More Related Content Similar to Digital bankleadgen (20) Digital bankleadgen1. Social Media Lead Generation
For Digital Bank
Sudarson Roy Pratihar
Data Science Evangelist and Technology Leader
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
2. Understanding Digital Bank
n Young's turned to 18 n Foreigners Relocating
n Pillars of Digital Banking
n Customer centric value
n Banking anywhere
n Disrupting and catalysing change
n Customers are bankers
n Go social
n With no compromise on
n Security
n Privacy
n Integrity
High Customer Life Time Value
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
3. Challenges
Computational and Data Collection
n Problem
n Diverse sources, huge data set and natural language
n Infrastructure set up cost and time
n Solution
n Use cloud infrastructure on-demand to start with
n Spot instances and EMR option of Amazon Web Services
(AWS) could be one potential way for cheaper bulk
processing of non-time critical data
Operational and Strategic
n Problem
n Understand customer, customers’ perception on needs and
preference
n But finance is not so frequently discussed on social media,
except negative experiences
n Identify and attract customer who can be brand
ambassadors
n Convert leads to customers
n Identify social medias to focus on
n Solution
n Promote discussions on social media on selected topics and
empower target segments to voice preferences in Bank’s
website through dedicated social page (and accessible
over mobile devices as well)
n Analyze social media data to understand needs of targeted
segments, identify social media groups to discuss
n Analyze social network to identify customer who can
influence
n Start with Social Medias such as FaceBook, Linkedin,
Consumer Association of Singapore for major data, but fill
in the gaps using Pintrest, Google+, Slideshare.
n Identify attributes to target influential customers and use
social media ads
n Redesign Bank’s landing pages and mobile app
Statistical and Analytical
n Problem
n Measure Social Media effectiveness
n Measure Campaign effectiveness
n Measure Bank Website and Mobile app effectiveness
n Solution
n Design experiments with randomly chosen samples of
target segments
n Draw statistical inference and significance from experiment
results
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
4. Process at a Glance
• Advice Marketers
• Run Campaigns
• Measure lead
conversions
• Statistical inference
• Campaign Strategy
• Product Strategy
• Marketing Incentive
Strategy
• Perception and
Preference Model
• Social network
analysis
• Field Experiments
and Identification of
main factors and
interactions
Modeling Strategy
Execution
Performance
Evaluation
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
ITERATIVE CYCLE
5. Modeling
Social Network Analysis
n Data Collection
n FaceBook ,Twitter and Linkedin Followers
n Consumer Feedback Sites to find most influential posts
n Technique
n Social Media Network is analyzed to understand network
spread, geography of spread, people who influence and
people who are connected with most influential people in
target segment
n Facebook and twitter data is analyzed to understand need
of customers. E.g. who has recently moved to singapore and
looking for local information
Perception and Preference Model
n Data Collection
n Secondary Data: Customer feedback from social
media about other bank’s offerings; Slideshare,
Consumer Association of Singapore
n Primary Data: Promoted discussion on Bank’s social
page and bank initiated discussions on social media
n Initially Digital Bank will have to rely more on
Secondary data.
n Technique
n Sentiment analysis of tweets, feedback to understand
customer mind and value of features in customer
mind
n Factor Analysis to group together multiple features to
get a dimension where Bank has opportunity to
attract customer
Statistical Analysis
n Data Collection
n Field Experiments in terms of conducting campaign on a
selected social media on small sample of target population
along with rest of target population working as control
group
n Technique
n Regression Analysis to understand impact of the particular
campaign on the given social media
n Campaign effectiveness and effectiveness of social media
can be judged
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
6. Strategy
Digital Strategy
n Modify Digital Bank’s website’s landing
page and Mobile Site based on
Modeling exercise
n A few iterations of field experiments
may be needed to test effectiveness of
landing page and mobile site
Campaign Strategy
n Selection of Social Medias based on
Modeling exercise
n Selection of attributes of target customers
based on modeling exercise
n Use these attributes to select leads and
follow up through marketers using various
channels
n Selection Campaigns and Ads based on
Modeling exercise (Field experiment)
n Control group: A small segment of target
group would be still not exposed to direct
campaigns, so that performance can be
compared
n Factor Analysis to group together
multiple features to get a dimension
where Bank has opportunity to attract
customer
Product Alignment / Incentive Strategy
n Observations on Customers’ preference
and perception to be passed to product
management if there is a significant
deviation
n Based on priority decided on banking
products, incentive for Marketers may
be changed to prioritize sell of certain
products
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
7. Execution
Advice to Marketers
n Based on strategy formulated, advice
marketers to participate and watch
for selected social media discussion
n Watch list of keywords and hash tags
n Use software (developed on data
mining algorithms) to generate a list
of target individuals and contacts
along with ranking based on
influence
Run Campaign
n Contract with social media for
campaign – provide keywords and
other attributes derived in strategy
phase
n Run campaigns
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
8. Performance Evaluation
Performance Measurement
n Gather metrics from social media –
such as Reach, Frequency, Cost per
Thousand, tracking of leads
n Keyword performance
n Conversion of leads to customer and
lifetime value
n Conversion of referral through the
lead who became customer
Statistical Analysis
n Regression and ANOVA analysis of
effectiveness of campaigns against
the control group
n Analysis to determine factors of lead
conversion and effective growth of
customer base through referral
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar
9. n AWS EMR spot instances for low cost batch processing
n AWS Ec2 instances for analytics system hosting
n For real time analytics dash board, spark will be used
Alert
Generation Feedback
capturing
Event
Ingestion
HDFS
Sync
Data
sync
CDR
Data
CDR
Data
Query
Flume/Kafka
HDFS
Hadoop
Cluster
Exploratory
Data
Processing
Python
Spark
Continuous Process
Pyspark/MLLIB
Analytical
adjustment
and
pattern
detection
Reporting
alert/mail/feedback
MYSQL
Social Media
Data
AWS – EMR
Landing zone
Customer
acquisition data
from Bank’s
system
Copyright © 2015 All rights reserved by Sudarson Roy Pratihar