Sie sind auf Seite 1von 9

WHITE PAPER

www.sybase.com
The Importance of Social Network
Analysis for Communication
Service Providers
Churn prevention and customer acquisition demands
new approaches to Business Intelligence
With commentary by: Sarah Wallace, Analyst at Heavy Reading
Social Network Analysis (SNA) is becoming popular among wireless operators, though many are wary of openly
discussing how they are using SNA due to potential backlash over subscriber privacy issues. SNA is a type of analytical
scoring system that enables communications services providers (CSPs) to analyze CDR and IPDR data to identify social
calling circles among friends and families. Such analysis will have both defensive and offensive implications.
From an offensive perspective, this analysis can help operators target or up-sell offers based on subscriber
preferences and the games or applications that are popular among the subscribers circle of trust.
From a defensive perspective, SNA allows operators to identify the key influencer through calling patterns in a
specific social group. This calling circle and community analysis identifies the impact of one consumers decision
to leave for another operators calling plan and the effect this has on the rest of a social circle.
Analytic algorithms are used to determine with a certain level of probability the likelihood, if an influencer churns,
for the rest of that social circle to churn as well. The next step is to determine which customers are worth keeping,
and how to actively execute compelling marketing campaigns to prevent them from churning. The churn event for
the rest of the social circle typically occurs within four to seven days after an influencer leaves for another provider, so
time is of the essence when providing proactive retention offers or other communications to potential churners.
SNA is a key tool enabling CSPs to execute compelling marketing offers or other communications to potential
churners within hours of the influencers churn event, and within days for the rest of the social circle. Before sending
an offer to a potential churner, additional data feeds need to be analyzed including handset analytics, customer
satisfaction and network performance, to name a few. In this way, SNA can be used in defensive and offensive
strategies to extend beyond churn to actual revenue generation, helping to identify which customers are best
targeted with which offers.
By targeting influencers, SNA also helps overcome the challenge of viral marketing or word of mouth promotion
(or criticism) of products. The algorithms used in SNA help analyze who are the key influencers, the ones that could
cause a ripple effect and possibly a negative viral spread of a product or service. This pinpointing of influencers helps
to optimize viral marketing efforts, making the entire process efficient and profitable.
An example of an influencer is head of household or the ultimate decision-maker for a family plan. Although
sometimes influenced by children who want the latest and greatest smartphone or messaging phone, this person
ultimately makes the final decision. Once the head of household determines that they will churn to a new operator,
the rest of the family plan will likely follow. An opportunity still exists for cross-product analysis to determine
whether that decision will affect the rest of a triple-play or quad-play bundle. By using SNA to analyze this event,
operators can use direct and immediate messaging to perhaps save the rest of the bundle.
SNA is becoming increasingly popular among operators who want to get a better understanding of customers
needs. This type of analysis of influencers is of utmost importance. SNA provides CSPs with timely up-selling offers
that come across as appropriate and not pushy. SNA helps operators use customer data accurately to track not only
an individual customers habits and preferences, but those of their social circle as well giving valuable insight into
customer trends.
In this competitive environment, operators must analyze messy data incoming from everywhere and everything,
including Customer Care, Product/Service/Device Portfolios, Cost and Billing, and Network Service Quality. SNA looks
at customers social habits and usage to gain an understanding of all these customer care elements.
To effectively deal with this massive amount of data, SNA requires a next-generation of Business Intelligence (BI)
architecture to capture, filter, clean, organize, analyze, and process incoming data and events as they occur, alongside
massive amounts of historical CDR data. This all-embracing understanding of churn probability and the next-
best-offer enables operators to do a better job in retaining customers and gaining new ones. In the world of CSPs,
traditional BI systems are hindered by looking at the past, analyzing only historical CDR data. With SNA, operators
need to look forward as well, in a predictive fashion. If applied intelligently, SNA will stifle churn by enabling
operators to find patterns buried within massive amount of call data in order to identify influencers and improve the
delivery of services to specific social networks.
