Franck Gallos de la société Ericsson enchaînera sur l’analyse des usages des services d’IP TV des grands opérateurs Telco. Franck détaillera comment la corrélation des données des usages IP TV avec des informations externes comme les données météorologiques ou sociales (événements politiques, sportif, vacances scolaires) permet de contextualiser les statistiques géo localisées pour un meilleur ciblage publicitaire. A noter que ce projet est arrivé second au Trophée de l’Innovation Big Data Paris 2014.
Hadoop User Group, le 11 Juin à la Tour Eiffel avec Infotel
2. Ericsson- HUG Paris | 2014-06-11 | Page 2
Agenda
1. Ericsson networked society & analytics
2. Projet Media statistics
3. Exemple : IPTV & meteo
1. Influence des facteurs métérologiques
2. Calculs prédictifs
3. Ericsson- HUG Paris | 2014-06-11 | Page 3
Operator’s challenge and opportunity
Growth in mobile broadband, users and usage 2013 to 2019…
250 M to
750 M
PCs and tablets
6.7 BN to
9.3 BN
Mobile subscriptions
1.9 BN to
5.6 BN
Smartphone
subscriptions
2.1BN to
8 BN
Mobile broadband
subscriptions
10 times
45% CAGR
Mobile data traffic
Growth opportunitiesData boom customer care Calls
Problem Resolution
time Source: Ericsson Mobility Report
4. Ericsson- HUG Paris | 2014-06-11 | Page 4
Telecoms trails behind other industries in
brand loyalty
NPS Benchmark for US industry groups 2012
Source: Informa Telecoms
NPS is a focus for telecom operators
5. Ericsson- HUG Paris | 2014-06-11 | Page 5
improving customer experience across the lifecycle is
crucial to increase NPS
Key factors driving
NPS
Improving customer experience
across the entire lifecycle is crucial
By breaking down loyalty drivers, we can
understand which areas that are most important
for improving NPS
Source: Keeping Smartphone Users
Loyal, Ericsson ConsumerLab 2013
GET
Simplicity, clarity,
personalization
FIND
Availability, variety,
relevance, transparency
SET UP
Accuracy, speed,
efficiency
USE
Speed, quality, accessibility,
reliability
GET HELP
Accessibility
Speed
Resolution
PAY FOR
Cost control
Simplicity
MODIFY
Simplicity, clarity,
personalization
6. Ericsson- HUG Paris | 2014-06-11 | Page 6
Big data analytics is key to
boost customer experience
HOW OPERATORS RESPOND
Idea-to-
Implemen
tation
Plan-to-
Provision
Lead-to-
Service
Service-to-
Cash
Experience
-to-
Resolution
PREPARATION
DELIVERY OF
CUSTOMER’S
DESIRED EXPERIENCE
BIG DATA ANALYTICS
TAKE ACTON
GAIN INSIGHT
MEASURE
FIND
Availability, variety,
relevance,
transparency
SET UP
Accuracy,
speed,
efficiency
USE
Speed, quality, accessibility,
reliability
GET HELP
Accessibility
Speed
Resolution
PAY FOR
Cost control
Simplicity
MODIFY
Simplicity,
clarity,
personalization
WHAT CUSTOMERS WANT
7. Ericsson- HUG Paris | 2014-06-11 | Page 7
ANALYTICS WITH ERICSSON
An agile, open, multi-vendor approach
that converts big data
and domain knowledge into
real-time, actionable insights for a
wide range of use cases
Terminal Probes
and DPI
RAN
Traffic
Nodes
Core
Traffic
Nodes
Control
Plane
Product
&
Service
Catalog
Fault and
Performan
ce
Trouble
Ticket
Charging
& Billing
CRM Social
Network
Marketing Engineering
Customer
Care
Networks
Mediation / Correlation / Filtering
Knowledge Extraction / Business Logic / Data Mgmt
Exposure / Insights / Action
White Paper :http://www.ericsson.com/news/130819-big-data-analysis_244129227_c
8. Ericsson- HUG Paris | 2014-06-11 | Page 8
Agenda
1. Ericsson networked society & analytics
2. Projet Media statistics
3. démonstration
4. Exemple : IPTV & meteo
1. Influence des facteurs métérologiques
2. Calculs prédictifs
9. Ericsson- HUG Paris | 2014-06-11 | Page 9
Ericsson
Media Statistics
Media statistics
User infos
CSP, type,…
Director
Actors
Analytics
Logs, STBs
Metadata
Open data
Comedy
Action
…..
