Video Stream Quality Impacts Viewer Behavior: Inferring Causality Using Quasi-Experimental Designs

Abstract

The distribution of videos over the Internet is drastically transforming how media is consumed and monetized. Content providers, such as media outlets and video subscription services, would like to ensure that their videos do not fail, start up quickly, and play without interruptions. In return for their investment in video stream quality, content providers expect less viewer abandonment, more viewer engagement, and a greater fraction of repeat viewers, resulting in greater revenues. The key question for a content provider or a content delivery network (CDN) is whether and to what extent changes in video quality can cause changes in viewer behavior. Our work is the first to establish a causal relationship between video quality and viewer behavior, taking a step beyond purely correlational studies. To establish causality, we use Quasi-Experimental Designs, a novel technique adapted from the medical and social sciences. We study the impact of video stream quality on viewer behavior in a scientific data-driven manner by using extensive traces from Akamai's streaming network that include 23 million views from 6.7 million unique viewers. We show that viewers start to abandon a video if it takes more than 2 s to start up, with each incremental delay of 1 s resulting in a 5.8% increase in the abandonment rate. Furthermore, we show that a moderate amount of interruptions can decrease the average play time of a viewer by a significant amount. A viewer who experiences a rebuffer delay equal to 1% of the video duration plays 5% less of the video in comparison to a similar viewer who experienced no rebuffering. Finally, we show that a viewer who experienced failure is 2.32% less likely to revisit the same site within a week than a similar viewer who did not experience a failure.Index Terms-Causal inference, Internet content delivery, multimedia, quasi-experimental design, streaming video, user behavior, video quality.

Improving QoE prediction in mobile video through machine learning

“…The problem of QoE assessment in HTTP video streaming is already well-known and well studied, and different QoE models for video streaming have been proposed in the past [7], [10], [12], [13], [15], [21], [23]- [25]. Today it is well accepted that stalling (i.e., stops of the video playback) and initial delay on the video playback are the most relevant KPIs for video streaming QoE [12]- [14], [23].…”

“…Besides pure video quality modeling, other papers [6], [9], [10] have addressed the problem of user engagement prediction for HTTP video streaming.…”

“…There are different tools [5], [6], [19] which are capable of monitoring application-layer metrics which are highly correlated to QoE in video streaming services. Buffering events or stallings, video quality/resolution switches and initial playback delay are accepted today as the key application-layer metrics which can be used to predict the QoE undergone by the video watcher, using different models proposed and investigated in the literature [7], [10], [12], [13], [15], [21], [23]- [25]. Out of these metrics, stalling is the paramount one, specially when it comes to mobile video watched in small end-devices such as smartphones; in fact, in [11] we show that QoE for video streaming in modern smartphones is actually slightly impaired by video resolution changes.…”

On the trajectory of video quality transition in HTTP adaptive video streaming

“…The following objective QoE metrics are known to affect user experience [11,15,19] and are used in evaluating the impact of the proposed model on the two modified algorithms.…”

“…-Convergence time: is the time taken to settle at the sustainable video rate. -Start-up Delay: is defined as the amount of time it takes a player to download a predefined number of chunks before the playback starts [11].…”