Finding Meaning in Multiscreen Metrics

It is hugely beneficial to all living creatures that a byproduct of photosynthesis is the creation of oxygen. We'd be in big trouble if that wasn't so. Though it may not be quite as important in the big picture, the pumping of content through IP-based networks creates its own serendipitous byproduct: Trillions and trillions of useful bits of data. This data is the raw material of analytical procedures that help ensure that services are being delivered correctly and that subscribers are being offered products and services in which they have particular interest. The idea is simple. Finding a needle in a moving haystack - delivering an on-demand piece of content to a tablet running a particular profile of one of several possible adaptive bitrate streaming techniques to a user identified by the unique signature of his or her machine - generates an avalanche of data. This data has lots of uses and can be a great help if put to work by the operator - and a great hindrance if it is neglected.Analytic information is vital to ensure that systems are operating efficiently and for operators' marketing departments hoping to keep content relevant and to find and monetize upsell opportunities. For the operations and technical departments, it is vital to keep end users' quality of experience at an acceptable level.There are two bottom lines here. The first is that the cable industry enjoys a big advantage over over-the-top (OTT) competitors who don't have access to a lion's share of these analytics because they only are tenants on the network.The second is that leveraging all this data is beyond anything the industry has done in the past. IP networks support such a rich and varied array of devices, content and end users that the amount of data generated is orders-of-magnitude greater than in the past. If companies aren't ready, they won't benefit from this input. Moreover, this isn't really an option. The inherently complex and expansive nature of the multiscreen universe means that performance almost certainly will suffer if advanced analytics aren't used.Examples of the importance of analytics are not tough to spot. They are, for instance, the linchpin in the measure of video quality. In today's environment, cable operators must measure quality of experience (QoE). This is a mix of quantitative and subjective measures that is more useful and more data-intensive than metrics-only quality of service (QoS) measures.QoE measurements can be tough, especially as multiscreen proliferates. Duncan Potter, the CMO of Seawell Networks, pointed out that the single stream of data arriving will provide content to a variety of devices using different ABR streaming systems and different profiles within each. QoE will need to be maintained for each of these sessions.Thus, a boy playing a video game in one room, his parents watching a Blu-ray movie another and a daughter streaming videos in a third must all be supported by a set amount of bandwidth. Analytics will be a key tool in gaming how to do this. "By looking at the patterns of use and change, what you can do at that point in effect is calculate a MOS [mean opinion score] or something akin to that from the QoE standpoint," Potter said. "If you can manage that in real time with tools such as ours, you can ensure a very consistent QoE."Operators can't consider excellent QoE an option. Their multiscreen offerings will be measured against their stable and high quality legacy services. The challenge is that the amount of data operators will be required to wade through in order to maintain adequate QoE is orders of magnitude greater than it is for those legacy services. It will be virtually impossible to scale older systems to track the trillions of log entries multiscreen generates quickly enough to help the QoE of an ongoing session, said Jon Haley, the vice president of development and product marketing for Edgeware. The answer, he said, is "big data" platforms that can create real-time pictures of sessions and how they are operating instead of dealing with the individual log entries of which each session is composed.The coming of big data engines and other sophisticated tools is good news for marketing departments. Fabien Maisl, the director of marketing for Witbe, points out that cable operators are starting at a disadvantage: Online retailers have spent a generation collecting detailed data on their customers. They have loads of data and know how to get more.Cable is in catchup mode. "Google and others have had one-to-one relationships with their customers for the past 10 years and have a very deep knowledge of who they are, what they are looking for, how to please them. Cable operators do not have that. They are looking for analytic capabilities to give them the same level of knowledge on how to please their subscribers and enter into relationships with them - not just charge them a monthly fee.”Carl Weinschenk is the Senior Editor of Broadband Technology Report. Reach in at [email protected].