INDUSTRY COMMENTARY
By: Sarah Wallace, Analyst at
Heavy Reading
Sarah Wallace is a Staff
Analyst for Heavy Reading, and
focuses on communications
services provider IT, including
all aspects of OSS/BSS. She
is particularly interested
in examining how service
providers will enable
customers to enjoy voice,
data, video, apps and gaming,
while at the same time trying
to survive in the competitive
landscape. Prior to joining
Heavy Reading, Wallace
previously served as an OSS
Analyst for RHK and as a
Senior Editor for TelOSSource
Magazine. She has worked
as a journalist for trade
publications in the call center
industry and has also served
as a writer/consultant in social
media. She holds a Bachelors
from The College of Wooster.
1
2
THE IMPORTANCE OF SOCIAL NETWORK ANALYSIS
WHY SOCIAL NETWORK ANALYSIS?
For communications services providers (CSP), retaining customers and cross-selling products and services to existing
customers is a top priority. Many studies show that annual churn rates now average between 3 to 4 percent a month
and retaining a client is only 1/5th the cost of acquiring one. Lowering churn rates will have a big effect on the bottom
line. Accordingly, many operators are shifting their Business Intelligence (BI) focus from that of customer acquisition
to customer retention. New BI techniques require operators to leverage the power of analytics technologies to look
forward, not just backward. Based on an analysis of massive amounts of customer data, what behaviors trigger churn
events? What actions have a higher likelihood of preventing a churn event? By using powerful analytical models to
detect potential churn in advance, and then executing proven next-best actions, operators are far more sophisticated
and efficient in their marketing campaigns and customer communications. These same analytical models can be
used offensively, enabling operators to more precisely target promotions and up-sell offers based on preferences and
behaviors exhibited among these interconnected social networks.
Figure 1. Depicted through KXEN, Social Network Analysis (SNA) is a type of analytical scoring system that enables communications
services providers (CSPs) to analyze CDR and IPDR data to identify key relationships among interconnected networks of friends and
families in order to better serve customer communities.
Social network analysis helps operators to identify individuals with the strongest influence over other members
within their particular communities. In most marketing disciplines, it is well understood that word of mouth remains
the strongest tool. Recommendations from experts and friends are often far more effective than advertising.
Smart operators use advanced analytics, combined with predictive techniques, to target key segments using
contextual advertising and behavioral targeting. This results in a better understanding of patterns of behavior within
communities, and also signals which influencers have the most credibility within a community.
This paper describes the
next-generation for business
intelligence and analytics
technologies specifically for
Social Network Analysis, and how
Sybase and KXEN are enabling
communications services providers
to harness and extract intelligence
from the massive amount of
customer data to reduce churn and
acquire new customers.
CHURN IS EXPENSIVE. A
TYPICAL TELCO OR CABLE
TV COMPANY SPENDS OVER
$200 TO ACQUIRE A NEW
BROADBAND OR PHONE
CUSTOMER. WHEN THE
CUSTOMERS SWITCHES,
YOU LOSE NOT ONLY THE
ACQUISITION COST, BUT THE
REVENUE THAT CUSTOMER
WOULD HAVE PROVIDED
YOU FOR A YEAR OR MORE.
CHURN COSTS AT&T, VERIZON,
COMCAST AND TIME WARNER
CABLE, TO MENTION ONLY
FOUR TELECOMS, BILLIONS OF
DOLLARS EVERY YEAR.
SOURCE: ELEVEN WAYS TO REDUCE
TELECOM CHURN, MIDDLETON
HUGHES, VICE PRESIDENT / SOLUTIONS
ARCHITECT AT KNOWLEDGEBASE
MARKETING.
