Telenor is a major mobile operator with over 150 million subscribers across Europe and Asia. It holds a 35.66% stake in VimpelCom Ltd, which operates in various markets in Europe, Asia, Africa, and North America, serving a total of 1.6 billion people. The document discusses Telenor's use of customer data for marketing activities, billing analysis, and gaining customer insights to better understand customer behavior and network usage. It also examines the evolution of adoption networks for various mobile devices and platforms like the iPhone and how social influences affect their diffusion over time.

1. Among the major mobile operators in the world
more than 150 million mobile subscribers *
31 000 employees
VimpelCom Ltd.
Present in markets with 1.6 billion people
Telenor Group
holds 35,66% of
the economic
ownership in
VimpelCom Ltd.
Russia
Ukraine
Italy
Europe
Kazakhstan
Georgia
Uzbekistan
Tajikistan
Norway
Armenia
Sweden Kyrgyzstan
C ambodia
Denmark Laos
Asia
Pakistan
Hungary
Bangladesh
Serbia Algeria
Thailand
Zimbabwe
Montenegro Burundi
Malaysia
C entral African
Rep.
A telecom perspective Bangladesh C anada
Kenth Engø-Monsen @ Telenor Group, Research and Future Studies
Pakistan
Workshop on time-varying network analysis, Sept. 19, 2012
India
*151 million customers in consolidated operations ; 360 million including VimpelCom Ltd (associated company)
What does Telenor need to do with Customer Data?
Marketing Billing &
activities Analysis
Customer Customer Insight & Social
Customer
ATL/BTL-activities acquisitio Prediction of network Billing
retention
n customer behavior analysis
The 360-
HANDSET DIFFUSION
Marketing of degree Context/net
Cross- & Up- Churn Member- Attribute- Producing
new products dissection of work-based
sell prediction Get-Member based insight bills
and services the insight! Collaboration with:
customers
Johannes Bjelland, Geoffrey S. Canright,
Rich S. Ling and Pål Roe Sundsøy
2 0 1 2.09.1 9 2 0 1 2.09.1 9

2. Anonymized CDR data—our starting point It is practical to define the adoption network
A number - Caller
1 2 3
Date & time
B number –
Receiving party
Type: Call, SMS, Data volume
Data, etc
Cell_ID: Location
IMSI: SIM card Social network Social network + Adoption network
adoption history
TAC: Handset
5 2 0 1 2.09.1 9
The iPhone adoption network evolution The iPhone adoption network evolution
Q307 Q407 Q108 Q208 Q308 Q408 Q109 Q209 Q309
2G release 3G release 3GS release
in US 39% 28% 50%
27% 38%
39%
0.1% 29%
27%
Mass advertising
NI: 36.7% NI: 36.9% NI: 44.7% NI: 51.3%
NI: 18%% NI: 45% NI: 52% NI: 50% NI: 37%
iPhone not yet
available in specific
market:
“Cracked iPhones”
bought in the 2 0 1 2.09.1 9
US. 2G 3G 3GS 2 0 1 2.09.1 9 2G 3G 3GS
Much traffic in US

3. =iPad user with iPhone
The DORO adoption network evolution
=iPad user with other
handset
Links are social relations based
on SMS+Voice
Q407 Q108 Q208 Q308 Q408 Q109 Q209 Q309
“The Apple Tribe”
Age distribution
5,00%
4,00%
3,00%
2,00%
1,00%
0,00%
• 54% of the iPad users also uses iPhone (5% market penetration of iphone) Non-Isolates
3.5% 4% 4% 4.9% 5.1% 6% 6.8% 10%
• If the iPad user is connected to another iPad user the chance of having an
iPhone is 72% Adoption of the Doro handset; an individual choice?
Or the choice of the user’s children who wish to be in contact with their
elderly parents?
Doro HandleEasy 326,328 Doro PhoneEasy 410 <25 25-55
2 0 1 2.09.1 9 2 0 1 2.09.1 9 Age:
Doro HandleEasy 330 Other Doro (338,345,409) 55-70 >70
The Mobile Video Telephony network evolution
iPhone users have high EVC
Q307 Q407 Q108 Q208 Q308
LCC: 4.7% LCC: 5.2% LCC: 2.6% LCC: <0.01%
Q408 Q109 Q209 Q309
2 0 1 2.09.1 9 2012.09.19

4. Apple vs. Android
You do what your friends do
If you have one friend with iPad,
your propensity to also buy iPad
will be 14 times higher.
Exploit the social circle to target
customers with high social product
Connected Connected Connected Connected Connected Connected
pressure
Apple users Android users Apple users Android users Apple users Android users
Q1 = launch Q2 Q3
Apple users form a ‘dense core’ of highly social users
2012.09.19 2 0 1 2.09.1 9
Building new weighted clustering measures
∑
Three factors:
wt ,1 wt , 2 3 wt ,1 wt , 2 wt , 3 Geometric
Connectedness
C= t
mean
Strength of T1/2 ∑ t
wt ,1 wt , 2
Relative strength of T3 to
T1/2 1  wt ,1 + wt , 2 wt ,1 + wt , 2 + wt , 3 
∑ 2

t
 2
+
3
at ,1at , 2 at ,3


Property: C=
wt ,1 + wt , 2
Reduce to the classical
unweighted version for
∑t 2 Arithmetic
Mean
identical weights
 
 1   w2 + w 2
2 2
   w2 + w2 + w2  
∑t  2   t,1 2 t ,2  + t ,1 t ,2 t ,3 
 at ,1at , 2 at , 3
    3  
   
CLUSTERING C=

wt2,1 + wt2, 2

Quadratic
∑ t
2 Mean
Collaboration with:
T. Binh Phan and Øystein D. Fjeldstad ‘Problem’:
Do we need more weighted clustering measures?
2 0 1 2.09.1 9 We need to figure out their properties!
2 0 1 2.09.1 9

5. The social network among customers
- On-net & Off-net
On-net customers Off-net customers
ON-NET AND OFF-NET
…. MISSING DATA
Who is an attractive customer?
2 0 1 2.09.1 9 2 0 1 2.09.1 9
Challenges / questions
• The ‘BIG’ question: When does a service take off?
Diffusion
• Who will adopt?
• Time-averaging over dynamic network data (current
approach) vs. true dynamic measures – Benefits /
Time-evolving value?
networks
• The problem of missing data (on-net vs. off-net)
• Multi-SIM behavior: ‘Identifying’ the same customer
• What are valuable structural properties of the
network (customer base)?:
Structure of • Centrality of customer set
social network: • Different roles: Originating and terminating
CHALLENGES • Which clustering measures are important?
• Competition with other network providers
2 0 1 2.09.1 9 2 0 1 2.09.1 9