3. According to Gartner 85% of
Fortune 500’s are not doing it.
According to Accenture, of those
who are doing it, 75% are failing.
Few can describe it and even
fewer know how to do it.
What is Big Data?
4. 1. Big Data Collection
(HDFS)
2. Big Data Processing
(Hadoop)
1. Data Mining at Scale
(Hive)
Breaking down the IT of Big Data
5. Big Data Tools
Words you May Hear
BlinkDB
CassandraHive
Python
Pig
Stinger
HadoopGiraph
Spark
GraphX
MLbase
You don’t need to be an expert in these tools, but knowing
how they are used goes a long way
Impala
6.
7. Image
Unstructured
Semi
Structured
Structured
• Click Streams
• Social Streams
• RSS feeds
• XML Documents
• Spreadsheets
• Relational
Databases
Data ecosystem, what is it, how to understand it.
Unstructured data is the goldmine,
it is growing while structured data
is shrinking. But to make big data
work for you, you need to structure
of the unstructured
9. How we Personalize Big Data and Marketing in Use
Combine the strengths of Google and Facebook’s methods with psychograph
techniques.
Listen, Adapt, Respond
Services co-created with customers and are interpedently with wider service
networks.
Benefits
People will log in more
Higher conversion and AOV
Better emotional bond between company and customer
Psychograph
Self
Facebook Self
Google Self
Clash between Today and
Future
Aspirational You
Present You
1-1
10. Sentiment
Expressed as
positive, neutral, or
negative, the
prevailing attitude
towards and entity
Behavior
These signals
identify persistent
trends or patterns in
behavior over time
Event/Alert
A discrete signal
generated when
certain threshold
conditions are met
Clusters
Signals based on an
entity’s cohort
characteristics
Correlation
Measures the
correlation of
entities against their
prescribed attributes
over time
Rate of Change
(Slow or Fast)
Quality
(Predictive or Descriptive)
Sensitivity
(Sensitive or Insensitive)
Frequency
(High or Low)
All signal types have certain qualities that describe how quickly signals can be generated
(frequency), how often the signals vary (rate of change), whether they are forward
looking (quality), and how responsive they are to stimulus (sensitivity)
Signals have attributes depending on their representation in time or frequency
domain can also be categorized into multiple classes
Signal Types
11. Timing/ Recency
Measure the
freshness of the
data and of the
insight
Source
Measure
sources’
strength:
originality,
importance,
quality, quantity,
influence
Content
Derive the
sentiment and
meaning from
tracking tools to
syntactic and
semantics
analysis
Context
Create symbol
language to
describe
environments in
which the data
resides
Clickstreams
Social
Articles
Blogs
Tweets
For each dimension, develop meta-data,
ontology, statistical measures, and models
High quality signals are necessary to distill the relationship among all the of the Entities across all
records (including their time dimension) involving those Entities to turn Big Data into Small Data
and capture underlying patterns to create useful inputs to be processed by a machine learning
algorithm.
Finding Signals in Unstructured Data
12. Behavioral
Patterns
1 to 1
Marketing
Product/Service
Compatibility
Market Trends Social
How the Data Becomes Customer Experiences
Crowd based user
actions drive
recommendations
Personalized
email
marketing
Recommendations
based on products
Use machine
learning
algorithms to
predict trends
Small world
network
communication
Algorithms analyze data
Data Capture Points, Experience Delivery Points, Metrics
Data Capture Ecosystem
13. The Data, Insights, Action Gap
The Data Insights Gap
Data to insights can often fall short
for a number of issues
- Difficulties in defining areas of
focus for external data
- Only gradual adoption of
exception analytics and
automated opportunity seeking
- Example (P&G / Verix Systems)
- Opportunity seeking business
alerts
- Value share alerts
- Out of stock alerts
- New Launch alerts
The Insights Action Gap
Processes and systems designed
prior to big data thinking
Examples:
- CRM
- Pricing: Buy now in-store pricing
- Supply chain and logistics
- Prevalence of operational ,
internal metrics
- Complex new concepts:
“Intents”
17. Image
Data Discoverers
Data Discoverers are setting the trend in what will be
common place in just a few short years.
