Becoming Data-Driven Through Cultural Change

1© Cloudera, Inc. All rights reserved.
Understanding Your Data
Journey
Dave Shuman
Industry Leader: Retail, Manufacturing, & IoT
Larkin Kay
Industry Leader: Communication

Data is abundant
…and cheap.
Keep all data online
as long as needed.

Computation
is affordable.
Ask bigger questions
as fast as you can.

60%
50%
By 2017,
of big data projects will fail
to go beyond the pilot phase.
or fewer organizations will have made the
cultural or business model adjustments
to benefit from big data.
Gartner “Predicts 2015: Big Data Challenges Move From Technology to the Organization” – November 2014

Get the right people.
Ratify your process.
Adopt modern technology.

ANALYTICS
DATA
MANAGEMENT
INFRASTRUCTURE
Big Data
Management
Traditional BI and Analytics Big Data Analytics
Assemble the right team
Tightly aligned, Mix of experts and innovators
DATA SCIENTISTS,
APPS DEVELOPERS,
ANALYSTS
DATA ENGINEERS
ARCHITECTS

Big Data Functional Areas of Responsibility
Executive
Data Science
Data
Engineering
Architecture
Development and Insights
Foundation and Strategy
Run and Support
Vision and Goals

A traditional BI and analytics organization consists of three main
components.
Analytics
▪ Business-oriented teams that use data models and data analysis tools
to develop reports, find insights – often using samples of data
Data
Management
▪ Data modellers that take requests from business users, find data to
satisfy those requests, develop models to answer the users
questions, and load those models into a data warehouse
Infrastructure
▪ Hardware and software specialists responsible for network, storage,
server, and software components needed to satisfy the analytics
needs of the organization
Description
Staff for Success

Analytics
Data
Management
Infrastructure
In the Big Data world, the Data Engineering team becomes
strategic since data is your most important asset and can be
transformed and used many different ways
Big Data
Management
Architects
Data Scientists
Data Engineers
It is critical
that these
three roles
be tightly
aligned
Description
Staff for Success

Curiosity
Math &
Statistical
Knowledge
Hacking
skills
Subject
Matter
Expertise
The hybrid data scientist
• Subject Matter Expertise lies
in the business
• Hacking skills can come from
existing IT staff or new hires
• Staff at least one true Ph.D
statistician for model
oversight across all teams
Important character trait
Data Science
A luxury is finding one or more
data scientists that cross these
disciplines
Your Data Scientist Team(s)

Often a centralized Data Science team can partner with the
business to identify data that differentiates, explore use cases to
solve, and help to jumpstart business teams. Be mindful not to
overbuild centrally.
Agility ▪ The team must be able to learn quickly and adapt
Skills
▪ Hybrid skills of computer science (hacking), domain expertise and
at least one true statistician. Data Science training.
Teams
▪ Often businesses find the domain expertise in-house, add in MS/Ph.D.
candidates from local universities and hire that one true statistician
Experts
▪ This team must be the “data experts” for the entire company in
order to fulfil the vision of sharing data for maximum innovation
Description
Staff for Success: Data Science-as-a-Service

Training is one of the biggest keys to success. The opportunity
to be trained attracts talent and shapes careers. Leverage on-
site training for team-building and skills development.
Description
Data Scientist
▪ Data Science at Scale using Spark and Hadoop (3 days)
▪ CCP: Data Scientist (Certification)
Analyst
Data Engineer,
DevOps
▪ Developer Training for Spark and Hadoop (4 days)
▪ CCP Data Engineer (Certification)
Administrator ▪ Admin Training for Apache Hadoop (4 days)
▪ Cloudera Certified Administrator for Apache Hadoop (Certification)
▪ Data Analyst Training (4 days)
Staff for Success: Training is essential

Adopt an agile approach
Successful projects start small,
and iterate to success
Get Data
Explore
and Analyze
Deploy
1. Get data you already have, or
create new data.
2. Explore and analyze, quickly.
3. Deploy your application.
…and repeat. Add:
More data, more users, more use cases, more
complex analytics; go real-time!

