big data maturity model a practical application cookbook
TRANSCRIPT
Big Data Maturity Model
A PRACTICAL APPLICATION COOKBOOK
September 2013
© 2013 SAP AG. All rights reserved. 2
However you look at it, Big Data is critically relevant
across industry
20%
15%
Improvement in all financial metrics
for those that leverage Big Data
(McKinsey & Co)
Organizations that will have
successfully achieved value from
Big Data by 2015 (Gartner)
90% Organizations that consider Big
Data an Opportunity to be mined,
not an IT Problem to be overcome
(TDWI)
© 2013 SAP AG. All rights reserved. 3
On the radar of Leadership
Big Data isn’t just one more technology
initiative. In fact, Big Data isn’t a
technology initiative at all; it’s a business
program that requires technical savvy
© 2013 SAP AG. All rights reserved. 4
Practical questions that must be addressed
How am I supposed to build this
business program?
What are the practical steps my
organization needs to take?
Who should lead this / is my
CIO’s team capable of delivering
on this or do I need to task
someone else?
When should I expect to start
seeing returns on this and where
will they come from?
How do I ensure we don’t boil the
ocean on this and turn it into
another Y2K?
What do we want to build as a
core capability and what do we
want to outsource to technology
vendors? What is the plan to
build the core capability?
Questions Executives are asking now
Comprehending the Vision
© 2013 SAP AG. All rights reserved. 6
The Vision: Turn Big Data into Intelligent Data, in real time
Answer &
analyze any
question
in real-time
Build a
photographic
memory of your
enterprise
Simplify your IT
landscape to transact,
analyze and act
instantly
Answer questions at
the speed of thought
Store and analyze
all data within your
enterprise
Access detailed
records from any
point in time
Personalized
Insights to
Everyone,
Everywhere
Build a culture of data-
driven decision making
Give all business
stakeholders access to
information wherever
they are
100101 011010 100101
Unlock
business value
from Diverse
Data Sources
Identify high-value
information from
mountains of data
Drive insights to
where they deliver
business value
Action 2
Action 1
Action 3
© 2013 SAP AG. All rights reserved. 7
Big data challenges IT departments are challenged with …
Action 2
Action 1
Action 3
Quick Insight,
AND
Best Action,
AND All Data,
AND
Long delays between
transaction, analysis,
and action
Complex queries can
take hours to answer;
and sometimes never
complete
Challenge identifying
high value insights
from mountains of
data
Difficulty providing
complete and
accurate information
for best decision
making
Complex architectures
to overcome current
DBMS limitations
Inability to effectively
store and analyze all
types of data
Right Audience,
BUT
Power users are a
bottleneck to getting
data to the right people
quickly
Difficult to drive insights
into business processes
and to mobile workers
100101 011010 100101
© 2013 SAP AG. All rights reserved. 8
Big Data high level process
The Big Data Process
Signal Processing: Focuses on connecting to relevant sources, monitoring changes, filtering which signals should be simply
noted, and which signals and data elements should be acquired into the architecture, either as a replication or an abstract.
Staging and Modeling: Information that is known in the platform requires analysis to identify which signals and patterns are
worth retaining and in what format, which sources are noise, or require additional processing. This can be a highly complex
area ranging from Data Quality improvements and auto enrichment all the way through to Predictive models to identify the
value of a source.
Analyze: Pattern finding, and understanding which Attributes are critical to understand and influence business outcome lies at
the heart of creating value through a big data program or initiative. Complex Analysis, pattern finding and driving true insights
too often is however the end of the road for big data projects.
Act & Automate: Value realization depends on being able to translate insights from Analysis into Action. These are either
behavior changes or moves to automate actions. It is here that scalable change is achieved and where business models are
effected.
Signal Processing
Staging & Modeling
Analyze Act & Automate
© 2013 SAP AG. All rights reserved. 9
Central Battle for “Big Data” – Data vs. Knowledge
Source: Martin Decision Sciences
Understanding the Nuances
© 2013 SAP AG. All rights reserved. 11
Big Data: A natural evolution of Analytics, one that we
only now have the ability to take advantage of
Business Intelligence Big Data
Raw Data
Cleaned Data
Standard
Reports
Ad Hoc
Reports
& OLAP
Generic
Predictive
Analytics
Predictive
Modeling
Prescriptive Modeling
What happened?
Why did it happen?
What will happen?
What is the best that
could happen?