More people will want to use their data and the
consumerization of data and technology will continue.
As this trend goes, only organization that learn to merge
the various disciplines of strategy, analytics and IT, will be
successful
Data as a Lifestyle
19. Search On-sites Sensors Re-marketing
Customer
Feedback
Signals Hub
Social
Personalization Products
Customer
Service
Digital
Marketing
In-store
Creating Customer Signal Hubs
20. Where we are Going
How we organize our data is getting more customized and
real-time for real bottom line improvements
0%
5%
10%
15%
20%
25%
Vendors Hadoop Customized Customized
Realtime
Big Data Technology Evolution
Personalization Technology
Evolution
25. Vision &
Goals
Governance
Execution
Clearly articulated vision for marketing and data
use, precisely defined goals with how to measure.
Defined scope of the product.
Market strategy, customer segmentation,
prioritization, org focus, measurement and
incentive systems
Production process, flexibility at scale, efficiency,
relationship management, benchmarking, metrics,
initiatives
How work gets structured
26. Strategy
- Define
the goals
Social
Define how to
engage
IT
Assemble the
Technology
Analytics
Make sense
of the Data
Linguistics
Distributed Processing (Hadoop)
Algorithms Development
Cross team Customer Experience
Improvement
Data science is a discipline for making sense of unstructured as well as
numerous data sets at scale
Develop Your Team
27. Listen
•Listen to the data streams
Share
•Share the data with the rest of the organization
Engage
•Engage to the data to find the insights
Innovate
•Innovate new ideas from the insights gained from the data
Perform
•Perform insightful actions from the data to create better customer experiences
Always Remember: Data,
Insights, Actions
28. Print
Radio SEO and PPC
Social
Predictive
Marketing
Television
You Are
Here
Human History of Marketing
Image credit:
www.conducthq.com
Using Data for Marketing in the Future
Predictive Marketing
29. • Extreme machine learning
• Collaborative predictive analytics
• Scale-invariant intelligence
• Neural networks for machine perception
• Real-time interactive big data
visualization
• Graph all the things
• Large scale machine learning cookbooks
• Collecting massive data via crowd-
sourcing
“Without big data analytics, companies are blind and deaf, wandering out
onto the web like deer onto a freeway.”
Big Data: 2014
30. • Personalization everywhere
• Company and consumer collaboration in service design
• Predictive location based selling
• Digital Concierges
• Real time event networks
• Graph and signal hubs merge for better understanding of
ad placement
• Large scale channel disruptions
• Marketing becomes more analytical
Big Data visionaries pose existential threats
Predictive Marketing: 2016
31. What’s Next: Combining contextual and analytical approaches provide a more complete picture of how customers interact
with the firm
Both approaches privilege observation and
understanding what people actually do
and look for opportunities to fix, improve
and innovate.
Robin Beers, founder of Business is Human
32. Location
Analysis
Graph Analysis
App and Device
Analysis
Customer
Feedback
Personal Event
Networks
Social
Personalization Digital
Concierge
Real-time
Service
Better Ad
Performance
True Omni
Signal hubs will
become new centers
for data, helping to
create better
customer insights
Predictive
Analytics
Creating Customer Signal Hubs of the Future
33. Although IT can build the systems, it will still be left to analyst and marketers of all types
to create the actions needed to engage customers
How Predictive Marketing is Shaping Up
34. Web
PDS
Email
ECC
Personal Event Network
Appt
Scheduler
Add to
Calendar
Confirmation
Email
Add
Confirmation
and Appt to
PDS
Using the digital concierge system, we can
create easy to use appointment systems,
capturing the data and using it for future
personalization efforts
Appointment setting with a Digital Concierge
35. Image
Engaging millions at a time
Data Monetization
- Keep it
- Sell it
- Partner with it
- Share it
Marketing of a Mass
Personalized Scale
36. Processes are lined, linear chains of cause and
effect.
A service is different. Processes are designed to be
consistent, personalization services are not
consistent but individualized and co-created. The
differences are not superficial but fundamental.