SNA DEMANDS A NEW APPROACH
With any CSP, the biggest challenge with advanced analytics, especially predictive analytics, is dealing with
tremendous amounts of data. Social Network Analysis requires operators to look at not just one customer at a time,
but to analyze hundreds of behavior events across thousands or millions of customers. Effective predictive models
look at both live data streams and data histories for large numbers of customers, applying complex analytics to
determine the potential value of a customer or to offer a promotion with the highest likelihood of success. Added to
the challenge of managing and analyzing these massive data sets, is the need to do it quickly. The real value of Social
Network Analytics lies in the ability to precisely target cross-sell or up-sell offers to an appropriate influencer and
social circle in the moment.
Many operators have realized that the data management systems implemented in prior decades to support OSS/
BSS systems fall short in addressing these new requirements for Social Network Analysis. Using traditional BI tools in
combination with relational databases often ends up with a compromise between IT and their users forcing IT staff
to impose limits on the number of users able to access to data, for example, the complexity of queries they can run, or
the amount of historical data from CDRs users can analyze.
Figure 2. Traditional systems were not designed for todays analytics requirements.
This transition, from traditional BI and reporting to the advanced analytics techniques that underlie successful SNA,
is creating multiple challenges for IT departments. Traditional Enterprise Data Warehousing (EDW) infrastructures
now in place are proving inflexible in addressing the massive data volumes and real-time insights that SNA demands.
Clearly, a new analytics solution is required to support Social Network Analysis, one that can scale beyond advanced
analytics of massive data by a small group of users to an elastically expandable solution capable of delivering real-
time insight to a large number of front line staff in marketing, customer service, or billing.
SYBASE IQ, PURPOSE-BUILT FOR ADVANCED ANALYTICS
Unlike relational databases designed to support business transactions, Sybase IQ is a column-based analytics server
that is purpose-built for advanced analytics. Analytical queries, by nature, involve only certain attributes. Sybase IQ
stores and retrieves attributes by columns, rather than by rows. Since each column is stored separately, the data it
contains becomes its own index. In contrast, the typical row-based databases requires indexes and views to be built
on request; a time-consuming process. Moreover, Sybase IQ compresses and tokenizes the data, optimizing it to use
less space on disk and in memory. This dramatically reduces the input/output burden on the processor, memory, and
disk. With Sybase IQ, there is no need to create materialized views, cubes, summary tables, and multiple data marts to
get more speed out of the system. Together, these features produce query speeds up to 100 or even 1,000 faster than
traditional row-based solutions.
3
Traditional systems were not designed for todays analytics requirements
Detect patterns Limits on availability of historical data
Decades of detailed data
analyzed in seconds
Ad hoc and drill-down
analysis
Analysis takes hours or days,
not minutes or seconds
Interactive analysis of
real-time data to quickly
identify solutions
Real-time reaction
Unable to handle complex business
decisions based on real-time data flow
Instant enterprise-wide
data analysis
Quick re-programming IT backlogs, unacceptable delays
Business user driven
analytics minimizes IT
involvement
Users want to Todays reality Target capabilities
4
Deploying Sybase IQ as a purpose-built analytics server to improve query speeds while reducing systems overhead
represents only one part of the new BI and analytics architecture needed for SNA. It is important that the architecture
handle days, weeks, or months of CDR data together with live data feeds. By analyzing handset, network, or services
data as it is occurring, operators can better detect and respond to customer management KPIs.
Operators also need to exploit stream data in order to identify next-best actions for clients and their communities,
optimize business processes based on real-time intelligence, and make offers typically online when users are
likely to be most interested.
Figure 3. Unique real-time analytics architecture combines speed + depth of analysis
The in-memory Event Stream Processing (ESP) solution from Sybase derives intelligence from events as they are
happening by monitoring stream data from the network. By providing the brains to absorb, aggregate, correlate,
and analyze events, Sybase ESP can trigger an alert or an appropriate response to business-critical events, while also
providing high-level information on the current state of the business.
