Building an Effective Data & Analytics Operating Model A Data Modernization Green Paper

Building an Effective Data & Analytics
Operating Model
A Data Modernization Green Paper
A Consulting Green Paper for CIOs, CTOs, CDOs,
CMOs, CFOs and CEOs
By Ranjan Bhattacharya, Julian Flaks, Anne Lewson, and Mark Hewitt

P.O. Box 211 Hampton Falls, NH 03844 | info@eqengineered.com | eqengineered.com | 617.448.4255
2
Green Paper Versus White Paper
The term white paper originated with the British government, and many point to the Churchill
White Paper of 1922 as the earliest well-known example under this name.
White papers are a way the government can present policy preferences before it introduces
legislation. Publishing a white paper tests public opinion on controversial policy issues and
helps the government gauge its probable impact.
By contrast, green papers, which are issued much more frequently, are more open-ended.
Also known as consultation documents, green papers may merely propose a strategy to
implement in the details of other legislation, or they may set out proposals on which the
government wishes to obtain public views and opinion.

3
Table of Contents
Green Paper Versus White Paper ...............................................................................................2
Executive Summary ....................................................................................................................4
Introduction.................................................................................................................................4
A Modern Data Ecosystem.........................................................................................................6
The Data Analytics Maturity Journey..........................................................................................8
Descriptive Analytics................................................................................................................8
Diagnostic Analytics ................................................................................................................9
Predictive Analytics..................................................................................................................9
Prescriptive Analytics...............................................................................................................9
Building the Data and Analytics Operating Model (D&AOM)....................................................11
Discover Phase......................................................................................................................11
Assess Phase.........................................................................................................................12
Roadmap Phase ....................................................................................................................12
Execute Phase .......................................................................................................................14
Conclusion................................................................................................................................15
Authors......................................................................................................................................16

4
Executive Summary
This is the age of analytics—information resulting from the systematic analysis of data.
Insights gained from applying data and analytics to business allows large and small
organizations across diverse industries—be it healthcare, retail, manufacturing, financial, or
others—to identify new opportunities, improve core processes, enable continuous learning
and differentiation, remain competitive, and thrive in an increasingly challenging business
environment.
The key to building a data-driven practice is a Data and Analytics Operating Model (D&AOM)
which enables the organization to establish standards for data governance, controls for data
flows (both within and outside the organization), and adoption of appropriate technological
innovations.
Success measures of a data initiative may include:
• Creating a competitive advantage by fulfilling unmet needs,
• Driving adoption and engagement of the digital experience platform (DXP),
• Delivering industry standard data and metrics, and
• Reducing the lift on service teams.
This green paper lays out the framework for building and customizing an effective data and
analytics operating model.

5
Introduction
As companies digitize their business practices, new business models and technologies create
vast amounts of data which offers the opportunity to gain strategic business insights.
Figure 1 below lists a number of key benefits that can accrue across the enterprise from a
mature data and analytics operating model.
Figure 1: Benefits of a Data and Analytics Operating Model
With state-of-the-art tools, industry focus and strong business cases, it might be surprising
that organizations often struggle to unlock the potential of data and analytics. However, data
is only ever simple when viewed from the outside.
Turning business data into actionable intelligence is a journey with many challenges along the
way. Just as the written word is notoriously open to ambiguities and interpretation, the world
of data lives within overlapping nuances of real world-problem domains, software system
behaviors, and the idiosyncrasies of data sets and their most outlying points of data. Truly
meaningful analysis of data can be challenged by blind spots in any of these overlapping
concerns. These challenges are compounded further when a system involves integrations
between disparate sub-systems each with their own data sources, or when legacy schemas
co-exist with modernized counterparts.
Since the challenges of data quality emerge from the entirety of a business, true data maturity
is best thought of at an organizational level. It is critical to adapt business operations to the
strategic vision of the business, while developing the right capabilities in terms of both
technology and talent. It is not enough to just add a powerful technology layer to existing
business processes.

