Unlock Your Customer Data
with Data Visualization and
Customer Segmentation
Dr. Richard Hackathorn
Bolder Technology, Inc.
May 14, 2009
Sponsor
Speakers
Richard Hackathorn Andrew Cardno
President and Founder, Chief Technology Officer,
Bolder Technology, Inc. BIS2
Unlock Your Customer Data
With Data Visualization and
Customer Segmentation
Richard Hackathorn
[email protected]R.D. Hackathorn
A Tough Business Problem
• Knowing your customer
Understanding the behavior of your customer
Aligning your products/services to this behavior
• Each customer is different
...whether a person (B2C) or a company (B2B)
...changing with the seasons and phases of the moon
...depending on many unobvious factors
• Customers are not rational but ...
they are predictable
R.D. Hackathorn Slide 5
Predictably Irrational by Dan Ariely
• Do people make rational decisions?
Deep question! What is rationality?
• Our brain tricks us constantly
Recent events weight more than past events
A random event that is positive will lead to bad habits
• Customer behavior is a summation of these
irrational decisions
R.D. Hackathorn Slide 6
A Tough Business Problem
• How can you ‘assist’ your customer to...
Recognize your product/service, brand, logo?
Match your product/service with their needs/wants?
Decide whether your product/service is worth it?
Make a purchase of your product/service?
• Would it be of value if you knew about:
Growth opportunities in customer base
Effectiveness of marketing campaigns
Frequency of visits to revenue and profits
Customers who were ‗overdue‘ for a visit
...and so on
R.D. Hackathorn Slide 7
An Example – Telecommunications
• Facing cost reductions, evolving technology
alternatives and shifting market niches
• Customer retention
• Service up-selling
R.D. Hackathorn Slide 8
A Tough Business Problem
• If you knew your customer, you could:
Develop customized marketing programs
Highlight specific product features
Establish various service options
Design an optimal distribution strategy
Determine appropriate product pricing
Prioritize new product development efforts
Design of new product strategies (packaging, pricing)
• Impacts many functions and levels
across the entire enterprise
R.D. Hackathorn Slide 9
Customer Segmentation
• Segmenting
Compile demographics, etc about customers
Cluster customers based on similarity of attributes
Treat customers within a cluster as the same
• Targeting
Choosing segment to target with specific marketing
• Positioning
Designing marketing for a specific segment
• Problems with this approach
Incomplete and inconsistent data
Clustering algorithms with complex interpretations
Large number of segments
R.D. Hackathorn Slide 10
It feels like...
You are looking through a keyhole
R.D. Hackathorn Slide 11
It feels like...
your
customers
...into a room with
lots of activity
R.D. Hackathorn Slide 12
It feels like...
...but
you really
want to
see like
this!
R.D. Hackathorn Slide 13
Need for a New Paradigm
• How do you distill meaning out of data?
Lots of data, powerful query/report tooling
• Computational-centric analytics
Analysis comes from processing algorithms
Such as data mining tools, predictive analysis...
• Visual-centric analytics
Analysis comes from visual perception
...But a new generation of visual tools
R.D. Hackathorn Slide 14
Toward Visual-Centric Analytics
visualization
computation
data warehouse
computational-centric
R.D. Hackathorn Slide 15
Toward Visual-Centric Analytics
visualization
computation
computation
data warehouse data warehouse
computational-centric visual-centric
R.D. Hackathorn Slide 16
Toward Visual-Centric Analytics
• Achieving the balance
R.D. Hackathorn Slide 17
Cross-Levels and Cross-Functions
strategic
tactical
operational
computation
data warehouse
cross-levels
R.D. Hackathorn Slide 18
Cross-Levels and Cross-Functions
strategic
marketing
tactical manufacturing
operational sales
computation computation
data warehouse data warehouse
cross-levels cross-functions
R.D. Hackathorn Slide 19
Visualizing Cross-Enterprise
computation
data warehouse
typical
R.D. Hackathorn Slide 20
Visualizing Cross-Enterprise
computation computation
data warehouse data warehouse
typical depth across levels
R.D. Hackathorn Slide 21
Visualizing Cross-Enterprise
computation computation computation
data warehouse data warehouse data warehouse
typical depth across levels breadth across functions
R.D. Hackathorn Slide 22
The Power of Seeing
• Seeing the business value
Closely coupled to the business value chain
Naked data that shows business dimensions
• Seeing the whole picture
Data in context with overview and detail
High-density content in shape, texture, color...
