Big Data In Telco
Introduction
It is not the strongest that survives, nor
the most intelligent, it is the one which most capable of adapting to change -
Charles Darwin
Big Data has been the latest fashionable
technology for quite some time now. While its applications and potential
applications are no longer a doubtful, what still continues to perplex users is
how relevant it can be to their business and more importantly how to
conceptualize and build a Big Data stack that is suitable?
From a professional angle, I have been
participating in regular engagements with Telco enterprises, trying to find out
reasonable ways to push Big Data based solutions to them, and have observed
certain key characteristics
ONE. In traditional Telecom the
scope of solution is usually big, often complicated by diverse requirements set
by the standard bodies and Govt. authorities. This conventionally prevented
many players to enter the playing fields and operators were stuck with only a handful
of well-known vendors. Growth of interest in Big Data and the technology to
process the same transformed the playing fields, because it no longer remained
a Telecom problem but became a Computer Science/IT problem like choosing the
right Database or Operating System. The Playing Fields were made more flat
making entry of several vendors large general purpose one to small niche ones
alike
TWO. There is no standard for Big Data.
This is a disadvantage as well as an advantage. Disadvantage because solutions
need to depend largely on precedence and best practices, and often on experimentation
before the right direction can be found. Advantage because it keeps the lines
of innovation open and space for both Open Source and Proprietary solutions. This
situation also leads to an environment where customers might have any strong
keenness for the exact technology as long as some basic expectations are met
a) Cost Effective: In most cases
because of lack of a clear and complete use cases, customers will want to reuse
their existing hardware, software and human infrastructure. Cloud Hosting – pay
as you use – with zero setup expenditure is an inviting option for most.
However vendors interested in selling infrastructure often try to deter them by
bringing up topics of security and performance. But those are really secondary
concern as long as you are not running some time critical business or
accidentally leaking key privacy information of your subscribers into the
public.
b) Avoid Lock In: This is why Open
Source based solutions keep winning. In the Big Data space, Open Source has
continued to define the de facto standards and vendors adopting Open Source
components in their solution automatically gain some advantage. By remaining
compatible with the Open Source distribution in terms of the Interfaces, the Configuration
and (though often overlooked) the Data Format, vendors offer an assurance that customers
are not locked in and in case they find a strong reason, can always replace the
core engines.
c) Mix,
Match, Integrate: Customers have more or less agreed that there is no
comprehensive Big Data solution and it is always pragmatic to mix and match the
most popular and appropriate software for each category (eg. storage,
computing, analytics, real time computation, reporting, visualization and so
on). It is also unlikely that customers will throw away their previous and time
proven investments in similar tools (like BI Tools, Analytics Workbench,
Database etc.) and adopt a big bang change. More likely than not, they will
think of integrating these tools with a compatible ‘Big Data Core’, where they
could retain the same familiarity of older tools, and yet expanding their
capability to ‘logically’ crunch larger amounts and variety of data
THREE. Telcos have large amount of
data both from machine sources (KPIs from network probes, signaling data,
transactional records and so on) and human sources (customer data, interaction
data, behavioral information and so on). Some of this information is very unique.
For example mobility information (while it is fairly possible to spoof your IP
address, it is virtually impossible to anonymize the mobile cell area you are
in), customer demographics (age, gender, occupation, country and so on because
in most cases Govt. regulations require such information to be furnished to the
Telecom operator for protecting both the consumer and the service provider), Financial
indicators (Billing Data, Mobile Money etc.) and so on. The key problem to
solve is how to use this information to promote consumption, increase retention
and improve operational efficiency
Telco
Perspective
The
key problem Telcos face is the lack of a killer service like voice or messaging
which can stand out in their Big Data offering. Big Data technologies thus
remain invisible to the common consumer in terms of variety or newer services
offered to him/her and spent instead to provide the conventional services with
better experience. In other words there is less innovation in the actual
service being offered but more on how it is offered.
