Confused about the right technologies for your digital marketing and analytics needs? You’re not alone. The challenges are complex and the range of possible solutions potentially bewildering.
In this webinar, Gary Angel, President and CTO of Semphonic, and Krishnan Parasuraman, CTO of IBM Big Data Solutions, demonstrate a common-sense approach to finding the right solutions for your company. Underlying the approach is a deep intellectual framework that highlights why digital marketing analysis requires new technologies and approaches. Using that framework, a whole range of specific digital marketing tasks (from full attribution analysis to social media marketing)are examined in light of the unique stresses each places on your technology stack. The result? A powerful way to match your specific business goals and requirements to the array of new technologies now coming online.
If you’re thinking about, evaluating, planning or designing a technology stack for digital marketing, this webinar is specifically for you.
What you’ll take away:
1. A much deeper understanding of why certain types of analysis in digital challenge traditional architectures – even when data volumes aren’t enormous
2. A true working definition of “big data” and a way to evaluate what it means beyond the hype
3. A detailed examination of the most common digital marketing business requirements in terms of their specific challenges to your technology stack
4. A good understanding of how IBM’s Big Data Solutions fit within that framework
See the recording of this webinar at http://www.semphonic.com/webinar-choosing-a-big-data-technology-stack.html
The Ultimate Guide to Choosing WordPress Pros and Cons
Choosing a Big Data Technology Stack for Digital Marketing
1. Choosing
a
Big
Data
Technology
Stack
for
Digital
Marke7ng
Gary
Angel
Krishnan
Parasuraman
President
and
CTO
CTO,
IBM
Big
Data
Solutions
2. Your
Hosts
Gary Angel, Semphonic President and Co-Founder
20+ years experience with BI & database marketing
15 years experience with digital measurement
Leading industry expert, speaker, blogger and Semphonic practice
leader for advanced analytics
Selected: Digital Analytics Association (formerly WAA) Most Influential
Industry Contributor: 2012
Krishnan Parasuraman, CTO Big Data Solutions, IBM
15+ years experience with Large scale distributed information systems
Background in product development, consulting and technology management
Leading authority on big data technologies such as massively parallel data
warehousing and Hadoop
Author of the book Harness the Power of Big Data
3. Talking
Points
for
today’s
discussion
• Challenges
with
Digital
Marke7ng
and
Analy7cs.
What
makes
this
problem
so
unique
and
different?
• Why
is
it
hard
to
use
tradi7onal
database
technologies
to
analyze
Digital
Data?
• What
type
of
framework
would
we
use
to
evaluate
and
select
the
right
technology
stack
for
Digital
Analy7cs
need?
• How
does
IBM’s
stack
address
the
needs
of
Digital
Marke7ng?
4. Introducing
Semphonic
Founded in 1997 and exclusively focused on digital measurement and digital customer analytics
Deep expertise in traditional Web analytics solutions (Omniture, GA Premium, IBM, etc.) AND in the
use of advanced technologies for warehousing, integrating, and analyzing digital data.
Practice focused on high-end customer analytics including:
Digital Segmentation
Site optimization and Personalization
Customer Analytics
Attribution Analysis & Media Mix Modeling
Digital Data Models for the Warehouse
5. Two
Worlds
Divided
BI
&
Customer
Analy3cs
Web
Analy3cs
and
Digital
• Tradi7onal
BI
and
Customer
Analy7cs
teams
have
deep
methods
and
powerful
tools.
But
digital
data
is
surprisingly
different
and
challenging.
• Digital
Measurement
professionals
lack
the
tools,
the
methodology,
and
the
exper7se
to
do
mul7-‐channel
analy7cs.
6. Our
Goal
is
to
Bring
Them
Together
Statistical Models
Proven Actionable
Online Behavior
Demographics Email Marketing
Database
Web
Marke3ng
Analy3cs
Database-Driven Event Driven Social
Old SaaS
List Enhancement
Customer Driven
8. Digital
Analy7cs
is
a
Paradigm
Big
Data
Applica7on
Digital
Measurement
is
a
paradigm
case
of
big-‐data:
• Lot’s
of
data
– Millions
(hundreds
of?)
events
per
day
– Lots
of
data
per
event
• Lot’s
of
key
High
Cardinality
variables
– Page
Name,
Product
Sets,
Referrers,
Campaigns,
Keywords
– and
Customers
• Focus
on
Detail-‐Level
Analy7cs:
– Customer
Life7me
Value
– Full
mul7-‐touch
aYribu7on
• Lack
of
meaning
at
the
Row-‐Level
– In
digital,
meaning
exists
in
a
collec7on
of
records.
