The Role of Chief Data Officer in the 21st Century
| Cutter Consortium
Products & Services by Practice Area
Which Product Is Right for You?
CUTTER PREDICTS
Weather and Whither: IT Trends in 2015
with Joseph Feller, Dennis A. Adams, and Jim Love
Featured Consultant
Consulting & Inquiry
BI Training
Vol. 13, No. 2
by , Senior Consultant, Cutter C and Sid Adelman, Principal Consultant, Sid Adelman
& Associates
Fifteen years after enterprise resource planning and over
two decades into data warehousing, many business executives
are still frustrated over their inability to trust their
company's data. They have spent millions on new technologies,
only to find that the state of their data assets has
deteriorated. This significantly reduces the business value
of their investments. One big reason for this continuing data
chaos is that companies do not manage their data as a
business asset, and there is no one watching the store. In
this Executive Report, we look at the role of chief
data officer and why this role is so important.
Just imagine: a CEO approaches a CFO and asks for an
accounting of the company's financial assets. The CFO gives a
vague response indicating a lack of knowledge of the corporate
bank accounts, has little idea what is in each account, and has
no idea about the status of accounts receivable. When asked about
the intended use of the corporate assets, the CFO replies, "There
is no plan for their use." The CFO (and probably the CEO) would
soon be pursuing new personal interests.
Now imagine this: a CEO approaches a CIO and asks for an
accounting of the company's data assets. The CIO gives a vague
response indicating no knowledge of the inventory of data, has
little idea where the data is stored, and has no idea how much
data is duplicated and whether or not it is consistent. When
asked about the intended use of the data assets, the CIO replies,
"There is no plan for their use." Interestingly enough, the CIO
would not get fired but would keep the image of an executive not
that relevant to the organization.
BUSINESS CASE
How can we best put it? Most organizations do a terrible job
of using and securing data. Let's be clear: data is a critical
asset just as are inventory, cash, buildings, personnel, and
accounts receivable. Data is critical to running a business.
Without good data, we don't know our customers, employees,
products, suppliers, agents, status of our hotel rooms, seats on
a plane, or the status of our supply chain. Without good data, we
are unable to make the correct operational, tactical, and
strategic decisions that differentiate those organizations that
live and those that die. As you go beyond the headlines of failed
organizations or failed programs, you'll find important examples
of information that was unavailable, mishandled, wrong,
misunderstood, or misinterpreted. These failures led to the
demise of the organization.
We have organizations where the consumers of the data do not
trust the results, where countless hours are spent validating
data, and where important reports are being generated from
private databases and spreadsheets. And these information
consumers have every right not to trust the data as they weigh
the risks of taking actions based on suspect information. We have
incorrect reporting going to governmental and regulatory agencies
and public companies with exposures from SOX, financial
organizations with exposures from noncompliance with Basel III,
and healthcare organizations violating HIPAA regulations. With
this incorrect reporting come serious fines and the loss of trust
from stock analysts, customers, suppliers, banks, and regulators.
In some cases, it even results in the abrupt resignation of the
CEO or other senior managers.
And the problem is, no one is watching the store! No one in
the organization takes responsibility for the quality, the
standards, the meaning, the security, the metrics, the
integration of data, or the coordination of data among the
various divisions and lines of business (LOBs). There are no
tie- no one has the authority to make determinations,
set standards, and assess the tradeoffs of risk, cost, effort,
responsibility, and accountability.
There are very few organizations that have stepped up and
established the role of a chief data officer (CDO) with the
responsibility and authority to deal with the many and complex
interrelated issues, both political and technical, that swirl
around data. This Executive Report makes a strong case
for the CDO position, along with the role this person must play.
It also suggests the talking points to sell the position and how
to counter many of the possible oppositions to creating this
position in the organization.
BUSINESS VALUE
A company's data has value. However, to date, data assets have
not been shown directly on companies' books, although it is
sometimes there as goodwill, and sometimes it is partially
reflected in the price of the stock. The following examples make
the value of data more apparent and should justify formal
enterprise-wide management of data assets, and thus the position
Companies That Sell Customer Data
A number of companies sell information about their customers.
Some of this is credit data that includes income, demographics,
credit experience, and credit scores. Companies also collect
information about customers' buying habits from infinity or
loyalty cards. This data is sold back, sometimes to the
organization originating the data, sometimes to competitors, and
always to organizations that feel that by knowing more about
their customers, they can target market more intelligently and
cost-effectively.
Internal Information Gathered About Customers
Companies capture data on what their own customers are buying,
which ones are returning merchandise, and what services they use.
In addition, companies should capture the concerns, kudos,
suggestions, and complaints of customers. It is estimated that
for every customer who expresses him or herself, there is a much
larger number of customers who have the same opinion but don't
let you know. Several companies miss the signs of
dissatisfaction, and the response to problems is slow, expensive,
and often results in a public relations disaster. This
information can also identify the customers who habitually
purchase underpriced teaser products, which causes the company to
lose money.
