Post on 16-Jan-2017
transcript
Variety
Old Paradigm – Small Data New Paradigm – Big Data
- Limited volumes processed- Terabytes- Hardware defined processing
- Full available Data Set processed- Petabytes- Data can be processed in cloud and
mostly software defined
- Analytics for basic reporting, segmentation, network planning
- Limited data sources used- Unstructured data is mostly unused
- Analytics used widely for prediction and recommendation
- All available sources used- Unstructured data processing is
hugely utilized for data refinement
- Limited speed of processing- Hours and Days- Waterfall PM, slower time2market
- Unlimited speed of data- Seconds and Hours- Agile PM, Fail Fast approach
- Poor range of formats being processed
- Difficult to check the quality- Poor data protection that can hurt
quality
- Any format of data- Data quality cross check- Full-scale depersonalization and
ultimate protection
WHAT IS BIG DATA FOR TELCO? SHIFT FROM OLD PARADIGM TO A NEW ONE
Volume
Velocity
BIG DATA IS A POOL OF ACTIVITIES
intended at
processing the data a company owns (internal and external)
so that to open new revenue opportunities,
minimize costs
and enhance UX.Veracity
Data Source Source BriefCurrent Value
Extraction
Difficulty of
Extraction
Difficulty of
ProcessingPotential Value
Billing logsCall details, Traffic, Revenues, Balance, Debt, Services
used, ARPU, MOU, Age, Gender, Roaming 3
1 1 4
Radio Network, Call Tracing
SystemsPoint of Interest, Location Analysis, Real Time
Tracking, Frequency of visits 2
2 3 5
SMS dataSender's numbers (including B2B senders), Semantic
and Sentiment analysis, 1
1 1 3
Device Management Systems History of devices, Functionality, Cost, Brand 2
1 1 3
DPI, Gn/Gi/S1Type of data traffic, Applications, OTT usage, Pages
visited, Search quiries, Apps installed/used, Page 2
3 4 5
Call Centre InfrastructureCall Center Logs, Call Center Speech, Complaints,
Requests, Profiles Refinement 1
3 3 3
NetworkNetwork logs, Signalling data, Network faults data /
Incidents 3
3 3 3
ERPOrders, Procurement, Corporate Documents and
interaction 1
1 1 1
IOT infrastructure NFC data, M2M data, Sensor data 1
2 4 4
CRMComplaints, Profiling details, Location data,
Requests, Client emails 3
2 4 5
Web Infrastructure IP adresses, Transactions, Basket Analysis 2
2 3 4
Other Internal TV, Media, Fixed lines, Financial Dat (Hyperion) 2
3 3 3
Social networks (FB, VK, Twitter
and alike)SNA, Alpha leaders, Hubs, Sentiments and tones,
Engagement, Rich Customer Profiling 0
4 5 5
Mobile Applications Usage, Preferences, Profiling 0
4 5 5
CSP Exchange Data exchange with other operators 0
1 1 4
Financial and Insurance
Institutions Score exchanges, fraudulent customers 1
1 1 4
Retail Cheque, Preferences, Location, CRM 0
1 1 3
Web Crawling Sentiment, Interest Profiling 0
4 4 4
GovernmentTransportation, Weather Forecast, Real Estate,
Urban Statistics 1
2 2 3
Research Companies Behaviour analysis etc. 0
3 3 4
Other Third Party Data Other data 0
3 3 3
WHAT DATA CAN TELCO RELY ON?
Internal Data
The data generated from all internal sources starting from traditional billing and core network and finishing with logs generated from
web sites and various applications
External Data
The data generated from unusual external sources
WHAT TELCO MIGHT NEED THIS DATA FOR?