Co-created value requires a relationship
Marketing of the Future: Process vs Service
37. Marketing as a service relies on the ability of an
organization to learn from customer’s responses
and to listen and adapt to those signals.
Causes of success are never revenue, costs, profits,
etc.., those are lagging indicators or effects.
What matters are the activities that generate the
profits, activities that create long or short term
value. You can measure that via personalization as
it is a leading indicator activity if done correctly.
Marketing is about Listening and Learning
38. An organization’s data is found in its computer systems, but a
company’s intelligence is found its biological and social systems
--- Valdis Krebs, researcher
Linking things changes things: social networks are good at
habit building. As behaviors are repeated, they form stronger
associations over time. You form strong bongs with people in
your life with whom you spend the most time, the same can be
said in a social interactive personalization model, customers
will form strong bonds with organizations they interact with the
most over a given period of time.
Small world networks: people banding together to achieve a
wide variety of shared objectives. These are the most powerful
types of social networks and the way to truly engage customers
is to beyond just social network sites and to get into the small
world networks as a valuable member of the network.
Marketing and Social
39. Start small, and remember, everyone else
is in the same boat
Online Resources
What You can do now
Expanding a marketers knowledge of big data, what you need to know.
Data ecosystem, what is it, how to understand it.
Most of what is needed to make marketing better is still to be explored.
Understand how to see your customer online
Signals help to make sense of the various types of data so you can use them in new ways.
How to understand signals in unstructured is not the same as structured data.
Take data, understand it, process it, extract value, visualize, communicate, measure
The Action Gap is still a big problem for many companies, understanding the cap will reduce the learning curve
A shift from systems and forecasts to activities as our center of design needs to take place.
Data alone can’t predict an unpredictable social animal known as the human being
Focus on the activities of people, not so much predicting them because we can’t do that good of a job with what we have.
The Human Motion Graph is emerging as a new way of understanding customers movements and how they relate to your product or service.
Fitbit as an example of data discoverers
Data as the new self discovery tool
Leads to consumerization of IT IT needs to adapt to be social
This means teaming up with marketing and letting marketing joining the conversation with data
How is big data used? How is it helping?
Signal Hubs are the Future of understanding customers
Realtime and customization are the future, faster response times.
How to set up your data practice.
How is big data used? How is it helping?
Different stages of maturing a company goes through.
Set up a structure
Understand the vision, make it clear
Governance: have structure around the tasks
Execution: Know how to get it done and why
Leadership plays a very important role in defining the vision and goals along with how they wish to see the program governed. Most people don’t understand personalization so having the right structure in place helps to ensure a good foundation for growth.
How is big data used? How is it helping?
How is big data used? How is it helping?
How is big data used? How is it helping?
How is big data used? How is it helping?
How is big data used? How is it helping?
How is big data used? How is it helping?
How is big data used? How is it helping?
The product is an intermediate step, not an end in itself, even after the customer buys, there is still a relationship after the sale that can take place beyond the product. With a process, this isn’t the case, once the final step is complete, you are done.
A process has one customer, the person who receives the final results, a service at its core a relationship between the served and server. At every point of interaction the measure of success is not a product but the satisfaction, delight, or disappointment of the customer.
“Most corporate systems were not built with customer delight in mind.” Fred Reighheld, Fellow, Bain and Company
Learning organizations evolve with the customer and personalization helps you understand how to evolve.
Ritz Carlton, the staff is trained to listen for guest preferences, not always stated in the form of a direct request. The staff is trained to look for intent and then act upon it. This is why the Ritz Carlton’s service is legendary, they have learned how to perfect personalization in the physical space and is a model to follow for Best Buy and can be done with our own personalization efforts.
Continuous improvement is natural!
Anatomy of a social network:
Brokerage: A person or group that connects different clusters together.
Closure: Building trust within a cluster, the closer you are the stronger the trust.
Betweeness: Critical linking member between other nodes in the cluster.
Closeness: How easily a person can make connections
Degree: Number of connections
Developing a social aspect of personalization requires a high degree of network fluency, situational awareness, influence, compatibility and a fair amount of luck.