Call Center Data
Companies that have call centers capture data for types of
calls, requests for information, problems, complaints, and
suggestions from customers. By analyzing this data, companies can
take appropriate action that might stop a customer from leaving
for a competitor. In addition, companies can use this data to
effectively target market to customers, even some disgruntled
ones. Information about a potential problem can route the call to
a supervisor or the department that specializes in rescuing
customers about to jump ship.
Click-Stream Data
Almost all companies have a website available to their
customers, potential customers, partners, suppliers, and
competitors. In addition to collecting their names, addresses,
email addresses, and comments, the websites also capture how
their users navigate through the site, where they click, what
information they read, and what information they act on. This
data provides insights into what your customers are considering,
and it opens opportunities for selling new products and services.
It also can help turn shoppers into buyers and can promote
products that the online customer is likely to purchase.
Demographics
Demographic data about customers includes gender, age, income,
assets, home ownership versus rental property, interests and
hobbies, education level, number and age of children, residential
location, if customers have a second home, information about
investment preferences, and even magazines read. All this
information is invaluable in how you market to and serve your
customers, and how you can capture a larger percentage of their
wallets and loyalty.
Channel Preferences
Most companies allow their customers to use various channels
to transact business. Some channels are more expensive than
others. The idea is to move customers into a less costly channel
or to offset expenses by charging for the more expensive ones.
For example, banks encourage their customers to use ATM machines,
telephone companies urge customers to pay their bills
electronically, airlines have gone so far as to close their
offices in cities thereby forcing their customers to book online,
and so on. Learning about a customer's channel preferences gives
the company the information it needs to support preferred
channels, make them more affordable, and keep the customer who
would change providers if a preferred channel was either
unavailable or not known to exist.
Direct Retailers
The companies that send out catalogs need to know what their
customers have bought in the past, their customers' propensity to
only purchase specials, and the specific types of items their
customers are interested in. Many of these direct retailers send
out specialty catalogues that target, for example, women's
clothes or climbing gear. Accurate data is critical to these
direct marketers. Since these catalogs are expensive to print and
mail, reducing the number of catalogs distributed saves an
expense that goes right to the bottom line.
Loyalty Cards
Loyalty cards are also known as membership cards or club
cards. Retailers can capture purchasing information and market
basket data in addition to some personal customer data. When
integrating this information with coupon usage and the customer's
demographics, the retailers can better understand price
sensitivity for upcoming promotions. This data is of such
importance to retailers that they are willing to pay for it.
Another example is casinos. They are able to watch the activity
of their patrons and reward serious play with free hotel rooms
and discounts at the buffet. This tactic is extremely effective,
as their patrons are incredibly loyal.
Travel Data
Airlines, hotels, and travel websites such as Travelocity and
Expedia capture travel-related data. They know what flights
customers have taken, their status with the airline, whether they
flew business or coach, their seat preference, how they purchased
their tickets, and if they ordered special meals. Hotels know how
many nights a customer spent in their hotel, if the stay was
personal or business or for a conference, what the customer ate
at their restaurant, what videos he or she watched in the room,
special requests such as an allergenic pillow, smoking
preference, Internet connection usage, and amenities selected
(e.g., room service). Rental car companies know their customers'
automobile preference, how often customers rent, accident history
with their rentals, and if customers rented the car for business
or pleasure. This incredible wealth of data allows airlines,
hotels, travel websites, and car rental companies to target
market to customer preference and the ways their customers like
to be serviced.
PROPOSAL FOR A CHIEF DATA OFFICER
In the Information Age, it is indisputable that data is a
profitable commodity that companies can sell. Therefore, it is
equally indisputable that all companies need to elevate their
data assets to the same level of importance as their other assets
and to create the position of a CDO who is in charge of these
precious data assets.
Qualifications
Dr. Anne Marie Smith, principal consultant at Alabama Yankee
Systems, LLC, summarizes the qualifications for a CDO position as
Bachelor's degree in finance, business, computer science,
or other relevant field
Fifteen-plus years of relevant data management experience,
including architecting data management solutions and
designing/developing data governance policies/standards
Strong project leadership and management skills to lead
organizational change to effectively meet strategic and
tactical goals
Strong written and verbal communicat
must be equally comfortable discussing the enterprise
information management (EIM) strategic perspective with
executives and implementation details with operational staff
Organizational/political agility with the ability to drive
large, cross-functional data management programs involving
coordination with multiple stakeholders
Ability to help define and articulate the enterprise data
management strategic vision and translate it into tactical
implementable steps
Solid knowledge of the organization's industry and its
challenges in the use of data and information
Certification in data management and understanding of the
various technologies of data management within an
organization
Strategic decision-making skills with a high degree of
Importantly, in an interview with Derek Strauss, CDO of TD
Ameritrade, he places a lot of emphasis on the second bullet
above by saying, "The most important qualification for a CDO is
that he or she must be a data person" -- not a database person,
but a data person.
