The country manager of Teradata India spoke at the QlikView Business Discovery World Tour in India on the growing importance of Big Data and the need for user-driven BI. View the presentation and how QlikView Business Discovery & Teradata are looking at Big Data & Business Discovery.
2. Agenda
• Big Data Current State and Challenges
• Big Data Journey - How to start?
• Big Data Scope – An Indian Perspective
• Big Data Business Cases
– BFSI
– Telecom
• Big Data Analytics – What should be the underlying architecture?
– 3 Pillars of Big Data Analytics
• What you need to do next…
3. Big Data – Current State and Challenges
Confusion on business
cases which are valid in an
Indian scenario
Clarity on types of BIG Data
Lack of clarity on what is
the right data architecture
to follow for the best ROI
Clarity on types of possible
BIG Data technologies
Lack of clarity on
implementation options
and what is the right
model for you
4. Big Data Journey…How to start?
The four commandments…
1
Become the first mover! If
not, you will be already
late…
2
Start small and identify
convincing business
4 cases
Learn to let go of
stuff that does not
work
3
Once value is
established…
Expand !!
5. Big Data Scope
An Indian Perspective
• Telecom
• Banking BIG Data scope
1. Network performance
• Auto monitoring,
• Insurance BIG Data scope
2. Fraud detection
3. Customer churn detection
• Retail 1.
2.
Customer 360
Anti-fraud and risk
4. Credit risk analysis
5. Micro- segmentation
Analytics Maturity
3. Cards analytics 6. Social media analytics
4. Telematics
BIG Data scope 5. Sentiment analytics
1. Web analytics
2. Recommendations Traditional analytics
3. Life cycle marketing Traditional analytics 1. Operational KPI’s
4. Sentiment analytics 1. Operational KPI’s 2. Handset analytics
2. Warranty analytics 3. Next best offers
3. Cross-sell/ up-sell 4. Usage analytics
Traditional analytics 4. Regulatory reporting 5. Campaign management
1. Operational KPI’s 5. Forecasting
2. Trend analysis 6. Event based marketing
3. Claims analytics
4. Supply chain analytics
5. Regulatory reporting
Low Medium High
Sector Maturity
7. BFSI: Business Use Cases
Customer 360
Customer’s transaction and interaction footprint Actions
• FB: “My money transfer’s stuck for 3 days”, Finally my
own house!”, “ I am getting married, yippee”, “Likes” Ritu • Offer a packaged account with
Kumar, “Moving to Mumbai”
• Linked in: Job change to VP @ Mumbai
home loan and education loan @
• Twitter: “ Made my car down payment”, following HBS, relationship pricing
Social Stanford, “ Hate service of XYZ bank” • Convert salary account to high
Media
save account
• She complains that she has been provided an add-on card
• Offer timely apology for issues/
without her applying for it grievances
Call center • She complains about a fund transfer 3 times across • Offer car insurance at best rates
last few days • Reverse any charges
• Offer marriage planning services-
• Her internet banking usage falls rapidly
• Her regular salary credits stop happening offer from Ritu Kumar for marriage
• She makes some high value transfers ensemble
Internet • She makes a payment to enroll for GMAT • Offer gold @ special prices
Banking • WEBLOGS • Offer vouchers @ xyz mall
Customer
• Offer specialized help for home
• She adds her dad as a payee loan transfer to your bank
• She makes large ticket payments every month to a real • Offer concierge services for travel
Mobile
Banking
estate firm bookings
• Help her in finding apartment in
Mumbai
• Her card transactions are dropping fast
• Help her get utilities in Mumbai
• Her purchase behavior suddenly changes
ATM/POS
• She requests for FD liquidation
• She checks if she can move her home loan to your bank
Branch
8. BFSI: Business Use Cases
Sample output: Common paths to account closure
Most Favored Paths
Events
Outcome
Multiple Fee Reversal and Viewing Product/Rates and Offers happens in the last mile for Account Closure. A list of
such customers who are in the midst of displaying this pattern is prepared.
This would be a business insight output from the Big Data analysis.
9. BFSI: Business Use Cases
Location, Behavior, Navigation Based Fraud Protection
• Is there account activity happening across multiple channel locations in
a specific time-span warranting further review?
• Distance calculation between subsequent access locations and
flagging off transactions which are physically impossible. E.g. first
access from Hyderabad and second access from Russia in 5
minutes
• Is there a sudden change in the customer’s buying behavior?
• Customers buying pattern follows a ratio:
• 30% Rental, 10% entertainment, 20% shopping and
suddenly it changes to 80% shopping
• Customer usually transacts in a certain set of shops, outlets
and it suddenly changes to a different pattern
• Is there an identifiable pattern of navigation that a customer follows
and is a certain access very different from that pattern?
• Customer logs in and clicks the account summary details link and
then clicks funds transfer and or FD opening, etc. Suddenly the
pattern changes to change mobile phone number, initiate
transfer to new beneficiary, password change, etc.