Explore,
Enrich, Analyze
Operationalize
Collect,
Create
More ways to
serve your insights
New data sources to
your data asset
More complex
analysis
Add …
… over time
The Data Journey: Iterate to Success

Site clickstream
Unique visitors,
enter/exit URLs
CRM transactions
Service trends
Reports,
email outreach
Dashboards,
page design
Explore,
Enrich,
Analyze
Operationalize
Collect,
Create
Explore,
Enrich,
Analyze
Operationalize
Collect,
Create
Single Data Set Analysis

Clickstream + CRM
Cross-dataset identity matching,
path analysis of potential issues
Enriched profiles,
issue identification
Explore,
Enrich, Analyze
Operationalize
Collect,
Create
Multi-Data Set Analysis

Preference matching,
predictive offer creation
Enriched profile + inventory
Reports,
Page design,
Email outreach,
Recommendation engines
Explore,
Enrich, Analyze
Operationalize
Collect,
Create
Predictive Modeling

“Companies that don’t continue to experiment,
companies that don’t embrace failure,
they eventually get in a desperate position where the only thing they can
do is a Hail Mary bet at the very end of their corporate existence.
Whereas companies that are making bets all along,
even big bets, but not bet-the-company bets, prevail.”
Jeff Bezos
Business Insider Interview
December 2014

Lower risk
▪ Risk of funding long-running projects with limited business value is
small. Use daily results to improve the process or change course.
Lower costs
▪ Can run infrastructure, data and insights workstreams in parallel.
Avoids large build-out of infrastructure and data before insights.
Communication
▪ With clear short-term results, enables a continuous communications
stream showcasing results or failures
Team
▪ Can start with small team, and add additional scrum teams as value is
determined and investment is available
Agile methodology provides actionable results more rapidly and
measures the value gained at each step, in small iterations. Agile
should be applied to data and insights project workstreams.
Description
Leverage Agile Methodology to Reduce Risk

Agile Approach: Quick, Iterative and Necessary
• Individuals and interactions (over Processes and tools): self-organization and
motivation are important, as are interactions like co-location (physical and
virtual) and teaming people to create a data scientist (each person brings 2 out of
3 key skills- Comp Science, Math, Business knowledge).
• Working software (over Comprehensive documentation): live software, data,
schemas, etc is more useful and welcome than documents in all meetings.
• Customer collaboration (over Contract negotiation): requirements cannot be fully
collected at the beginning of the Use case development cycle, therefore
continuous business user involvement is vital.
• Responding to change (over Following a plan): agile methods are focused on
quick responses to change and continuous development.

Building a Big Data Culture

Description
An essential key to success is having a strong Executive Sponsor for the
overall Big Data mission including advocacy for creating/collecting data
and Business Stakeholders for individual use cases.
Profile
▪ An executive focused on change, and willing to take risk to ensure the
success of the business via the Big Data initiatives.
Education
▪ Use every opportunity to bring the topic in front of potential
sponsors and stakeholders. Share industry and business potential
ROI models (heeding the warning not to overstate).
Advocacy
▪ Build big data success stories from within the business. Advocate for
the use of data in new ways. Support the proactive collection of data
and lead the charge to assign value to data.
The Important Role of the Executive Sponsor

Description
Use different vehicles and forms to enable collaboration
Meetups
▪ Bringing together the larger big data across the company to share
interests, learnings, and wins.
▪ Team led
Big Data Days
▪ Transfer of information through executive led thought leadership.
▪ Include experts from across the business units, vendors, partners.
▪ Cross-domain focussed.
Hackathons
▪ Allow developers to build new applications designed to boost
business.
Communications and Collaboration

How an EDH is Different than Traditional Approaches
1. Economically feasible to store more data
2. Powered to predictably process large data sets
3. Ability to build your data asset at linear scale
1. Collect data in native format – enables agility
2. Build history of activity by collecting data prior to its use
3. You can have near real-time access to data, plus a view of history
4. Security at the data layer increases flexibility and ability to protect privacy
5. Create community data and drive innovation by sharing across your business
Extreme performance
and efficiency
Analytic agility

Start at Cloudera University
On-site training OnDemand training Classroom Virtual Classroom
www.university.cloudera.com

Becoming Data-Driven Through Cultural Change

More Related Content

What's hot

Viewers also liked

Similar to Becoming Data-Driven Through Cultural Change

More from Cloudera, Inc.

Recently uploaded

Becoming Data-Driven Through Cultural Change