Improving the Orange customer experience through mass data processing and analysis

INTRODUCTION
HOW
PIPELINE
KPIs / KQIs
CEI / NEI
TPT / HWLOCATION (HOME / WORK LOCATION)
USES CASES
ARCHITECTURE

Analysis, design and construction of the CEM (Customer Experience Management) project.

> 20 M

users

> 15 TB

per day

> 400 K

events per second

Orange is one of the leading telecommunications companies nationwide. It provides global mobile, landline, internet, mobile internet and IP television services.

In order to better serve its customers and reduce the churn rate, it was keen to manage and improve the user experience.

CHALLENGE

Be able to manage the user experience

The aim was to leverage the data generated by users when interacting with their mobile phones to analyse it and determine the user experience in order to improve it.

WHAT WE DID AND HOW WE DID IT

Analysis, design and construction of the new big data CEM (Customer Experience Management) solution

The main source of information was the data generated by the sensors deployed on the Orange network which send information every minute on the different transactions/events (voice and data) of each user of the Orange mobile network.

The designed solution consisted of harnessing a series of performance and quality indicators (KPIs/KQIs) from the ORANGE mobile network to use to generate a Customer Satisfaction Index (CSI).

The created architecture comprised the following structure:

Apache Kafka is a distributed streaming platform used to receive the information generated by sensors on different topics in real time.
Apache Flume is a distributed service that reads the information left by sensors on Kafka topics to move them in real time to Data Lake (Hadoop and HDFS).
The Hadoop Distributed File System (HDFS) is used for data storage (Data Lake).
Spark is an analytical engine for processing large data volumes, with modules for streaming, SQL, machine learning and graph processing harnessed to process data stored in Data Lake and to calculate different KPIs and KQIs. All in order to generate the Parquet files to subsequently create boards with HIVE.

HIVE is a data warehouse infrastructure that provides tools for data analysis and query that we can use to create the different boards with the KPIs and KQIs already generated for subsequent use with MicroStrategy.
MicroStrategy is a data analysis platform providing all the functions to design and develop analysis and mobility apps to access the boards created in HIVE and create the different dashboards.