Three key patterns can be observed
a) Promote Consumption: The
general approach is to increase consumption by users by providing them the right
kind of service stack. It does not necessarily mean developing new services but
finding out suitable offers which will gain acceptance and hence drive more
consumption. Personalization and Contextualization are the key factors here. Recommendations
(based on preference, mood, locality, usage), special tariff, partnership with
OTT service providers are some ways Telcos can boost the consumption.
b) Increase Retention: Unlike many
businesses, Telcos mainly thrive on regular revenue assurance from its user
base and hence retaining them becomes highly significant, especially when
attracting a new user is costlier by several magnitudes in view of slow service
innovation, customer inertia and slow onboard time due to physical barriers
involved (eg. point of sale interactions, device interactions etc.). While
customer churn is protected in some ways because of these constraints and
improvement in Quality of Service guarantee, attractive tariff plans
rewards/bonus etc. help to increase customer retention and any eventual churn
possibilities.
c) Improve Efficiency: Increasing
efficiency mainly concerns about saving money. Its main purpose is to help to
do more or at least the same things with lower expenditure. One of the key
aspects of efficiency improvement is cutting down on the waste activities, and
this is where Big Data has been able to be proven effective. It is now possible
to identify with reasonable accuracy the potential buyers, exact pain areas, detailed
behavior patterns and so on. Improving efficiency is reflected into activities
such as network optimization (eg. re-distributing cell areas or BTS antennas
based on traffic pattern), Farud Management (eg. detecting habitual churners), IT
management (eg. commodity hardware vs. custom hardware) customer care (eg. 360
degree view of the customer to serve him/her better), precise marketing (eg.
avoid selling products to irrelevant groups) and so on.
The
State of Telco Big Data
The
ultimate aim of any business is to grow the revenue and reduce costs. Big Data
technologies have been quite effectively used to analyze data, identify waste
in infrastructure and processes and remove them. So in a way they have proven
to be useful in ‘Reducing Costs’. However when it comes to ‘Generate New
Revenue’, they kind of fall short in terms of new service offering and depend
almost entirely on promoting consumption of the existing ones.
In fact Telcos don’t have too many products
which can be called their own. As a result Big Data application in most cases
has been limited to realizing recommendation and personalization services. Such
as personalization of Value Added Services – such as Ring Back Tone, Mobile
News etc., Tariff Plan recommendation based on user’s voice, data or message
consumption history or carrying out intelligent advertisement campaigns. The
lack of native services, prompt Telcos to ‘resell’ or have ‘symbiosis’ with OTT
services like instant messaging, entertainment, maps, advertisers etc.
Another key bottleneck Telcos face today is
inadvertent ‘Data Leak’. While the operator can theoretically capture and
analyze data happening through their channels, there is rarely any deep
correlation and exploration of data generated at various sources. To give an
example, most Telcos for example are also ISPs (Internet Service Providers) and
can theoretically capture the consumer’s wireline Internet behavior and
recommend matching services in the wireless domain. Imagine a user who exhibits
a pattern to watch videos on his/her fixed broadband and may be enticed into a high
speed 4G connection with a special discounted tariff plan or a trial pack
knowing that he/she has a high chance of ending up subscribing.
Vendor
Perspective
Traditional Telco vendors have the biggest
advantage of deep knowledge of the Telco Data Science as well as the Telco operators
and their business problems. Hence the key value from Telco Vendors can come by
exploiting these aspects and not competing with pure Technology vendors
who are in the business for a longer time and more mature. In fact Big Data is
potentially emerging as a general purpose technology like OS, Database,
Programming Language and so on and in the end there will be only few key
vendors. However a thriving ecosystem is still possible when companies start to
focus on the applications side, to utilize the Technologies and Tools to
address the business needs.
- Telco Vendors are no different. They can create values by providing out of the box tools to support the business of Telco Operators better for their specific day to day use cases.
Some common cases may be:
- Develop and Provide prediction models for customer churn, product recommendation, service recommendation, fraud detection, network optimization etc.
- Share data and events as commodities to third party applications (OTT providers)
- Cooperate with OTT providers to provide rich experience to the consumers, data comes as a tool
- Offer innovative services and improve quality of operation by correlating multiple data sources fast
- Enable real time decisions to provide instantaneous personalized services
...... and so on
Comments
Post a Comment