If your business is built on delivering video over the internet at scale, it only makes sense that any quality evaluation should start and end with the customer watching the product you’ve delivered — the viewer. Their opinion, at the end of the day, is the only one that matters. The customer, as the saying goes, is king.

Traditionally, however, quality metrics have been network-centric rather than viewer centric — largely focused on quality-of-service (QoS) rather than the viewer’s quality of experience (QoE). This problem is so pervasive, we use the term Viewer Score instead of QoE.

Quality of service metrics have been defined by network-service-level parameters such as network throughput, package drop rate, video bit rate, network delay, etc., and sometimes include integrity checks that indicate whether the videos and associated audio and caption information are properly reproducible at user devices.

These metrics are fine, as far as they go. They provide a ballpark idea about the general health of delivery networks.

What these metrics don’t do is make meaningful predictions of viewer satisfaction, and they aren’t great at assisting in timely repair of video quality degradations or helping with allocation of bandwidth resources.

Comparing two apples in a grocery storeSSIMWAVE’s technology starts and ends with the viewer’s quality of experience in mind. This achieves three things. One, it makes the viewer far happier. Two, it’s far more accurate. And three, it generates cost savings.

Unlike QoS, Viewer Score is viewer-centric, directly measuring the acceptability, satisfaction, happiness or enjoyment of the end-viewer. In many scenarios, Viewer Score makes substantially better quality predictions than QoS.

For example, the same video streamed to two subscribers with identical network conditions who are using different viewing devices (e.g., TV vs. smartphone) could have identical network QoS scores but very different perceptual Viewer Score.

Only SSIMWAVE’s Viewer Score can produce reliable, immediate predictions of viewer satisfaction. Based on those predictions, quality problems can be identified, localized and fixed, and resource allocation, i.e., bandwidth, can be optimized, generating cost savings.

Bottom line? Focusing on the Viewer Score rather than QoS makes sense, both from the point of view of an end-viewer’s satisfaction, and from the point of view of cost.

The original article was published on Medium.com. Jan 4, 2019.