How Datazoom does Data

Based on feedback we receive from customers and partners, the Datazoom engineers are constantly working to expand the array of supported data points available to flow through Datazoom’s Data Pipes.

We bucket data points into three general categories based on its nature. These buckets include Event, Metadata, and FluxData.

With these buckets serving as our “X-Axis”, our “Y-Axis” would include the way we prioritize data for the sake of our Engineers’ technology sprints.

First, we will explain the Event, Metadata, and FluxData captured by Datazoom:

1. Event Data is generated by user interactions with the player and non-user generated occurrences affecting the stream. User invoked data points include items like Play, Pause, and Volume Change would be included in this bucket. Non-user invoked events include buffering, quartile, and bitrate shift to name a few.
2. Metadata includes data which describes Event and FluxData. These points include items like Device, Session_ID, and CDN.
3. FluxData is data collected at predefined intervals as a sort of “heartbeat”. Such data would include current Bandwidth, current Rendition Bitrate, current Volume Level, etc. FluxData can be updated on-the-fly within the Web Console. Currently, we support FluxData collection at some pre-defined intervals (currently 1, 5, 10, 15, 30, 60 and 300 seconds) and also allow users to define their own custom interval.

Now that we understand the overall fashion in which Datazoom buckets data, we can go into more detail about how we further break it down for our technology sprints. In our release notes, we will refer to supported Collector data points in terms of one of four particular Levels. What follows serves to clarify precisely what this terminology means:

Level 1: Basic Data Points (14+ items)

Level 1 data points pertain to general audience data. Supported points include Play, Pause, Seek and user-initiated Event data or general metadata, as well as Flux Data like bitrate, play-head position, and buffer fill.  

Level 2: Quality of Experience Data Points (71+ Items)

The majority of Level 2 Event, Metadata and FluxData points pertain to Quality of Experience information which is generated by the fashion by which the asset is streaming to the end user. These data points include Content Quartiles, ABS Rendition Shifts,  and Audio among others as well as Client Metadata.

Level 3: Ad Playback Data Points (57+ Items)

Where Level 1 and Level 2 data concerns the video asset itself, Level 3 data concerns data generated during ad playbacks. They include user interaction events and metadata, as well as Quality of Ad Experience data and timing events to name a few.

Level 4: Custom Data Points

Every business, every team, is different. Sometimes, our customers will want to collect innovative, unconventional data points to suit their particular needs. Our team will work to meet your goals and demands by enabling the capture of such data as requested. Click here for more information on requesting custom data points.  

So now you know! If our release notes say that “Datazoom now supports Level 2 support for Bobplayer,” you can rest assured that we’ve added detailed QoE information to the array of data Datazoom is capturing.

You can also check out the specific data points supported per player by clicking here.