Sybase advanced analytics solutions combine Sybase ESP real-time monitoring, in-database analytics, and large-
scale data management. Sybase ESP analyzes high volumes of live data streams contained in systems such as routers
and network switches, while Sybase IQ mines massive amounts of data found in CDRs, and processes answers in
seconds. This combination of speed with depth of analysis is needed for effective SNA.
Sybase IQ includes PlexQ, a Massively Parallel Processing (MPP) architecture that accelerates highly complex queries
by distributing work to many computers in a multiplex grid configuration. Unlike shared-nothing MPP architectures,
PlexQ utilizes a shared-everything approach that dynamically manages and balances query workloads across various
compute nodes. PlexQ can improve the performance of a query by breaking it up into pieces and distributing those
pieces for concurrent execution across multiple server nodes.
Sybase advanced analytics technologies provide the foundation for Social Network Analysis, but predictive models
are needed to extract value from massive amounts of live and historical customer data. KXEN, a Sybase partner, helps
CSPs use predictive analytics to make decisions, extracting the data that is stored in Sybase IQ and Sybase ESP.
Advanced customer segmentation
LiIesIyle
Purchase criIeria
Usage paIIern
LocaIion, eIc.
Assign probability to future events
ProbabiliIy oI cusIomer churn
ApproximaIe producI, service or
device success raIe
lraud and revenue leakage
PrepaymenI percenIage, eIc.
Forecast impact and costs
of decisions
CosI Io launch a service
CosI Io serve cerIain cusIomer
segmenIs and launch services in
cerIain zip codes, eIc.
Advanced root cause analysis
keason Ior churn
Pre-empIive service assurance
CorrelaIion oI real-Iime evenIs
Service lmpacI Analysis, eIc.
Sybase Analytics in the Data Center
ApplicaIion
Servers
Near
keal-1ime
UpdaIes
Sybase lQ
Ops DaIa
OIher
source
(CDks)
Processing wiIhin
sec/min/hours
ETL
GSM Network
SGSN
GGSN
8SC
SIeam DaIa
Memory
Disk
EVENT STREAM PROCESSING
IQ
AlerIs
AcIions
Processing
wiIhin millisecs
SIream DaIa
Low LaIency
Real-Time Customer Intelligence Architecture
With this combined solution, CSPs can scale their environment to deliver carefully tailored promotions, campaigns, and
retention programs in order to improve the effectiveness of customer acquisition, cross-sell, and churn prevention efforts.
EXTRACTING VALUE FROM SOCIAL NETWORKS WITH KXEN INFINITEINSIGHT PREDICTIVE MODELS
Consolidation, ongoing regulatory changes and a shift to next-generation services have resulted in a proliferation
of products and pricing plans. SNA provides CSPs with a vital grasp of customer behavior in this highly dynamic
market. Predictive models enhance viral marketing, and also help with an overall increase in usage and ARPU through
targeted analysis. InfiniteInsight Social, KXENs solution for SNA, automatically builds social graphs from actual
customer interactions, whether it is a call, an SMS text, an email, or even a financial transaction. CSPs can accurately
target influencers, followers, and communicators by putting a value on specific customers social influence.
Figure 4: Example of a Social Graph depicted through KXENs InfiniteInsight. Dots represent communities (dense groups) of
customers. The bigger the community is, the bigger the size. Communities where groups of customers have an iPhone (iPhone
community) is in red. It is interesting to observe how close are such iPhone communities, probably due to a partially viral adoption of
the device. Further tests were made to prove that the iPhone adoption was correlated at user level and at community level.
KXENs InfiniteInsight enables
operators to optimize viral
marketing campaigns, helping
them to gain insight into customer
communities, strength of customer
relationships among individuals,
and the ability for individuals to
influence others.
5
6
Defensive Strategy: Reduce The Risk of Churn
Predictive analytics is an effective weapon against churn and it is imperative to have an early warning system
to identify those customers at risk of churning and suggest offers that can help retain them. KXEN builds churn
predictive models to pro-actively detect the customers who will churn by including social variables such as churn rate
in communities and social pressure.
