March 25, 2014
Charles Sayers
Experience Innovation
SapientNitro
csayers@sapient.com
predictiveintelligencemaking better business decisions with just enough data
what we’ll look at
 coping with data volume
and velocity
 dashboards: the good, the
bad, the ugly
 the guesswork behind most
predictions
 activity : straw modeling
 activity : data by design
© 2014 SapientNitro
we’re generating
a LOT of data
REALITY
© 2014 SapientNitro
%
of the data in the world today was
created in the last 2 years alone.
- IBM -
© 2014 SapientNitro
This is the volume of photos
uploaded to Flikr in a single day
© 2014 SapientNitro
we’re generating
too much, too fast
PROBLEM
© 2014 SapientNitro
© 2014 SapientNitro
dataisaccumulatedfaster
thanourabilitytounderstandit
ibm
We are drowning in information
and starving for knowledge.
- John Naisbitt
© 2014 SapientNitro
innovations in wearables
and biometrics will drive a
surge in physiological data
© 2014 SapientNitro
a new
wave of
data is on
the horizon
© 2014 SapientNitro
190,000people with deep analytic sills as well as
1.5millionmanagers and analysts will be needed by 2018
to fill jobs in Big Data
© 2014 SapientNitro
we can only process
so many numbers at
one time
PROBLEM
© 2014 SapientNitro
%
structuredandunstructureddata
of analytics projects will deliver insights based on
by 2015, more than
therearemanydatasolutions
tohelpyoumanageandmine
© 2014 SapientNitro
REPORTING
ANALYSIS
MONITORING
PREDICTION
what
happened?
why did it
happened?
what’s
happening now?
what might
happen?
predictive analytics
dashboards and scorecards
OLAP and visualization tools
query, reporting and search tools
BUSINESS VALUE
COMPLEXITY
BUSINESS
INTELLIGENCE
TECHNOLOGIES
Data types most beneficial for 
service and sales improvement
Data types most beneficial for 
market and trend analysis
Internal Databases
External Website Content
Social Media Content
Internet‐Connected Device 
Data
Location‐based Data
Internal Log Files
RFID Data
Sensor‐Generated Data
Competitive Intelligence
Customer Segment Data
Market Condition Models 
and Data
Sales Results and Forecast 
Data
Internal Customer Data
Third Party Market Data
Point of Sale Data
Combined in an actionable, real‐time context, 
data can drive better market analysis and improve 
customer intelligence
SOURCE: “Big Data and the Democratisation of Decision,” Alteryx
weneedtousedatatosolve
problems,notfindproblems
© 2014 SapientNitro
thevalueofintelligence
companies
that use
predictive intelligence
median ROI
145%
annual customer retention
+6%
companies
that don’t use
predictive intelligence
median ROI
89%
annual customer retention
-1%
VS.
© 2014 SapientNitro
trending
data
patterns of continuous
change
• sales
• conversion
• retention
• journey
dependent
variables
controllable factors that
influence change
• marketing
• pricing
• branding
• response
kpi’sthatinfluencethefuture
behavioral
data
patterns of contextual
interaction
• churn
• dwell time
• engagement
• task completion
uncontrollable factors
that influence change
• weather
• economy
• seasonality
• disasters
independent
variables
© 2014 SapientNitro
what’sthedifference?
analytics
understanding
the factual past
intelligence
VS. anticipating the
near future
© 2014 SapientNitro
dashboards
© 2014 SapientNitro
3 types of
dashboards
1. Strategic
2. Operational
3. Analytic
© 2014 SapientNitro
© 2014 SapientNitro
strategic
simple, concise; contains aggregate
metrics representing overall health
(Executive Dashboards, Financial Dashboards, Planning Dashboards)
© 2014 SapientNitro
operational
monitor real time operations; track
and alert deviations from the “norm”
(Network Monitors, Health Monitors, Manufacturing Dashboards)
© 2014 SapientNitro
analytical
interdependent, interactive; provide
ability to futurecast and test what-if scenarios
(Campaign Response, Sales Forecast, Investment Prioritization)
anatomyofapredictivedashboard
analysis
reportingmonitoring
prediction
© 2014 SapientNitro
anatomyofapredictivedashboard
analysis
reportingmonitoring
prediction
© 2014 SapientNitro
© 2014 SapientNitro
insight
is the key to
predictive intelligence
whichcomesfirst,
thedataortheinsight?
