Harnessing big data for effective and cost-efficient CEM
Of all the technology changes in the past three years perhaps the most significant has been the revolution that is underway with data. So called “big data” has enabled progressive enterprises to act smarter, be operationally more efficient and treat their customers better.
At the heart of the revolution has been the rapid acceptance of open source technologies collectively referred to as Hadoop. Hadoop clusters and technologies are being adopted by CSPs to support their current data needs taking advantage of the lower cost infrastructure and its ability to scale to support large data sets, says Marc Price, VP Technology, Openet.
But adoption is just beginning. On the telco side, a recent report from McKinsey suggests that relatively few telcos have adopted big data architectures, and associated analytics technologies, to derive meaningful benefits. However, a small group has achieved outsized profits from such investments.
The case for CEM
The telecoms industry is in the unique position of having an abundance of data at the CSP’s disposal to gain an understanding of all subscribers’ customer experience. However big data challenges of volume, variety and velocity of data are more significant in telecoms than other industries. They must be resolved with solutions fit for purpose to avoid unnecessarily costly and unmanageable environments.
Finding a strong use case for big data is the most likely route to gaining credibility within a CSP. One of the strongest drivers has been the use of Hadoop for customer experience management (CEM). CEM requires the combining of data from many different sources to create a 360-degree view of each subscriber. This requires the use of very large data sets and the running of complex models to understand different aspects of the customer.
Big data technologies, including technology covered within Hadoop distributions, provide an important role in supporting an analytics infrastructure that scales to viably meet the rigorous demands at CSPs.
Big Data challenges
CEM looks to understand a customer through the collection and analysis of data points that systems, surveys, interactions or external data have captured about a subscriber. This data is used to monitor each subscriber and the service performance, and by so doing enable CSP’s to manage each customer’s experience.
But these CEM projects will need industry specific knowledge to ensure that only relevant data is captured, from the right data source with the correct frequency needed to support the CSP’s requirements. This can be helped by the use of industry specific solutions for data preparation.
One of the problems with big data projects is that historically data preparation typically takes 50-70% of a big data analytics project’s time and effort. With data volumes and complexity increasing it’s clear that CSPs will need to examine methods and systems to reduce the timescales and effort needed in data preparation.
Knowledge is power
Specialisation in telecoms data is as key here. As telco data is unique, having a deep understanding of this data and the business drivers specific to the industry serve to reduce the workload of data prep. This includes the cyclical effort of finding data, building models and refinement by supporting the more common business requirements, and having knowledge of what different data sets to collect and how to collect them.
Coping with the big data challenge requires specific industry knowledge to reduce and right-size data volumes, pick out the data attributes that need to be processed in real-time and normalise the data to reduce the multiplicity of different data types. This right-sizing of data can then provide richer and useful data sets to analytics.
The ability to reduce data volumes and the corresponding associated costs are done at the data preparation stage in each big data analytics project. Although costs of data storage and compute resources are falling, growth in data volumes is outstripping any savings. It is therefore important to manage and reduce the impact of the data volumes to prevent increasing costs associated with data storage and analysis.
The entire scope of a CEM solution is daunting for CSPs, who understand the implications of gathering and analysing very large data sets. The cost of creating a fully scoped CEM solution that covers the complete customer journey requires most systems and departments in a CSP to participate in providing data and accepting automated actions. The cost justification for initiating a CEM project has to therefore be very clear.
According to a recent paper by Analysys Mason, having happy customers pays dividends for communications service providers. From the revenue uplift (happy customers have been shown to buy almost 2 services from a CSP, versus unhappy customers buying 1.5 services), to churn reduction (unhappy customers are 2 to 2.5 times more likely to defect to alternative providers than happy customers), to increased usage (happy customers consume about 20% more goods and services than customers who are reported as unhappy or detractors).
CEM projects are potentially unwieldy if all aspects of customer experiences are included, so CSPs need to focus initially on the areas of highest impact, specific data sets and a subset of subscribers.
The author of this blog is Marc Price, VP Technology, Openet.
Comment on this article below or via Twitter: @ VanillaPlus OR @jcvplus