When viewers interact with a video player on your site, their actions trigger player events. Some of the player events return video metrics that you can use to analyze and improve your viewers' video experiences.
Below is a visual representation of player events in a timeline. The table below explains all the bold items. The gray events are included as reference points.
Use the setupError event to track critical errors in the player setup process.
Use the buffer event to track how often the player enters a buffering state and evaluate the quality of your adaptive stream. Since this event fires when the player enters a buffering state, this event may fire multiple times during a session.
Use the reason property of the buffer event to differentiate normal buffering ("reason": "buffering") from stalls ("reason": "stalled").
Use the adError event to track critical errors in the ad process.
Use the firstFrame event to track video starts, or plays. The firstFrame event fires once per media item, when playback has begun.
We do not suggest using the play event. The play event can be fired several times per media item. For example, when a view pauses and resumes content, resuming the content triggers a play event.
Use the visualQuality event to evaluate your quality of service (QoS). This event fires when the rendition of an adaptive stream changes. The visualQuality event has two relevant properties:
• mode: The mode property returns one of two values that correspond to how the visual quality has been changed: "manual" or "auto". To evaluate the quality of your stream, track when "auto" is the value returned. When "manual" is the value that is returned, a user has manually changed the visual quality. When "auto" is the value that is returned, the visual quality has been changed automatically by the player, based on bandwidth calculations.
• level: The level property returns characteristics of the new rendition, like bitrate, width, and height.
Use the time event to track quartiles on video on demand (VOD) assets. You can use the duration and currentTime properties of this event to calculate the percentage of the video viewed. Then, you can determine quartile events.
When tracking the time event, be sure that the analytics ping is dispatched only once. Since the time event is fired frequently, use a flag or another mechanism to prevent multiple firings.
NOTE: Live streams do not have a duration. Therefore, you cannot track quartile events.