Cost Optimization
New Revenue Streams
Inte
rnal
mo
net
izat
ion
Exte
rnal
mo
net
izat
ion
Enhancing UX
- Network planning- Supply chain- Channel Performance- Sales performance- Revenue assurance- Churn prevention- Retail optimization- Improving cross/up-sale
- Product and service design- Fraud Prevention (Banking and other)- Marketing- Customer complaint prevention
- Smart city services- Retail planning- Digital advertising- Insurance and finance scoring- Marketing research- Utilities- Healthcare- Data Brokerage- Data hub- Recommendation engines support- Converged B2B services
64%
22%
14%
Share of Opportunity in 2019
InternalMonetization
Big Data as aService
Big Data DrivenBiz Models
Euro 359mn 2015
2019
15-19
Euro 1,526mn
Euro 4,380mn
Detecon estimations for Europe
- Less than 40% of Big Data initiatives expected to result in new revenue streams- The most promising revenue-generating initiatives are in City Planning,
Healthcare and Advertising
Based on Gartner evaluations
Future Cash Cows
• It is going to be a long way for telcos to reach maturity in big data processing and value extraction
• Internet peers however are already at the top level of maturity which may result in fierce competition and dramatic devaluation of data telcos currently dispose
Big Data metamorphosis
Small Data Paradigm
• Reformatted project and process management• Full-scale recommendation and prediction engines• Fully anonimyzed, inventoried and protected data• Large number of products including internal fraud
and risk prevention. New digital revenue streams• Large number of partners from Internet community
• Formulated Big Data strategies and implementation
• Advanced Cross-sell/Upsell• Mature churn prediction and
prevention process• Large range of white labeled digital
and IoT services
• Small Data Paradigm• Slow decision making process• Lack of digitalization of the
business• Small penetration of convergent
products
• New revenue streams reaching 10-30% of total revenues
• Smart and Soft Pipe• Significant M&A activity in the
Internet domain• NewGen services and products
2015 2016 2017 2018 2019
BIG DATA TRANSFORMATION STAGES AND OUTCOMES
Vodafone
Telefonica
Telstra
SKT
SingTel
Orange
DT
DOCOMO
AT&T
Telcos Big Data Activities
Most of the telcos
are here
Gartner There are a lot of activities in Big Data domain however revenue implication of these activities is still low or unreported
• Identification of clear priorities and a development
plan for internal products
• Participation in the formation of a product market
with external monetization
• Development of mechanisms for the purchase and
use of external data
Demand and USP creation
Building new paradigm Infrastructures
Competencies
Processes, project management
• Creation of holistic Big Data IT infrastructure
• Implementation of agile development principles for Big Data infrastructure, on-demand development processes
• Spin-off Big Data Initiatives
• Agile project management
Legal Risk Management
Nurture and cultivate new competencies:• Data Science • Data Governance • Product Management• System architecture• DevOps
• Creation of a unified system for managing the risks associated with Big Data products
BIG DATA TRANSFORMATION PILLARS AND PREREQUISITES
*Source: Gartner, Key Trends in Analytics, Big Data and Data Science, 2014
• The most important issue with big data is whether it can add
significant value to the business of the telco beside its cost-
optimizing effects
• Legal risks are not perceived as the most crucial ones though
might be a showstopper for almost all profitable external
services and products
Purpose:
Increase in awareness regarding benefits of “Big Data” related approaches throughout the top management
Identification of most relevant use cases / pilot project set-up
Alignment with IT Roadmap
Activities
Big Data use case identification
Big Data use case description, covering data requirements and expected benefits
Valuation of use cases (high level) and short list derivation
Elaboration of Business Case for shortlisted use cases
Prioritization of use cases
Set-up of big data technology roadmap for BI
Big Data Use Case Evaluation
Purpose:
Validation of the business value of a selected use case
Activities
Identification of business requirements
Vendor assessment for technical solution components
Proof of concept and trial setup of the preferred technical solution
Execution of pilot project and performance monitoring
Preparation of Go/No Go decision based on detailed analysis of pilot results
Big Data Pilot Project
Purpose
Development of overall Big Data strategy and implementation plan to fully leverage the benefits of Big Data
Activities
Develop the vision, targets, target segments, technology architecture and roadmap
Optional: Design of a Telco Center of Excellence for Big Data
Optional: Design a Business Unit “Big Data as a Service”
Adaptation of organization and relevant processes to Big Data logic
Run RfP & Vendor selection for Big Data technical solution
Plan technical integration
Launch & operational support
Full Implementation & Launch
STEPS TO BUILD BIG DATA CAPABILITY IN TELCO
Small steps. Pilot projects. Proof of concept and proof of value. Formulation of the strategy
Board approval of the strategy and CAPEX. Execution stage
A ZOO OF SOLUTIONS TO CONSTRUCT THE ARCHITECTURE
Knowledge Build Up
What is the best way to control the new technology ?