R.D. Hackathorn Slide 23
The Power of Seeing
• Seeing in a glance
Intuitive perception...once trained
Stimulating the ‗a ha‘ moments
• Seeing as a group
Decisions made across functions and across levels
Resulting in coordination across the enterprise
R.D. Hackathorn Slide 24
A Tough Business Problem – Wrap-up
• Knowing your customer
Understanding the behavior of your customer
Aligning your products/services to this behavior
• Seeing your customer behavior
Perform a value-based customer segmentation
...Using visual-centric analytics
...Gaining insights into behavior of specific customers
• Telecommunications...
Finding the good customers who will probably flip
Offering the right service to the right customers
R.D. Hackathorn Slide 25
TM
Unlock Your Customer Data
With Data Visualization and
Customer Segmentation
Andrew Cardno
CTO, BIS2
26
Introduction to Super Graphics
SPATIAL TEMPORAL PIVOTAL
QUARTAL INSPATIAL
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 27
All Rights Reserved.
May 2009
Introduction to Super Graphics
Present the results visually Example Super Graphics TM
using Super Graphics. Easily identify patterns and meaningful
relationships in the data
vizbybis2 is a highly
configurable tool.
Allows you to directly Access the vizbybis2 GUI or use
vizbybis2 as a mashup
query the database.
EDW friendly because it
CRM, 3rd
works on the data in place. XML Party BI
Present Visualizations Tools &
Other
Applications
Millions of customers:
Query Data
across Thousands of branches / locations
doing Millions of transactions
with Billions of interactions Directly access the data e.g.
• Regular queries
buying Thousands of products / services • Spatial queries & functions
supplied by Thousands of vendors
due to Thousands of Promotions
serviced by Thousands of employees
EDW
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 28
All Rights Reserved.
May 2009
Enhancing Your Customer Data
Mashup on context maps (such as GoogleTM Maps which
accelerate deployment and are continuously updated).
Spatially visualize your database!
Demographic
& other data
+
zip codes, roads,
polygons etc.
BIS2 SpatialXchange
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 29
All Rights Reserved.
May 2009
Using Industry Solutions to
Unlock Value
retailViz TM telViz TM
…see and understand: customer bahaviors, …see and understand: customer revenue,
instore management, channels, campaigns, lifetime value, churn, network performance,
SCM… campaign effectiveness…
insuranceViz TM gameViz TM
…see and understand: customer profitability, …see and understand: how to optimize
claims, risk factors, cross-sell / upsell, agent revenue, factors that influence what customers
performance, lapses… play, rate of play, when, how long, where…
entertainmentViz TM manufacturingViz TM
…see and understand: demand, costs,
…see and understand: customer preferences,
operational efficiency, resource management,
distribution management, promotions…
production/supply risks…
moneyViz TM
…see and understand: customer profitability,
financial risk, customer retention, competition,
costs, customer interactions…
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 30
All Rights Reserved.
May 2009
Good Customer Segmentation
is the Key to Unlocking Value
Dimension Reduction
DATA
Clustering
Identify Noise Clusters Cluster Remainder
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 31
All Rights Reserved.
May 2009
Good Customer Segmentation
is the Key to Unlocking Value
Dimension Reduction
DATA
Clustering
Identify Noise Clusters Cluster Remainder
Use Clusters to Build
VISUALIZATION
Super Graphics
Clusters of
Visualize from WHOLE to PART results Deep customer
insights of
Dimensional clusters
• Marketing Results Analysis
Inside customer
Understand All • Responders Cyclical interactions
patterns of based on clusters
• Related to Physical Location clusters
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 32
All Rights Reserved.
May 2009
Good Customer Segmentation
is the Key to Unlocking Value
Dimension Reduction
DATA
Clustering
Identify Noise Clusters Cluster Remainder
Use Clusters to Build
VISUALIZATION
Super Graphics
Clusters of
Visualize from WHOLE to PART results Deep customer
insights of
Dimensional clusters
• Marketing Results Analysis
Inside customer
Understand All • Responders Cyclical interactions
patterns of based on clusters
• Related to Physical Location clusters
INSIGHT
Key Business Percent (%) of whole or other For example, A map showing
Performance variable – where are the percent (%) of responders
Drivers (BPDs) variations from the whole against whole
Understanding through Super Graphics to develop insights
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 33
All Rights Reserved.