9. Why
it’s
Challenging
These
unique
aspects
of
digital
data
make
it
difficult
for
most
tradi3onal
technology
stacks
to
support
effec3ve
digital
measurement
and
analysis:
Large
Row
Defeats
systems
not
setup
to
op3mize
full
table
scans
Volumes
ONen
creates
unmanageable
indexing
sizes
Creates
basic
load
and
availability
issues
High
Defeats
many
classic
OLAP
strategies
Cardinality
Forces
full-‐table
or
index
scans
of
the
data
Focus
on
Defeats
aggrega3on
strategies
Detail
Level
Analy3cs
No
opportunity
for
fixed
aggregates
to
succeed
Lack
of
Defeats
simple
aggrega3on
strategies
Meaning
at
the
Row
Level
Defeats
tradi3onal
row-‐based
ETL
11. Why
Digital
Data
IS
DIFFERENT
• Here’s
why
your
tradi7onal
BI
and
Customer
Analy7cs
folks
struggle
with
Digital:
– There
are
no
domain
experts
– Nearly
all
digital
data
is
stream
data
– Unlike
transac7on
data,
digital
streams
don’t
aggregate
cleanly
– Digital
Data
o]en
contains
a
hidden
topographic
structure
Most
modeling
Tradi7onal
data
Unlike
and
analysis
modeling
relied
Transac7on
Hidden
systems
provide
on
domain
data,
digital
Structure
skews
row-‐based
experts.
These
stream
data
Basic
Sta7s7cal
analysis.
don’t
exist
in
doesn’t
Analysis
Analy7cs
data
is
digital
aggregate
stream
data.
13. Aggrega7on
of
Streams
• Aggrega7on
of
streams
is
cri7cal
to
effec7ve
digital
measurement
– This
isn’t
because
of
performance
(though
it
helps)
– Digital
data
has
meaning
as
a
collec7on
not
a
single
row
– So
no
maYer
how
powerful
your
processing
system,
you
need
to
understand
whole
sequences
of
behavior
• Tradi7onal
aggrega7on
doesn’t
work:
Transaction Page View
Total Transactions Total Page Views
14. Why
Streams
MaYer
• The
single
biggest
driver
of
digital
analy7cs
measurement
is
the
need
to
de-‐silo
data.
• Proper
answers
to
ALL
of
these
ques7ons
require
mul7-‐channel
data
integra7on.
– Where
do
mobile
apps
fit
in
the
broader
customer
journey?
– How
does
web
engagement
translate
into
offline
sales?
– How
do
my
best
offline
customers
use
the
digital
channel?
– What’s
the
Predicted
Life7me
Value
of
a
Digital
Lead?
– What’s
the
value
of
a
Facebook
Fan?
– What
impact
does
Posi7ve
Social
ChaYer
have
on
Brand
Affinity?
– What
content
on
my
Website
is
most
effec7ve?
15. A
Quick
Primer
on
Joins
Joining
one
type
of
Customer
Record
to
another
yields
a
single
row
per
customer
with
an
easy
to
use
combined
record.
Joining
a
Customer
Record
to
Geo
or
Census
data
yields
a
single
row
per
customer
with
an
easy
to
use
record.
16. And
Why
Streams
are
Pain
Combining
streams
like
digital
and
mobile
–
even
with
a
join
key
–
just
yields
two
dis3nct
streams.
The
join
doesn’t
simplify
analysis.
Pu[ng
mul3ple
digital
data
sources
on
the
same
box
WITH
join
keys
doesn’t
really
solve
the
problem.
17. Every
Sub-‐Channel
has
Dis7nct
Streams
• One
of
the
HUGE
challenges
facing
a
digital
technology
stack
is
that
almost
every
digital
source
is
quite
different.
– One
of
the
most
common
failure
points
we
see
is
the
assump7on
that
pudng
the
data
in
one
place
makes
it
useful.
– Given
the
challenges
of
stream
analy7cs
in
a
single
sub-‐channel,
asking
the
analyst
to
join
streams
on
un-‐modified
data
is
overly-‐op7mis7c.
– An
effec7ve
data
model
has
to
provide
a
means
of
unifying
sub-‐channels
in
a
coherent
structure.
19. Don’t
Have
a
Homer
Moment
(D’oh)
• Crea7ng
a
strong
founda7on
for
assessment
begins
with
your
business
purposes.
Each
of
these
puts
different
stresses
on
the
underlying
technology
stack:
Advanced
Web
Customer
Analy3cs
Modeling
Personaliza3on
Email
Targe3ng
Site
Loyalty
Program
Merchandising
Enterprise
Personaliza3on
Analy3cs
Analy3cs
Dashboarding
Social
Media
Opera3ons
(Call
Analy3cs
Avoidance,
etc.)
20. Here
are
the
Key
Decision
Vectors
• We’ve
matched
the
business
func7ons
to
the
following
key
aYributes
of
various
big
data
technology
stacks:
The
goal
is
to
help
you
assess
what
technology
trade-‐offs
best
fit
your
needs.