© 2013 SAP AG. All rights reserved. 12
Leveraging Big Data requires capabilities beyond
traditional BI
Use Cases
People & Skills
Standards & Processes
Governance
Architecture
© 2013 SAP AG. All rights reserved. 13
Big Data Maturity Model is driven by Use Cases
Us
e c
as
e s
op
his
tica
tio
n
Business Value extraction through data insight
No Big Data Capabilities
Search for a significant item in big data
Understand the big picture from all available data
Generate change in response to shifts in data automatically or manually
Use big Data to predict outcomes and adjust processes accordingly
Often a single step
Coping Understanding Managing Innovating Ignorance
© 2013 SAP AG. All rights reserved. 14
Maturing from BI to Big Data is an iterative process
Use case
• Utilizing design thinking to identify Use case approach in line with Strategy
• Define the use case by Outcome
• Establish clear targets, SLAs, KPIs, for core use case
• Establish creative “free Friday” approach to Edge issues
Discovery
• Identify potential data sources and their quality
• Evaluate Technical requirements for connectivity (capture vs index)
• Create rapid prototype (visualization, model, Exploitation)
• Check End-User acceptance of prototype
First Evaluation
• Did the data meet the requirements for the outcome
• What is the likely cost vs benefit calculation
• How many core issues and edge issues remain and are their in your control to fix
• What is the realistic timeline to deliver the first benefits
• Does the maturity you encountered match the solution
Information Architecture Design
• Explicit definitions of information profiling, Source analysis, intervals, security, rules
• Explicit definition of component architecture and re-use (Integration, Staging, Transformation,….)
• Load management – explicit design of work-load management on functions such as NLP, Predictive models,….
• ILM policies, Information ownership, governance, master data control, meta data,…
Implementation
• Technical implementation
• Information flow testing
• Analysis
• Automation,
• Feedback loops, controls, change control,…
Steps to Get There
© 2013 SAP AG. All rights reserved. 16
Step 1: Define / Refine your Use Cases
Look at what your organization is already doing that could be improved - or what
others in your peer group are doing that could be emulated
What could be done better with Big Data techniques, such as mashing up social data
with product data, or developing new cost models based on consumer profiles?
Apply ‘Design Thinking’ to surface potential new use cases
This approach can help you to identify brand new, previously unconsidered use cases
(e.g., http://www.sap.com/campaigns/2013_05_design_thinking/index.epx)
Categorize use cases by outcome
These may be cost-focused rather than revenue-focused, e.g. by better predicting crime
hot spots you could redirect police patrols
Prioritize and validate the use cases
Rank them in order of anticipated benefits. If the goals are valid, what decisions need
to be made that would move the needle on those goals?
Establish clear targets, SLAs and KPIs by Stakeholder
If you don’t define success, how will you know what success looks like? Definition of
success can vary by stakeholder: be clear on how the Use Case will help each
© 2013 SAP AG. All rights reserved. 17
Step 2: Data Discovery
What data exists, what is its size and composition, where is it located,
how it is managed, what are its dependencies, how does it integrate
with other systems?
Check under the bed
Don’t overlook the ordinary
Mine opportunities in machine data
Confidentiality and compliance
Pay attention to neglected data
Look outside
Read the signs
See the forest for the trees
?
© 2013 SAP AG. All rights reserved. 18
Step 3: Initial Evaluation. Get started, experiment, and
learn along the way
Did the data meet the requirements for the outcome?
What is the likely cost vs benefit calculation?
How many core issues and edge issues remain and are they in
your control to fix?
What is the realistic timeline to deliver the first benefits?
Does the maturity you encountered match the solution?
© 2013 SAP AG. All rights reserved. 19
Step 4: Information Architecture Design
First, understand the
complete Logical Architecture
possible…
…then layer in Component
architecture to determine
where there are gaps
© 2013 SAP AG. All rights reserved. 20
Step 5: Implementation
Get executive buy-in, support and sponsorship
Allow business managers to translate the goals into
objectives and define success metrics
Put the necessary IT infrastructure, applications and governance processes in
place
• Technical implementation
• Information flow testing
• Analysis
• Automation
• Feedback
Learnings and Support
© 2013 SAP AG. All rights reserved. 22
Early Learnings from Big Data projects
• Form follows function: Start with “Business” not with Technology
• Prioritize on 2 dimensions: Business Value and Feasibility
• Determine which capabilities are core for you to build and which can you
outsource
© 2013 SAP AG. All rights reserved. 23
Help is at Hand