10. BFSI: Business Use Cases
Sample output: Graph Analysis for Fraud Detection & Prevention
• Identify complex networks of
relationships
- Users to transactions
- Transactions to means of payment
- Identities to individuals
• Fraud identified by graph relationships
- Clusters of connections
- Patterns of activity between
connections
• Understand impact of fraud
- Trace flow of money and goods
- Identify users impacted by fraud
11. Telecom: Business Use Cases
Social network analysis -Churn prevention through pattern recognition
• Core Business Challenge
– Negative experience with the core
network product can spread through
a subscriber’s network
– Identifying the negative experiences
and providing remedies is essential
– Proactive and/or knowledgeable
responses to at-risk customers can
• Business Questions Answered
– Which key subs are likely to churn?
– Who within the subs network is also
at risk?
12. Telecom: Business Use Cases
Network System Optimization – Prediction of network quality
• Core Business Challenge
– Prediction of the next failure point is elusive
– Network traffic patterns materially impact
customer experience by impacting
accessibility and retain ability
• Business Questions Answered
– What is the ‘normal’ condition of the network
by geography and time?
– What deviations from ‘normal’ constitute a
threat to service?
• Insights Gained
– Analysis of massive data sets creates new
benchmark or ‘fingerprint’ of network
performance
– Deviation patterns trigger immediate
response or preventative actions
– Previously unknown vulnerabilities informing
network investment
14. What is needed for effective discovery?
UDA Unified Data Architecture : Addressing variety of big data sources
• UDA is the only truly integrated analytics solution unifying multiple technologies that work behind the scenes
in the travel value chain
• UDA is a cohesive and transparent architecture that leverages best-of-breed and complementary values of
data warehousing and data discovery and open source Hadoop
• UDA represents a ‘room-to-grow’ approach
Discovery Analytics Innovation Analytics
Cross Channel Purchases Event Triggers
Golden Path analysis Marketing Integration
Attrition Paths DISCOVERY INTEGRATED Customer Behavior Analytics
Fraudulent Paths PLATFORM DATA WAREHOUSE Customer Segmentation
Attribution analytics Customer profitability
Customer Sentiment Analysis Customer valuation
Unique customer ID
Sessionization
Cross Platform Aggregation
BUT Also many new data types
CAPTURE | STORE | REFINE
EMAIL CUSTOMER FEEDBACK & USER SOCIAL COMPETITIVE CHANNEL FLT & MTCE AIRPORT
CAMPAIGNS DATA GEN CONTENT MEDIA DATA DATA DATA OPERATIONS
15. The 3 Pillars of Effective Big Data Analytics
Data Scientists Business Analysts Marketing Front-Line Workers
Engineers Customers / Partners Executives Operational Systems
Pillar 1:
The storage
LANGUAGES MATH & STATS DATA MINING BUSINESS INTELLIGENCE APPLICATIONS
DISCOVERY INTEGRATED
PLATFORM DATA WAREHOUSE
Step 1: Data Capture
• Store high volume Big Data from detail
web interactions, mobile applications,
search engines, industry competitive
sources and social media in Hadoop
“Cold” data store
• Cleanse data and preprocess for CAPTURE | STORE | REFINE
subsequent analysis
EMAIL CUSTOMER FEEDBACK & USER SOCIAL COMPETITIVE CHANNEL FLT & MTCE AIRPORT
CAMPAIGNS DATA GEN CONTENT MEDIA DATA DATA DATA OPERATIONS
16. The 3 Pillars of Effective Big Data Analytics
Data Scientists Business Analysts Marketing Front-Line Workers
Pillar 2: Engineers Customers / Partners Executives Operational Systems
The Discovery
Platform
LANGUAGES MATH & STATS DATA MINING BUSINESS INTELLIGENCE APPLICATIONS
Step 2: Discovery
• Use Connectors for Hadoop to
extract individual customer data
from Hadoop into discovery
platform
• Use SQL-H to query less
frequently used data in Hadoop DISCOVERY INTEGRATED
• Use nPath to explore patterns to PLATFORM DATA WAREHOUSE
determine consumer most
frequent paths to purchase
• Use Graph Analysis to identify
strength of connection in social
network
• Analytics Portfolio contains 50
commonly used routines
CAPTURE | STORE | REFINE
EMAIL CUSTOMER FEEDBACK & USER SOCIAL COMPETITIVE CHANNEL FLT & MTCE AIRPORT
CAMPAIGNS DATA GEN CONTENT MEDIA DATA DATA DATA OPERATIONS
17. The 3 Pillars of Effective Big Data Analytics
Data Scientists Business Analysts Marketing Front-Line Workers
Pillar 3: Engineers Customers / Partners Executives Operational Systems
The
Warehouse Step 3: Innovation
LANGUAGES MATH & STATS DATA MINING BUSINESS INTELLIGENCE APPLICATIONS
• Move discovery results into
using connectors
• Use combination of SQL & SQL
MR analytics to create new
customer data analytic sets
• Use increasingly rich 360o view
of customer to drive continuous
DISCOVERY INTEGRATED product innovation
PLATFORM DATA WAREHOUSE Create new marketing and
social interactions using
Event Based Management
triggers
Initiate real-time and right
time messaging, based on
travel cycle or life cycle
Integrate expanding user
CAPTURE | STORE | REFINE generated feedback, content,
gamified participation, etc
EMAIL CUSTOMER FEEDBACK & USER SOCIAL COMPETITIVE CHANNEL FLT & MTCE AIRPORT
CAMPAIGNS DATA GEN CONTENT MEDIA DATA DATA DATA OPERATIONS