Category
Data report
Reco. Engine
Advertising
Systems
(box,portal,…)
10. Ericsson- HUG Paris | 2014-06-11 | Page 10
Big data project
› Cluster Hadoop
› CORE IPTV
› CDN LIVE/VOD
› METEO
STATIONS
VOLUME VELOCITYVARIETY
› 16 To
› Terasort benchmark for 1 To Data Volume with a response time of 04’28s
12. Ericsson- HUG Paris | 2014-06-11 | Page 12
Agenda
1. Ericsson networked society & analytics
2. Projet Media statistics
3. Exemple : IPTV & meteo
1. Influence des facteurs métérologiques
2. Calculs prédictifs
13. Ericsson- HUG Paris | 2014-06-11 | Page 13
CROSSING TV & WEATHER
› 1 To of raw IPTV data
› Data Meteo : 100 weather stations
› 6 months of weather and IPTV logs data
› Response time= 00:25:00
14. Ericsson- HUG Paris | 2014-06-11 | Page 14
WEATHER INFLUENCE On VOD
Consomptions
› Monday Tuesday Wednesday thursday
› friday saturday sunday
› Such consomptions influences are not taken into account by marketing divisions
› Such usage analytics data could now feed any recommendations operational systems
15. Ericsson- HUG Paris | 2014-06-11 | Page 15
TOP « ONE day weather » INFLUENCE
ON VOD USAGE
› This Map displays the ranking
influences between French locations
› The lists indicates the names of the
most influenced by weather French
departments on Thursdays
› THURSDAY
rank department
1 Côtes d'Armor
2 Finistère
4 Ille-et-Vilaine
8 Loire atlantique
9 Mayenne
10 Morbihan
11 Maine et Loire
12 Vendée
rank department
3 Corse
5 Var
6 Hérault
7 Bouches-du-Rhône
16. Ericsson- HUG Paris | 2014-06-11 | Page 16
TOP « ONE day weather » INFLUENCE
On VOD usage
› Week end departures times in Paris
suburb and weather do influence the
VoD usages
› FRIDAY › SUNDAY
› Weather does not influence people
on watching VoD when returning
back home on Sundays in the north
of France while it does in the south
17. Ericsson- HUG Paris | 2014-06-11 | Page 17
PrEDICTIVE ANALYTICS
› Total numbers of VOD sessions per Sundays from July to December 2013
› Post-predictions validations obtained from 1To data volumes
› Response time= 00:25:00
18. Ericsson- HUG Paris | 2014-06-11 | Page 18
CAPACITY PLANNING
› Total numbers of VOD sessions for each Sundays from July to December 2013
› Linear growth differences between real numbers and forecasted ones is 3%
› The methodology used for this use case can be applied to individual TV
channels (ex : by planning capacities thanks to EPGs for instance… )
Slide no1: Growth in Mobile Broadband
Let’s look at some numbers showing the tremendous growth in numbers of users and usage of mobile broadband.
Subscriptions for mobile PCs and tablets are expected to grow from 250 million in 2012 to 750 million by 2019.
Total smartphone subscriptions: Will reach 1.9 billion at the end of 2013 and are expected to grow to 5.6 billion in 2019.Around 1 billion smartphones were sold in 2013, representing close to 60 percent of all mobile phones sold in Q4 2013.
Mobile data traffic is expected to grow at a CAGR of around 45 percent (2013-2019). This will result in an increase of around 10 times by the end of 2019.
Global mobile broadband subscriptions: In Q4 2013, mobile broadband subscriptions grew by around 150 million to 2.1 billion (40% year-on-year increase). Are predicted to grow to 8 billion by 2019.
Total Mobile subscriptions: 6.7 billion mobile subscriptions globally in Q4 2013. By the end of 2019 it’s expected to reach around 9.3 billion.
While the pace of change may seem fast, in reality it will never be this slow again.
This data boom means the end-user behavior shifts from being VOICE CENTRIC to DATA CENTRIC and creates a number of challenges for the operator.
NPS – Net Promoter Score = Measures the loyalty that exists between a Provider and a consumer
-Considered to be a significant metric to measure the operators’ CEM initiatives
-Easy to survey and analyze
-In combination with other metrics. NPS is an effective final dependent variable in a customer data model.
It is more important to understand what the factors are that make a customer a promoter or detractor than simply to know how many promoters an operator has.
Operators are starting to use NPS as a way to understand the impact that their different touch points have on customer expectation via a simple metric
NPS still evolving from a general overall score into an individual analysis. It needs to be analyzed in conjunction with other metrics across the business.
Focusing on the Right Use Cases, Driving Business Results
Delivering Actionable Insights, Based on Domain Expertise, Realized in User-Specific Applications
Empowered by an Agile, Real Time, Horizontal Big Data Analytics Platform
Multi-Vendor, Multi-Data, Multi-Purpose
Securing Customer Privacy