TPSA/Orange Poland built social networks and communities from 4.6 billion voice call records, representing 130
million distinct monthly links between customers, in just two hours. This improved churn model accuracy by 47%
using social community attributes.
Figure 5: KXENs InfiniteInsight helps detect communities of high churn risk. The red circle represents the detected community, the
blue icons represent customers (subscribers), and the red icons represent customers having churned last month. The blue icon with
the red cross (N6048988319) is an off-net, e.g. a non-existing customer. Four customers (5391, 5060, 3053 and 3932) represented in
blue, receive a strong churn pressure because they belong to a community of 8 people, in which 3 churners and 1 is an off-net. These 4
customers are at risk of churning (represented by an red arrow), and must be treated with care through dedicated retention strategies.
CHURN MODEL ACCURACY
IMPROVED BY 47% WITH
INFINITEINSIGHT SOCIAL
FROM KXEN AT TPSA/ORANGE
POLAND
Telekomunikacja Polska, part of
France Telecom Orange Group,
is the largest fixed-line provider
for voice and broadband services
in Poland. TP needed to explore
and define effective ways to
predict churn starting from
the CDRs derivatives. TP used
InfiniteInsight Social to assess
the value of SNA in empowering
customer retention strategies. The
prediction of churn was based on
four main types of analysis: churn
measurements, psychographic
analysis, geographical churn
analysis, and social network
analysis. InfiniteInsight Social was
used to detect communities of
networkers, bridges, leaders and
followers. The role of subscribers
within the communities, as well as
the other social variables created
by InfiniteInsight Social, were
then used to rebuild the churn
prediction model.
The contribution of the
InfiniteInsight Social improved the
accuracy of the churn predictive
model by 47%. Using a single
server, InfiniteInsight Social was
able to smoothly work on call
detail records (CDR) derivatives in
order to prepare links and detect
communities within the subscriber
base in two hours. All this was
done with an easy user interface
and short elaboration times.
Optimize Viral Marketing by Identifying Influencers
Viral marketing is a powerful marketing tool to identify key players or influencers that have an impact on marketing
efforts, but it is important to realize that viral marketing needs to be continually monitored and optimized. Timeframe
is critical with SNA and is very dynamic influencers and communities change from one day to another.
To be effective, CSPs need the flexibility and scalability in their architecture to analyze months, weeks, or days of
CDR information and generate the results in a timely manner.
By looking at social influencers around specific business questions such as churn, handset adoption, or service
adoption, CSPs can identify high potential of diffusion and can identify the people who will churn next month and
possibly take three friends along with them. Major European and Asian mobile operators have seen uplift in viral
marketing campaigns through SNA techniques, specifically with product adoption rates and churn rates, and have
experienced and overall increase in usage and ARPU.
KXEN provides an approach that includes more than just the transactional records between individuals, it bakes in
customer attributes and centrality measures structure-based indicators to let CSPs hone in on social communities and
understand influencer power and social pressures.
Figure 6: KXENs InfiniteInsight identifies influencers within social groups
Offensive Strategy: Acquiring More Customers With SNA
The combination of high penetration rates and high customer acquisition costs is driving CSPs to find more efficient
ways to locate new customers. Regardless of the channel (cloud, website, or online services), new levels of speed and
flexibility are needed to personalize acquisition campaigns.
InfiniteInsight Social builds social networks and obtains the list of off-nets (subscriber of a different network) that
are highly connected to on-nets (current customers), by detecting communities of high on-nets for example with
high revenue. Using predictive models built with InfiniteInsight, targeting the off-nets communities become much
easier. For example, using member-get-member (MGM) and refer-a-friend become communities to target for product
up-sell and cross-sell campaigns.
A major Canadian communication provider improved email open rates by 20% and achieved higher conversion rates
among social leaders in its refer-a-friend campaigns.