21. Decision
Vectors
Advanced
Web
Analy7cs
Advanced
Web
Analy7cs
&
Hadoop
Handling
Huge
Volume
Up3me/Load
90
Miminize
Data
Handling
Huge
80
Volume
Without
Disrup3on
70
Modeling
Up3me/Load
90
60
80
Miminize
Data
Without
50
70
Modeling
Minimize
Easy
Data
Disrup3on
60
40
50
Administra3on
30
Integra3on
Minimize
40
Easy
Data
20
Administra3on
30
Integra3on
10
20
0
10
Support
Integrated
0
Support
Integrated
Real-‐3me
Support
Marke3ng
Solu3ons
Real-‐3me
Support
Marke3ng
Solu3ons
Support
Support
Algorithmic
Support
BI
Tools
Algorithmic
Support
BI
Tools
Queries
Queries
Support
Stats
Exper3se
Available
Support
Stats
Tools
Exper3se
Available
Tools
Advanced
Web
Analy3cs
Hadoop
23. Most
Common
Failure
Points
Here
are
some
common
risk
points:
• Insis3ng
on
Too
Much
History
Data
Windows
• Using
a
single
technology
• Keeping
too
much
data
ETL
• Missing
Join
Keys
• Failure
to
Reckon
with
Streams
Integra3on
• Assump3on
that
a
key
is
all
that’s
necessary
• Ad
Hoc
Effort
instead
of
up-‐front
segmenta3on
Analy3cs
• Failure
to
understand
Topology
• Lack
of
structure
Data
Democra3za3on
• Tool
Complexity
• Single
Technology
Stack
Real-‐3me
• Unrealis3c
expecta3ons
24. Digital
Analy7cs
is
a
Paradigm
Big
Data
Applica7on
Digital
Measurement
is
a
paradigm
case
of
big-‐
data:
• Lot’s
of
data
– Millions
(hundreds
of?)
events
per
day
– Lots
of
data
per
event
• Lot’s
of
key
High
Cardinality
variables
– Page
Name,
Product
Sets,
Referrers,
Campaigns,
Keywords
– and
Customers
• Focus
on
Detail-‐Level
Analy7cs:
– Customer
Life7me
Value
– Full
mul7-‐touch
aYribu7on
• Lack
of
meaning
at
the
Row-‐Level
– In
digital,
meaning
exists
in
a
collec7on
of
records.
25. Digital
Analy7cs
is
a
Paradigm
Big
Data
Applica7on
Digital
Measurement
is
a
paradigm
case
of
big-‐
data:
• Lot’s
of
data
– Millions
(hundreds
of?)
events
per
day
– Lots
of
data
per
event
• Lot’s
of
key
High
Cardinality
variables
– Page
Name,
Product
Sets,
Referrers,
Campaigns,
Keywords
BIG
DATA
– and
Customers
PLATFORM
• Focus
on
Detail-‐Level
Analy7cs:
– Customer
Life7me
Value
– Full
mul7-‐touch
aYribu7on
• Lack
of
meaning
at
the
Row-‐Level
– In
digital,
meaning
exists
in
a
collec7on
of
records.
26. The
Big
Data
Plaform
Requirements
Analyze
Extreme
Volumes
of
Data
Impressions
Online,
Offline,
Social,
Behavior,
First
Party
&
Cookies
Third
Party
across
mul3ple
channels
Online
Registra3ons
Purchase
Transac3ons
Analyze
Wide
Variety
of
Data
In-‐Market
Intent
Structured
–
POS,
3rd
Party,
Transac3ons
Unstructured
–
Social,
Video,
Blogs
Influence
Semi-‐Structured
–
Cookies,
Impressions
Sen3ments
BIG
DATA
Social
Followers
Analyze
Data
in
Real
Time
Recommenda3ons
Likes
PLATFORM
Product
Recommenda3ons,
Real
Time
offers,
Targeted
Ads
in
Real
Time
Psychographic
surveys
Geo-‐Demographic
Discover
&
Experiment
3rd
Party
Segments
Ad-‐hoc
analy3cs,
data
discovery
&
Offline
Transac3ons
experimenta3on
Responses
Governance
Enforce
data
structure,
integrity
and
control
to
ensure
consistency
27. IBM’s
Big
Data
Plaform
Impressions
Netezza
Cookies
• Extreme
Performance
Online
Registra3ons
• In-‐Database
Analy3cs
Purchase
Transac3ons
In-‐Market
Intent
• Scalable
Appliance
Influence
Sen3ments
Streams
BIG
DATA
Social
Followers
• Act
on
Data
“In-‐Mo3on”
Recommenda3ons
Likes
PLATFORM
• Real
3me
analy3cs
• Alerts/Ac3ons
Psychographic
surveys
Geo-‐Demographic
3rd
Party
Segments
Offline
Transac3ons
Big
Insights
Responses
• Hadoop/
Unstructured
Data
• Complex
Analy3cs
28. Audience
Q&A
Download
the
Full
Whitepaper
at:
hrp://www.semphonic.com/a-‐big-‐data-‐technology-‐stack.html
Learn
more
about
IBM’s
big
data
solu3ons
at:
hrp://www.ibmbigdatahub.com
hrp://www.analyzingmedia.com