6
Key operational gaps may exist in an enterprise in areas such as:
• Data collection and sustained management,
• Data hygiene and consistency,
• Data governance and compliance,
• Data security, and privacy, and/or
• Adoption of data best practices, modern data architecture, technologies, and tools.
A Modern Data Ecosystem
The figure below portrays the high-level components of a modern enterprise data ecosystem.
Figure 2: A Modern Data Ecosystem
The main components of this data ecosystem can be described as:
• Data generation and collection
• Data aggregation
• Data analysis
• Data governance

7
In modern organizations, the volume, variety, volatility, and velocity of incoming data is
breathtakingly diverse. Data can be derived from:
• Structured data from databases, flat files, and other external systems via APIs,
• Streaming data from real-time sources like mobile phones, wireless sensors, etc., and
• Unstructured data from documents, chat transcripts, images, audio, and video
sources.
The difficulties of interpreting unstructured data are more immediately apparent. However,
even the most structured data, whether meticulously normalized relational databases or
documents with exacting schemas, should be considered in relationship to the layers of real-
world domains and underlying software. Fields prohibited from having missing data may end
up with meaningless placeholder values in them, and seemingly clear datapoints may be
interpreted counterintuitively behind the scenes by obscure software pathways. Users will
often have unexpected understandings of the data which arise from industry-specific or even
organization-specific usage customs.
Once the source data is collected, it must be cleaned, transformed, and loaded into various
repositories. This may include Operational Data Stores or ODSs, enterprise or cloud-based
data warehouses, NoSQL databases, or data lakes.
Only after this information is available in the repositories can insights and predictions be
gleaned from the data. Common outcomes include:
• Easy to consume interactive reports and dashboards using reporting and business
intelligence tools
• Patterns and predictions using tools and techniques of data mining, artificial
intelligence (AI), and machine learning (ML.) (See note below for additional information
on the terms AI and ML)
Coupled to this end-to-end data pipeline, should be a governance structure, spanning
business and technology with policies around master data management, data handling ethics,
data quality, security, and privacy.

8
The Data Analytics Maturity Journey
To help assess where an organization is on its data journey, it is helpful to look at a widely
used maturity curve adapted from Gartner.
The X and Y axes represent maturity and business value.
Figure 3: The Analytics Maturity Journey
It is important to note here that as an organization embarks on this maturity journey, it is not
necessary to be at the end stage before it can start extracting benefits from the data. For
example, it may be possible to run predictive analysis on a subset of the data even when at
an earlier stage of maturity. In fact, it may be argued that this journey will never end, as an
organization will need to adapt continuously to new challenges in its business environment.
Descriptive Analytics
The data journey must be built upon a solid foundation of data collection & integration, data
hygiene, and data governance. Without this foundation, it is impossible to build a mature
analytics pipeline.
Once the foundation is established, an organization is ready to enter the “Descriptive
Analytics” stage, in which operational and ad-hoc reports can be created to answer questions
such as, “What Happened?”. These reports are typically static in nature and produced as part

9
of batch jobs by standard reporting tools. Tools like Microsoft Excel, with its pivot table and
VLOOKUP capabilities, are also great tools for data exploration.
Diagnostic Analytics
The second stage of data maturity is called “Diagnostic Analytics.” At this stage, an
organization can answer questions such as, “Why did Something Happen?”. Interactive
dashboards using visualization tools like Tableau or Power BI can depict data graphically and
allow users to summarize a lot of data quickly and explore the data to understand what is
behind the numbers.
For example, organizations in consumer-oriented businesses like retail, finance, and health
care can leverage visualization outputs to enable product and services personalization for
customers.
Predictive Analytics
As an organization evolves to the stage of “Predictive Analytics,” it is transitioning from an
operational to a strategic posture. The organization can now begin to build forward-looking
models to ask questions like, “What can Happen Next?”. Organizations can utilize
sophisticated statistical and ML modeling techniques to identify patterns and relationships
from the data, and start predicting trends.
As examples, predictive analytics can help identify and categorize customers based on risk,
profitability, or purchasing patterns, or identify fraud from anomalous transactions data.
Prescriptive Analytics
The final stage in the data maturity journey is that of Prescriptive Analytics. At this stage, the
system can tell the organization, “What is the Best Outcome Possible?”. At this juncture, the
data pipeline can help the organization optimize outcomes like revenue or profitability. This
can lead to truly transformational change in how analytics becomes a natural part of the
business.
Examples of such strategic optimizations in industries such as retail and manufacturing are
streamlining operations, logistics and supply-chain in real time, or anticipatory resource
scheduling across multiple locations.

