Big-Data: What does it really mean to Your Organization?
The Key Challenges
Approach: Creating the right organization and framework
Gathering: Picking the right technology stack(s)
Analysis: Find Meaning(s) within the Data
2. Gary Angel, President of Semphonic
Co-Founder and President of Semphonic, the leading independent web
analytics consultancy in the United States. Semphonic provides full-
service web analytics consulting and advanced online measurement to
digital media, financial services, health&pharma, B2B, technology, and
the public sector. Gary blogis at http://semphonic.blogs.com/semangel
Scott K. Wilder – Partner @ Human1.0
Currently Founder and Digital Strategist at Human 1.0. Before that, Scott
was SVP/Social Media Architect at Edelman – Digital. Founded and
managed Intuit’s Small Business Online Community and Social Programs.
Before Intuit, Scott worked a AOL, Apple, Kbtoys/etoys, Borders, American
Express.Scott is also a founding Board member of the Word of Mouth
Marketing Association. He received graduate degrees from New York
University, The Johns Hopkins University and Georgetown University.
Scott’s blog is at http://www.wildervoices.com
Marshall Sponder – Founder WebMetricsGuru INC.
Marshall Sponder is an Author of the McGraw-Hill book, Social Media
Analytics, he is independent Web analytics, data and SEO/SEM specialist
working in the field of market research, social media, networking, and
Outbound Communications. Marshall is currently working with Principal at
WebMetricsGuru INC . Marshall also teaches Social Media Analytics and Art
at Rutgers University and UCI Irvine, Extension. Marshall’s blog is
http://www.webmetricsguru.com and book site is http://www.smabook.com
2
3. Agenda
• Big-Data: What does it really
mean to Your Organization?
• The Key Challenges
• Approach: Creating the
right organization and
framework
• Gathering: Picking the
right technology stack(s)
• Analysis: Find Meaning(s)
within the Data
3
4. The Big Data Shift
• More marketing dollars moved to digital.
• Data growing exponentially with more focus on Big Data
• Social and Mobile becoming increasingly important and
interconnected (and measureable).
• Companies in all sectors have at least 100 terabytes of
stored data in the United States; many have more than 1
petabyte and it continues to grow as more people are
online in social and mobile.
• Better ability to glean customer insights as a result of
improvements in semantic technologies.
• Desire to expose data externally (gov.org) and share it.
4
5. Org challenges
• Departmental:
• Analytic applications are often departmental by nature
• Departments deploy their own platforms for big data and analytics
• Many organizations today haven’t figured out how to leverage Big Data.
• Two thirds of executives believe that there is not enough of a “big data culture” in their
organization - this is particularly notable across the manufacturing sector
• Technology:
• Not all BI/DW technology stacks are designed for advanced analytics
• Lack of single digital platform
• Difficulty measuring effectiveness – unable to link data to individuals
• Complicated buying process/user experience
• Not adequately using data they already have
• Too much unstructured data to support decision-making
• Skills:
• Talent shortage
• Lack of expertise and experience
• Having a just-in-time agile mindset
• Ask the right questions
5
9. The new roles of digital marketers
Almost 60% of
organizations rely on ..
marketing to make
technology
recommendations
Leading to a mis-match
between the goals and
technology used to
execute.
2011 Digital Marketing 2.0 Study by research effort
between DataXu, SNCR and Human 1.0
9
10. But greater dependency on
in-house support and IT organizations
Over 60% of organizations
are relying more on
internal teams than
agencies
And only 35% agree that IT
is able to provide the tools
they need to optimize their
digital marketing
10
11. No organizational Kumbaya
Organizations struggle to make real-time decisions and to pull
insights from the large data sets created by digital marketing
CMO and CIO teams aren’t always partnering effectively
Strongly Agree Agree Neither Agree nor Disagree Disagree Strongly Disagree
40%
35%
30%
25%
20%
15%
10%
5%
0%
My digital marketing tools provide me with insights into how The CIO's team and the CMO's team in my organization have a My IT group's analysis of digital marketing data on consumer
demand for my organization's products and services vary in true partnership in using data to better understand the behavior permits real-time business decisions
real-time (depending on time of day, for instance) customer
2011 Digital Marketing 2.0 Study by research effort
between DataXu, SNCR and Human 1.0
11
12. So what’s the hold up?
60%
agree digital marketing can
reduce acquisition costs
However, a common
issue is not being
able to make a case
for and prove it to
company leadership Digital Marketing 2.0 Study by research effort
2011
between DataXu, SNCR and Human 1.0
12
14. Future Organizational Shift
CEOs will push for more analytics projects --they want to
exploit big data for growth
CFOs play bigger role on
signing off on costs
CMO will bring more
technical / business CIOs will play a bigger
intelligence types into with big data projects
their organization
14
16. Cardinality
Distinct Values per Variable
Traditional Data Systems relied on the ability to
• With lots of distinct values: aggregate most dimensions into small set of
– OLAP becomes difficult
distinct values to work well.
When your dimensions have lots of distinct
– Visualization is nearly impossible values (high cardinality), you’re dealing with
– In-Memory Systems struggle Big-Data.
16
17. Complex (and Dynamic) Relationships
Combining data from different tables:
• Joins put lots of stress on the
design
– Join strategies are complex and
hugely impactful
– Exposing the data model
becomes difficult
– Optimizing specific paths limits
query flexibility
Traditional Data Systems relied on a small number of static paths to expose
reporting data at the aggregate level.
When you have to join lots of tables and have unknown or dynamic needs to
combine data (all Analysis applications), then you are dealing with a big-data
problem.
17
18. Why Digital is Usually Big Data
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
• Lot’s of complex relationships and joins:
– Page -> Visit -> Campaign_Touch -> Visitor -> Channel
• Traditional variables don’t aggregate
meaningfully:
– Views, Page Time, Visits, etc.
18
19. It’s About Getting Your Hands on the Data
In the digital world, there’s little correlation between
size of enterprise and size of data. For most
organizations, the real challenges are around accessing
and integrating digital data regardless of it’s volume.
• Regardless of your data volumes, direct access to
the data presents new challenges to digital
analytics
– The need to model the data meaningfully
– New types of analysis and reporting
possibilities
– More complex technologies that aren’t always
SaaS
– New types of resource requirements and skills
19
20. Choice Vectors
Handling Very Large
Data
100
90
80
Ease of Management 70 Richness of
and Setup 60 Technology Stack
50 SQL-Server
40
30 Oracle
20 Teradata
10
0 Netezza
Availability of Aster
Ease of Integration
Expertise Hadoop
In-Memory
Appropriateness to
Cost / Size
Realtime
20
22. Good Questions Drive Results
Having the right organization for Big Data, choosing the right
technology, and developing a strong foundation for analysis are
ALL critical to success:
Organizational
Approach
Rich
Customer Technology
Segmentation Stack
Foundation
22
23. Thank you for your time
Gary Angel
gangel@semphonic.com
@garyangel
Blog:
http://semphonic.blogs.com/semangel/
Scott K. Wilder
scott@human1.com
@skwilder
Blog: www.wildervoices.com
Marshall Sponder
now.seo@gmail.com
@webmetricsguru / @smanalyticsbook
Blog: www.webmetricguru.com
23