The Customer Experience: CEM, QoS, and QoE is More than Just Alphabet Soup

Posted by Scott Sumner on Monday, August 17, 2015 with No comments
In the mobile space at least, performance is the biggest differentiation service providers have over cloud offerings and cheap, dumb pipes. Taking advantage of differentiated assets often means the difference between making revenue from mobile broadband instead of losing customers to Wi-Fi. For example, in the business services space providers can leverage prioritized, assured performance and premium data center connectivity to beat out cheap ISPs. But, customer experience management (CEM) is the key to maintaining this edge--and it's tightly tied to quality of service (QoS) and quality of experience (QoE) as measured by performance monitoring solutions.

Integrated SDN, NFV, Subscriber Presence Information 
and Network Visibility - to Drive Customer Experience

As a foundation for all services, the transport efficiency of the many, competing applications traversing today's networks--as well as the prioritization of each service's packets to meet QoS requirements--form the main influencers of customer experience. 

Service providers who can maintain excellent quality for a wide range of applications--such as low latency for VoLTE and SaaS, high throughput for 4k streaming, and burst handling for chatty IoT devices--will provide a seamless experience to their customers. Providers subscribe to the principle that you can't control what can't be measured, and that network layer performance optimization is the largest factor in improving QoE.

SDN: Coupling QoS Monitoring to Network Control


As software-defined networking (SDN) planning and deployment ramps up, 
service providers are developing big data analytics platforms that combine real-time network layer metrics with control-plane and application-specific KPIs. By correlating these different layers with the subscriber 'state'--where they are, what they pay for, what they are doing--analytics can create a more complete picture of CEM. 

This approach creates a live feed of application, network, and customer experience 'health' that can be used to drive SDN-control decisions on a per-subscriber basis--for example, allocate more bandwidth to a priority user watching a video--and then measure the QoE improvement immediately. When this is combined with machine learning, service providers have a path to a fully automated, experience-optimized network.

In the effort to optimize the customer experience, u
biquitous visibility is key to successfully integrate monitoring with control. Virtualization allows instrumentation to become ubiquitously deployed to the point where complete QoS--at all layers, all locations, all services and all subscribers--can be collected and analyzed together. 

This allows experience to be decoupled from infrastructure, and enables a different way to view network performance in general. For example, it may not matter that a network segment has significant packet loss if the customer experience is optimal. If things go wrong, being able to delve in the network layer to resolve any experience issues is fundamental, so it keeping 'network state' information fresh and available is crucial. 

Increasingly, CEM's purpose is not only to improve and maintain the best possible customer experience, but also to drive revenue. Network utilization and other more direct metrics can, more fundamentally, be used to drive up-sell opportunities for existing customers. For example, if a subscriber experiences regularly poor bandwidth performance on a 3G mobile network while watching videos, the operator can offer an upgrade package to their LTE service--and combine that with a video streaming package to increase ARPU further. This can be a major contributor to realizing a return on the significant network upgrade investments service providers have made. Operators in Latin America and some parts of Asia--where adoption of LTE has lagged far behind availability--face this challenge (and opportunity) daily.

Be Careful Not to Flood the Lake!

CEM will evolve toward correlating all aspects of the network with all levels of performance metrics (transport, control, and application layers) to create a highly versatile data lake for analysis. On the surface, more information would seem to be better--especially since operators don't always know what data will prove to be the future 'gold mine'--but caution must be taken to make sure data collected is accurate, precise, and worthy of storing. 

The next phase of CEM monitoring feeds is to use dynamic, adaptive data collection systems with the ability to increase granularity and coverage when issues, QoE, or changing usage patterns occur; the CEM system itself should specify what's needed and where. This is a logical step into the integration of CEM systems into SDN control. Result: QoE, instead of network performance Information, will drive configuration decisions, and determine the network's value to the operator.