Stephen Dilllon (Vodafone) – The Evolution of Customer Performance Through Big Data

This year we had a HUGE number of entries from many different industries for the Best Use of Data Science in a Large Company Award powered by FTI Consulting.

This Year’s Large Company finalists consist of Bank of Ireland, Depuy Synthes, Vodafone, Microsoft and Core Media, who’s teams were represented them in force at the presentation finals on August 17th in Croke Park.

Stephen Dillon, RAN Optimisation Engineer at Vodafone, has written this interesting and insightful piece on how Big Data is used to improve customer performance.

Here is what Stephen has to say:

The evolution of customer performance through Big data:

Telecommunications is an industry which marries well with machine learning and big data concepts. Data sets are large, with a medium sized mobile network (two to four million customers) generating between ten to one hundred terabytes of data per day. Traditionally mobile operators use key performance indicators (KPIs) generated by network elements (cells) to assess the performance of the network and the customers which it serves. Cell performance can be viewed as an aggregate of all the customers it serves. If a Cell meets performance KPIs it is inferred that the customers served by the cell will also have acceptable user experience, but this may not always hold true. The poor experience of a small percentage of customers may be masked by the good experience of the masses due to aggregation. Identifying and targeting customers with poor experience is key to improving customer satisfaction (measured by net performance score (NPS)) and churn reduction. Vodafone’s mission is to evolve current industry standards in the measurement of network and user experience by removing the aggregation of customer experience data. This will be achieved by developing KPI metrics at the individual customer level, leveraging existing (where possible) and new data sources, thus allowing Vodafone to achieve the required customer insights in an efficient cost effective manner.

Self-optimising network (SON) Evolution:

SON is a network optimisation tool which makes closed loop parameter changes to cells based on KPI breeches. Examples of Typical issues addressed by SON include tilt optimisation to increase the coverage area of cells, and neighbour additions to ensure users can move seamlessly between cells without suffering dropped calls. SON is an ideal platform to drive customer experience improvements and as such was chosen as the medium to begin our customer focused optimisation journey. Together with our SON partners (Cisco and CellMining) a project was proposed to address three use cases over a four-month period;

  1. Highway user experience, with the goal of optimising the experience of fast moving mobiles
  2. Enterprise assurance, to group and track the experience of large corporate accounts
  3. NPS detractor prediction, using machine learning techniques to correlate customer KPIs and attributes to NPS scores to predict potential NPS detractors in the network

New data sources were integrated into the SON platform to achieve the three use cases (an additional one terabyte of data per day).

The highway use case groups and extracts fast moving users KPIs and assesses their performance. Prototype SON algorithms were developed over a six-week period with the goal of improving the experience of highway users. A test cluster along the M1 (Dublin to Drogheda) was then used to verify the algorithms over a two-week period.

Improvements were seen across all relevant KPIs, and in particular a twenty-eight percent improvement saw achieved in dropped call rate, a metric crucial to highway experience (see Figure 1 below).

Figure 1: Highway optimisation use case

The Enterprise assurance use case grouped one large corporate account (eight hundred users). Performance of the account and individual users was monitored and optimised ensuring optimal account and user experience were achieved. A number of network changes were made to improve the experience of the account and individual users within the account. This use case will be further explored in phase two of this project (November 2017) where the monitoring of corporate accounts will be moved to the SON platform, ensuring that the experience of enterprise accounts is monitored in a closed loop fashion.

 

Understanding what drives detraction in a mobile network, and how to proactively address it, is a key mission for Vodafone. Leveraging machine learning techniques the NPS detractor prediction use case correlated twenty-five thousand NPS survey results with customer KPIs and attributes producing predictive models for detractors, neutrals and promoters. The model was then applied over the entire network to find potential NPS detractors. The eventual goal of this use case is to optimise potential NPS detractors via SON to ensure improved customer experience is achieved.

 

Vision:

Vodafone’s vision for data science in relation to customer experience management is to;

“Create a network with the customer at the core. A network where customer experience drives network optimisation to ensure our customers receive the best possible service levels which the network can provide, and to do this in real time, optimising as customer requirements and behavioural patterns change.”

To achieve this vision Vodafone Ireland are committed to driving innovation within the customer experience space, researching advances within our real time SON network optimisation tool to deliver our customers the best possible network experience. Our vision of a customer centric SON solution is well underway and next phase development will begin in November 2017. The next phase will further develop the Enterprise and NPS detraction prediction use cases by enhancing SON algorithms to assess individual customer KPIs in a weighted SON solution (Figure 2). Enhanced SON algorithms will ensure that individual customer experience issues are identified and addressed in a closed loop fashion. Weights will also be applied to customers depending on their quality of experience (e.g. potential NPS detractors will have a larger weight within the algorithms to ensure performance issues are addressed).

Vodafone's SON Vision

Figure 2: SON vision

The improvement of customer experience is an ongoing process and will continue to be a pillar of Vodafone’s strategy into the future.

You can be part of the DatSci Awards as well on the 21st of September in Croke Park, Dublin, Ireland. Be sure to get your ticket for a great opportunity to talk and learn from over 400 leading Data Science professionals in the Data Science community!

By | 2017-09-04T14:35:57+00:00 September 4th, 2017|

Leave A Comment