© 2014 SapientNitro
most
predictions
begin with
a guess
© 2014 SapientNitro
predictive
guessing
© 2014 SapientNitro
making structured and
informed guesses using
common sense and insight
predictiveguessing
usesunconventionalrelationships
© 2014 SapientNitro
predictiveguessing
doesn’trequirealotofdata
© 2014 SapientNitro
predictiveguessing
isaboutfindingtherightcontext
© 2014 SapientNitro
whatifwe’re
smarterthanwethink?
© 2014 SapientNitro
straw modeling
ACTIVITY 1
is about creating a visual hypothesis
based on readily accessible data,
common sense and intuition
© 2014 SapientNitro
guestimation worksheet
Storesales
-15
-10
-5
0
5
10
15
20
25
-10 -5 0 5 10
Net margin
sales margin
Abercrombie & Fitch
Amazon.com
Best Buy
The Gap
Home Depot
JC Penney
Macy’s
Nordstrom’s
Sears
Target
howwelldidtheseretailersperform?
1
2
3
4
5
6
7
8
9
10
© 2014 SapientNitro
takeaways
© 2014 SapientNitro
• “accuracy” is influenced
by data we (think) we
already have
• initial straw models
often point to logical
starting points
• all guesses should be
validated by real data
• it’s important to know
what we (and others) are
good at guessing
data by design
ACTIVITY 2
allows us to design educated
perspectives that point us to
relevant data clusters
© 2014 SapientNitro
list5innovationsyourcompanyis
consideringforinvestment
© 2014 SapientNitro
innovations
1
2
3
4
5
readiness
H/M/L
howwouldyouscoreeach
innovationbybusinessreadiness?
innovations
1
2
3
4
5
readiness
H/M/L
howwouldyouscoreeach
innovationbyvaluetocustomer?
© 2014 SapientNitro
innovations
1
2
3
4
5
value
H/M/L
businessreadiness customer value
Plotour
innovations
usingthis3X3
basedon
readinessand
value(H/M/L)
© 2014 SapientNitro
PRACTICAL INNOVATIVE
EMERGINGEXPERIMENTAL
Experimental
In-branch experiences involving
technologies and approaches that are in the
early stages of development and exploration
Practical
Market-ready solutions already available in-
branch to always-on consumers
Emerging
Promising in-branch experiences that will
require greater investment in infrastructure
to enable
Innovative
Future-facing experiences that offer greatest
opportunity in-branch differentiation both the
long- and short-term
readiness value
thisisourstraw
model
© 2014 SapientNitro
whatdatawillweneedtovalidate
ourguess?
© 2014 SapientNitro
readiness value
taking the
guesswork out of
guesswork
FINAL THOUGHTS
© 2014 SapientNitro
6 characteristics of
valuable data
1. Fast
Acquire, analyze and adapt in real-time
2. Centralized
Unify data repository and enable universal access
3. Actionable
Collect what you can use (not just what you can collect)
4. Diverse
Combine multiple data sources – across mode, place,
device, time and action
5. Retained
Data has value beyond original anticipation. Don’t
throw it away!
6. Scalable
Plan for exponential growth
capturethespark
Accept the fact that our best guesses
will require further refinement and
validation.
© 2014 SapientNitro
modelsimply
Complexity is confusing. Get to the point.
© 2014 SapientNitro
stayfocused
Concentrate on finding
actionable data
© 2014 SapientNitro
getdirty
Predicting the future is a messy business.
You won’t call everything right. Maybe some
of it will hit the mark. Maybe none of it.
It’s always better to err on the side of action,
than lose by doing nothing.
© 2014 SapientNitro
avoiddeflation
Don’t suck the air out of every prediction.
Strive to strengthen the strength of your
insights and rationale.
© 2014 SapientNitro
© 2014 SapientNitro
questions
thank you
Charles Sayers
Experience Innovation
SapientNitro
csayers@sapient.com
March 25, 2014
predictiveintelligence
making better business decisions with just enough data

Predictive Intelligence