.”
Next, inside our interface you set the data points you want to
capture (metadata, events, or FluxData), and capture frequency,
down to once per second. Last, you select the tools and input
the account identifier(s) for each tool you want to send data to—
called “Connectors”—and you’re done! You’ll start seeing your
data, in different analytics systems, in seconds.
Q: What’s on the horizon for data and video? What’s next?
If you think about it, we’ve not been able to use data in a meaningful
way on a per-stream basis. We pump data into analytics systems,
where people look at screens and try to take action on big issues.
People are the intelligence and optimization behind video. But since
video streaming happens on a second-to-second basis, people often
cannot react to a change in the time required to be impactful, and can
only take on one issue at a time.
In the future all data will be Functional data. Technologies like machine
learning and AI can handle the volume, variety, and velocity of data
created by video, and can be rolled out at scale to make changes
on a per-stream basis. But these platforms require data. Fast data.
Diane is a proven corporate leader. She was previously the VP of Global Sales
and Business Development at NicePeopleAt Work. She played a pivotal role at the
company in both sales and business development, shifting the company focus
from several products to their budding QoE Video Analytics platform. Before NPAW
she was the Director of Business Development at GetApp (Acquired by Gartner).
A California native, she is a graduate from Manhattan College where she earned
a Bachelors of Science in Marketing and was a Division 1 scholarship athlete who
competed and served as team captain on the indoor volleyball team.
DATAZOOM | THEWIZARD@DATAZOOM.IO | WWW.DATAZOOM.IO