10
A Note on Artificial Intelligence (AI) and Machine Learning (ML)
AI and ML are broad terms and are often used interchangeably. Both techniques are
based on analyzing large volumes of data and extracting patterns from it.
Specifically, AI is often used to refer to algorithms that can enable human-like
behavior in a machine. This kind of behavior can include problem-solving, decision-
making, and planning. ML on the other hand, refers to algorithms that can spot
patterns and identify anomalies in data that are hard for humans to see. ML systems
can be designed to be trained continuously on changing data and adapt its behavior.
The ability to create machines that can think, act, and learn independently of human
intervention has fueled a serious discussion of what is right, and what is enough, or
too much. Guildelines developed by Microsoft, Google, Apple, and others to ensure
transparent, principled, and ethical considerations around human dignity, rights,
freedoms, and cultural diversity are subscribed to, but much work remains to be
done. Where the line eventually gets drawn is ultimately our collective responsibility.

11
Building the Data and Analytics Operating Model (D&AOM)
A robust but flexible data and analytics operating model should not only support an
organization’s current needs, but also be adaptive enough for new strategic directions and
technological changes. To be effective, the model must accommodate existing organizational
and technological capabilities and resources as much as possible.
The high-level steps for building a Data and Analytics Operating Model are:
Figure 4: Phases in Building a Data & Analytics Operating Model
Discover Phase
For any organization, the Discover Phase is critical to understanding and identifying the
strategic imperatives of the business: “How will data and analytics be used to drive insights
and value?”. The answers to this question will inform the D&AOM design and architecture.
The three key influences on the D&AOM design are:
• Internal: related to the organization itself
• External: related to outside influences acting on the organization
• Foundational: related to all the factors that will impact the modeling initiative
Figure 5 below lists the various components of these three categories.

12
Figure 5: Discover Phase Components
Assess Phase
The Assess Phase helps define the Data Business Model and the Data Operating Model. The
Data Business Model identifies all the core processes, both internal and external, that
generate or consume data, across the entire data value chain. The Data Operating Model
touches all aspects of integrations across these processes and helps identify the key
technology and governance gaps across the data landscape.
Figure 6: Assess Phase Components
Roadmap Phase
The Roadmap Phase contains both a standardization and a planning step. Standardization is
key to ensuring data can flow consistently across systems, irrespective of its format or origin.

13
Figure 7: Roadmap Phase Components
This step also identifies the key components of the Reference D&AOM, shown below, as it
serves as the basis for planning and creating a roadmap.
Figure 8: Reference Data and Operating Model
The Reference D&AOM is like any other operating model but from the perspective of data. For
example, for the components listed in Figure 8:
• Manage Process: enables end-to-end integration of the worlds of business and data