May 2009
Spatial – Spatial Xchange (Story 1)
BIS2‘s Spatial
Xchange provides
key geospatial
data for BIS2 and
Partner
customers.
The geospatial
data is ‗load-ready‘
with Super
Graphics from
BIS2.
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 34
All Rights Reserved.
May 2009
Spatial – With Customer
Segments (Story 2)
Analysis of revenue and profit by customer segment, product and ZIP Code - Using
the vizbybis2 Spatial Super Graphic, show any predominance for revenue and profit by
ZIP Code. CLUSTER #1
Example questions include:
• Are there any regional
variations in consumption?
CLUSTER #2 And consumption over
time against any national
trends? This could be
against the norm and/or
revenue value per head of
CLUSTER #3 population.
• Is there any region which
is a leading region? We
could use the take-up or
churn from any leading
segment, what will it look
like in five years?
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 35
All Rights Reserved.
May 2009
Temporal – Call Patterns (Story 3)
Understanding call volume patterns against network utilization - Using the Temporal
Super Graphic, show call volume patterns against network utilization to better optimize
network utilization and reduce costs.
Example questions include:
• Network utilization by all products
and by product line.
• Network utilization by all
customers and by customer
segment.
• Network utilization by all
Network Utilization – geographic areas or particular
2001 - 2008
geographic areas.
• Network utilization - % from mean
– areas of the network.
• Network utilization – product line
or customer segment
increase/decrease year-on-year.
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 36
All Rights Reserved.
May 2009
Pivotal – Call Routing (Story 4)
Understanding call routing patterns - Using the vizbybis2 Pivotal Super Graphic, show
call routing patterns to better understand and optimize interconnect pricing and costs.
Interconnect
Agreement
Time or customer segments or product lines (over the network)
Example questions include:
• Call routing by all interconnect agreements and specific interconnect agreements.
• Call routing by cost by product line.
• Call routing by cost by customer segments.
• Call routing costs - % from mean.
• Call routing costs – product line or customer segment increase/decrease
year-on-year.
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 37
All Rights Reserved.
May 2009
Quartal – Customer Revenue (Story 5)
Identifying customer segment opportunities: 0. Top left:
(Older customers and low ARPU)
―Older‖
Customers
Customers who have been with us a long
time and have a low revenue per user.
1. Top right:
Split by Revenue@
(Older customers and high ARPU)
Rank History#
Customers who have been with us a long
time and have a high revenue per user.
0 1
2. Bottom left:
2 3
(Newer customers and low ARPU)
Customers who have been with us a short
time and have a low revenue per user.
Customers
3. Bottom right:
―Newer‖
Newer customers and high ARPU)
Customers who have been with us a short
Low ARPU High ARPU time and have a high revenue per user.
Rank Average Revenue Per User (ARPU) # History means the length of time someone has been a
Per Year customer. An ―older‖ customer means that they have been a
customer for a relatively longer period, than a new customer.
Split by Revenue Each quadrant is split by an equal amount of revenue.
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 38 @ Alternatively, could be split by ARPU.
All Rights Reserved.
May 2009
Quartal – Customer Segmentation
(Story 6)
Cluster 1 Cluster 2
Differences in customer
behavior—by customer
segment—can easily be
seen when the customer
segments are displayed
in the Quartal Super
Graphic.
Cluster 3 Cluster 4
The ability to change what is
displayed on the x and y axis
of the Quartal picture makes
Split by Revenue@
Rank History#
the Quartal Super Graphic a
powerful and intuitive
customer segmentation,
selection, and analytical tool.
Rank Revenue Per User (ARPU) Per Year
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 39 Split by Revenue
All Rights Reserved.
May 2009
Wrap-up
• Unlocking business value—fast and
economically.
• Volume, velocity and variety of data requires
new approaches.
• Good customer segmentation is the key to
unlocking value.
• Super Graphics enables you to unlock your
customer data.
Copyright © 2008 - 2009 – Business Intelligence Systems Solutions, Inc. 40
All Rights Reserved.
May 2009
Questions?
Speaker Contact Information
• If you have further questions or comments:
Richard Hackathorn
[email protected] Andrew Cardno, BIS2
[email protected]