KXEN has helped optimize viral marketing campaigns for many communication providers. Not only can predictive
analytics cause an uplift in viral marketing but can help with an increase in usage and ARPU through targeted
SNA analysis.
7
www.sybase.com
Sybase, Inc.
Worldwide Headquarters
One Sybase Drive
Dublin, CA 94568-7902
U.S.A
1 800 8 sybase
Copyright 2011 Sybase, Inc. All rights reserved. Unpublished rights reserved under U.S. copyright laws. Sybase and
the Sybase logo are trademarks of Sybase, Inc. or its subsidiaries. indicates registration in the United States of
America. SAP and the SAP logo are the trademarks or registered trademarks of SAP AG in Germany and in several other
countries. All other trademarks are the property of their respective owners. 07/11
iPhone is a registered trademark of Apple, Inc.
THE BENEFITS OF SOCIAL NETWORK ANALYSIS FOR CSPS
Converting the increasing amounts of subscriber and network data to determine the next best offer is a difficult
task that service providers need to tackle if they want to meet their business objectives, which center around
delivering superior customer experience and products to reduce churn. The right analytics architecture will play a
pivotal role in the success of operators, not only providing them with real-time intelligence, but also helping them to
maximize their revenue potential during short windows of opportunity.
Sybase advanced analytic solutions infuse organization with fast, flexible access to information. Coupled with KXEN
solutions, CSPs can pro-actively detect the customers who will churn and find efficient ways to retain them, while also
enabling more pinpoint offers and promotion to attract new customers.
The ability to analyze influencers and social circles is of utmost importance for CSPs. As Heavy Reading suggests,
once a head of household churns to another provider, in all likelihood, so will the rest of his or her family, and within
days, the rest of the social circle may churn as well.
SYBASE IN TELECOMMUNICATIONS
Today, Sybase serves hundreds of communications service providers of all types in North America, Europe, Asia, and
Africa. These providers depend on powerful Sybase technologies to answer questions that were previously impractical
or impossible, and deliver the business-critical information they use to ensure efficient network utilization, customer
satisfaction, and service reliability.
More than 350 operators are using Sybase technologies today to dramatically reduce the time to answer for
complex network optimization problems such as traffic engineering, capacity planning and routing, reducing
operating expenses and avoiding unnecessary capital expenditures.
Communications service providers (CSPs) rely on Sybase technologies to reduce churn and increase revenues
by answering customer assurance questions, making Business Support Systems (BSS) and Operations Support
Systems (OSS) data available to CSRs. Because queries that used to take hours now take less than a minute, they
increase first-call resolutions and reduce customer care costs.
Wireless and wireline service providers, MVNOs and cable operators are using Sybase solutions to increase ARPU
by identifying their most profitable customers, segmented by their behavior and network usage, and making the
right up-sell and cross-sell offers at the right time. They also compare network usage and service offering data to
identify profitable gaps in their tiered offerings, and grow revenues and market share.
The Sybase suite of mobile operator solutions powers messaging and management offerings from many of the
biggest names in telecommunications.
KXEN, THE PREDICTIVE ANALYTICS LEADER
KXEN is revolutionizing the way companies use predictive analytics to make better decisions. Based on patented
innovations, the companys flagship product InfiniteInsight delivers orders of magnitude improvements in speed
and agility to optimize every step in the customer lifecycle including acquisition, cross-sell, up-sell, retention and
next best activity.
For more information, visit
www.sybase.com/iqfor4g
and www.kxen.com
SNA BENEFITS
SNA is an accurate way of using
customer data to track, not
only an individual customers,
habits but their social circle and
communities as well, giving
insight into customer trends.
SNAs analytical scoring system
takes the guess work out of
possible churn.
SNA offers an opportunity
for more targeted up-sell
opportunities which come
across as timely versus invasive.
SNA is even more beneficial
to an operator who is trying
to become more targeted
in its selling and customer
relationship management.

Das könnte Ihnen auch gefallen