14
• Manage Data & Analytics Services: responsible for all aspects of data governance
and management, from acquisition, processing, reporting, and analytics
• Manage Project Lifecycle: manages data-oriented projects, utilizing standard project
management tools and techniques
• Manage Technology/Platform: addresses all aspects of technology—architecture,
infrastructure, applications—and the related support and change management
The road-mapping activity which follows the standardization step should subscribe to an agile
approach of use-case creation, prioritization, and iteration planning, making sure that high-
value items are prioritized and represented at the top of the backlog.
The business use cases that are identified in the “Use case creation and prioritization” step
can be grouped into the broad categories identified in the Discover Phase:
• Internal use-cases focused on internal business process optimizations
• External use-cases focused on customer-facing areas like pricing, growth, customer
satisfaction and churn, effectiveness of marketing spend etc.
• Foundational use-cases focused on areas like predictive maintenance, IT demand and
cost optimizations, fraud detection etc.
The overall project plan should be grounded in the big picture, while delivering value
continuously through short- and medium-term goals. This is critical because the sooner the
organization can extract value out of the data model, the easier it justifies the cost of this
effort. It is important to first identify the business use cases that an organization will get value
out of and then think of the data and effort to operationalize them.
Execute Phase
The execution phase follows naturally from the agile planning phase, with iterative projects
that focus on outcomes, not operations, and a “fail-fast” and “test and learn” mindset that is
critical for success.
Figure 9: Iterative Execute Phase

15
Each project should focus on delivering a capability with a well-defined business value.
Iterative value delivery combined with the outputs of the Discover and Assess phases, should
help to mitigate the risks and challenges previously discussed in the data space. However,
risks and impediments should be reassessed as projects progress, to help empirically assess
and address their impact.
Conclusion
The data domain is accelerating. Given the sheer amount of data generated, the importance
of learning how to utilize the information available to an organization must become an
imperative. It is essential to upskill oneself, and one’s organization and team in the data
domain to not be left behind. Harnessing the capabilities of a mature data and analytics
practice will allow organizations to create significant value and differentiate themselves from
their competitors.
For an organization to become truly data-driven, and to speak the language of analytics in its
day-to-day operations, the entire organization must commit to the journey, adopt an agile
mindset, and bridge the gap between technology and business.
Though there is no one-size-fits-all approach, the above framework can help an organization
build a robust and adaptable D&AOM, which can provide consistency in approach and shared
understanding to foster data competency. The framework helps tie the D&AOM to the big
picture strategic imperatives, and at the same time, smaller iterative wins help to build
momentum and expose the possibilities which greater data maturity will bring.

16
Authors
Ranjan Bhattacharya – Chief Digital Officer
Ranjan is passionate about building technology solutions aligned with business needs,
intersecting data, platform, and cloud. He is a believer in delivering value incrementally
through agile processes incorporating early user feedback. Outside of technology, Ranjan
loves to read widely, listen to music, and travel. Ranjan has a BS in Electrical Engineering,
and an MS in Computer Science from the Indian Institute of Technology, Kharagpur, India.
Julian Flaks– Chief Technology Officer
Julian is a relentless problem solver and hoarder of full stack expertise. Having thrown
himself headlong into Internet technology when best practices had barely begun to
emerge, Julian is happiest putting his experience to use unlocking business value. Julian
holds a Bachelor’s of Laws from The University of Wolverhampton, England and a Master
of Science in Software Engineering from The University of Westminster.
Anne Lewson – Principal Consultant | Project & Program Management Leader
Anne blends technical with practical approaches to deliver projects ranging from large data
mergers to more detailed, analytical solutions for a wide array of internal and external
stakeholders. Anne leads the project management practice by using an applied Agile
methodology suited to our clients' requirements. Anne has a B.S. in Computer
Technology/Computer Systems Technology from University of Nantes and has her PMP
and PMI-Agile certification.
Mark Hewitt – President & CEO
Mark is a driven leader that thinks strategically and isn’t afraid to roll up his sleeves and get
to work. He believes collaboration, communication, and unwavering ethics are the
cornerstones of building and evolving leading teams. Prior to joining EQengineered, Mark
worked in various management and sales leadership capacities at companies including
Forrester Research, Collaborative Consulting, Cantina Consulting and Molecular | Isobar.
Mark is a graduate of the United States Military Academy and served in the US Army.

Building an Effective Data & Analytics Operating Model A Data Modernization Green Paper

Recommended

More Related Content

What's hot (20)

Similar to Building an Effective Data & Analytics Operating Model A Data Modernization Green Paper (20)

More from Mark Hewitt (18)

Recently uploaded (20)

Building an Effective Data & Analytics Operating Model A Data Modernization Green Paper