By Roger Brooks, Ph.D., Pankaj Kumar, Mudit Jain, Megha Vij, Nandit Jain & Andrew Colby, Guavus
While service plan price will attract customers, it is their quality of experience (QoE) which will determine whether they churn (Ovum, 2017). Subscribers hold operators responsible for everything, from the content provider to their home, that affects their experience. When asked “What characteristics are important for a high-quality broadband service?” the top two responses were
- “100% reliable broadband service”
- “Good customer service”
When we look at the first of these, survey results (Incognitio, 2016) indicate that subscribers base their decision to recommend their broadband service provider on three factors:
- Service speed: 45%
- WiFi reliability: 31%
- Pricing and service bundling options: 21%
We can imagine numerous use cases to affect these key quality indicators (KQIs) and to which machine learning might be needed. However, appropriate data is not always available to support the use cases as significant portions of those data must come from the CPE and network devices themselves.
In this paper, we report on two examples of how machine learning can leverage data generally available from CPE and network equipment to address technical customer experience use cases.