Build your own knowledge base and experience
Hire external knowledge/consultants
Analytical Processing
How can analytical tools handle the necessary amount of big data ?
Look for solution on the market
Write your own code
Access to Source Data
How can you get access to data on mission critical systems (e.g. HLR) ?
Manage risk
Convince system owners
Big Data Platform Selection
Go with traditional BI tools or with new open source driven platform ?
Reliability and maturity of solutions
Cost of solutions
New capabilities
• Big Data is a multitude of critical blocks that together transform into a high-performance monolith.
• Each such block is provided by a host of companies and solution options you can pick from
• Big Data Technology
Source: BITKOM, Big Data technologies - Knowledge for decision makers, 2014
Data Management and Storage
Data Integration
Visualization
Analytical Processing
Data Manipulation
Data Connectivity
Data Ingestion
DashboardsAdvanced
VisualizationReal-time
Intelligence
Video / Audio
Geospatial WebText
Semantics
PredictiveData
MiningMachineLearning
Reporting
BatchProcessing
Streaming& CEP
Search &Discovery
Query
HadoopDistributedFile System
NoSQLDatabases
In-MemoryDatabases
AnalyticsDatabases(DW, etc.)
Transactional Databases
(OLTP)
Data analysis - be it big or small - needs a set of functional modules to do its work.
FUNCTIONAL COMPONENTS OF BIG DATA ARCHITECTURE
Data Governance & Security
Identity & Access Management
Data Encryption
Multi-client Ability
Governance
SAMPLE MODULAR DESIGN OF BIG DATA ARCHITECTURE AND CAPEX ISSUES
Large portion of CAPEX might be consumed to get data prepared for further ingestion and processing, e.g. preciseness of geospatial data, web traces inspection, structuring internal unstructured data.
Security aspects of Big Data require huge amount of CAPEX
A thorough PEST analysis should precede any Big Data development with an external component
Is data privacy a high profile political concern?
What is the regulatory framework?
Which data types are protected?
How long can I store data?
Who can I share it with?
How do I need to protect it?
How will the general public react to our business model?
What has to be expected in terms of press coverage?
How will privacy interest groups react?
Who are my competitors in the Telco industry / in other industries?
Which advantages / disadvantages may their business model have over mine?
Which competitors have the highest potential to create synergies through partnering?
Whom do I need to partner with to gain a competitive advantage?
How can I connect to these partners? (APIs)
Political Economic
Social Technological
BIG DATA PEST MATRIX
44%*
30%
70%
33%
Never or very rarely share their personal data
Would be unhappy if their personal data were shared withthird companies
Would agree to a company using their personal data for more relevant marketing
Would agree to a company using their personal data for the development ofnew products and services
*Source:Ernst&Young, Big Data Backlash, 2013
Since telcos operate under strict regulatory rules any unauthorized personal data usage might be prohibited. Moreover, the negative publicity around such Big Data products may heavily overweigh its positive outcomes
Consumers in many countries are not ready to embrace processing of their personal data
WHAT ARE THE MAIN RISKS TELCOS MIGHT FACE? SAMPLE BIG DATA RISK MAP
Legal risks/Data privacy
Fierce Competition with OTT
Lack of scale/Lack of demand
Competition with other telcos
Fail to deliver products and services
Small range of external products
Fail to construct an appropriate architecture
Fail to gain public support
Fail to gain support from the regulation
Devaluation of the data and poor profits
Poor execution
Lack of scalability devaluating future revenues
Early price erosion of data
There at least 4 highly probable showstoppers according to the Big Data Risk Map
- Competition with other telcosSince all the telcos have quite similar sample of data it will be difficult to differentiate services and products that will result in price damping
- Fierce Competition with OTTOTT players can compete in many ways with telcos. Combined they have comparably huge amount of data at considerably lower prices (LocateIt)
- Legal risks/Data privacyThreat of data leakage can have a very dangerous outcomes since telcos operate under regulatory set of rules. Moreover, many telcos are not allowed to process data except for purposes formulated by law
- Fail to gain public supportThe right and beforehand publicity of the services with positive externalies is absolutely the must. In Britain, geospatial service Smartsteps from O2 was boycotted while the T-Mobile’s MotionLogic gained support and is still active thanks to its right prepositioning and PR