CDO Activities
When asked about specific activities that a CDO typically
performs, Anne Marie Smith provides the following guidelines:
Provides vision and strategy for all data management
initiatives
Is a champion for global data management, governance,
quality, and vendor relationships across the enterprise
Is responsible for data management activities performed by
the EIM program, the business data stewards, and data service
Works with executives, data owners, and data stewards to
achieve data accuracy and process requirement goals for all
internal and external customers
Establishes data policies, standards, organization, and
enforcement of EIM concepts as established by the
organization
Leads the data governance council as executive sponsor
Is responsible for reporting on progress of enterprise
data management governance, including metrics
Leads the creation of program business definitions and
data management goals and principles for execution by the EIM
Oversees the monitoring of data quality efforts within the
organization and provides a central authority for the
resolution on data management issues that cannot be resolved
by the data governance council
Establishes data vendor management strategy and provides
oversight to support implementation by the EIM program and
coordinates with the IT organization through the CIO/CTO
Oversees the education of the organization on data
management concepts, the appropriate usage of data,
enterprise master data management and data quality concepts,
enterprise decision-support concepts, data vendor
capabilities, definition and appropriateness of data
management, rules on data access, and other data-related
Has executive responsibility for enterprise
information/data management budget and data-related systems
initiatives
Oversees evaluation of all data movement projects and
ensures the enforcement of the data strategy
RESPONSIBILITIES AND AUTHORITY
Principally, the CDO establishes and enforces a data strategy
for the organization. A data strategy includes, at minimum, the
following components, which we examine in detail.
Data Governance
According to the Encarta Dictionary "governance" is
defined as "having authority over something, controlling or
restraining something, having influence over or being the law for
something." Translating that to data governance, it means having
authority over data, controlling or restraining data, having
influence over, or being the law for data. Author and consultant
Danette McGilvray defines data governance as:
The organization and implementation of policies, procedures,
structures, roles, and responsibilities that outline and enforce
rules of engagement, decision rights, and accountabilities for
the effective management of information assets."
She further notes that data governance ensures that the
appropriate people representing business processes, data, and
technology are involved in the decisions that affect them, which
includes escalation and resolution of issues, implementing
changes, and communicating resulting actions.
Data Quality
An easy way to determine which of the five levels of data
quality maturity applies to your organization is to look at your
organization's current data quality improvement
activities:
Level 1: Uncertainty. The technicians in
the organization stumble over data defects as their programs
crash (abend), or the businesspeople complain. There is no
proactive data quality improvement process in place.
Basically, the organization is asleep and doesn't want to be
Level 2: Awakening. A few isolated
individuals acknowledge the dirty data and try to incorporate
some data quality disciplines in their projects. However,
there still is no enterprise-wide support for data quality
improvement, no data quality group, and no funding.
Level 3: Enlightenment. The organization
starts to address the root causes of its dirty data through
program edits and data quality training. A data quality group
is created, and there is funding for data quality improvement
projects. The data quality group immediately performs an
enterprise-wide data quality assessment and institutes
several data quality disciplines.
Level 4: Wisdom. The organization
proactively works on preventing future data defects by adding
more data quality disciplines to its data quality improvement
program. Managers across the organization accept personal
responsibility for data quality. Incentives for improving
data quality replace incentives for cranking out systems at
the speed of light.
Level 5: Certainty. The organization is
in an optimization cycle by continuously monitoring and
improving its data defect-prevention processes. Data quality
is an integral part of all business processes. Every job
description requires attention to data quality, reporting of
data defects, determining the root causes, improving the
affected data quality processes to eliminate the root causes,
and monitoring the effects of the improvement. Basically, the
culture of the organization changes.
Enterprise standards should include a uniform and repeatable
system development lifecycle methodology. There should also be
common standards for data naming and the use of abbreviations and
acronyms. In addition, you need standards for metadata capture as
well as logical data modeling. Businesspeople should establish
data defect thresholds for the most significant data and have
data quality improvement standards tied to those thresholds.
Standards should include the identification of data owners and
data stewards on the business side. Testing and reconciliation
standards also apply to data quality improvement. Security and
privacy standards must be enforced. Standards are often linked to
In too many organizations, standards are not really standards
-- meaning they do not have to be followed. They are more like
recommendations or guidelines, and as such, they are often
ignored. In addition, employees are rarely rewarded for following
standards. Usually, speed is what is being rewarded, and since
standards can slow down an activity, the smarter workers will
choose success (speed) over a standard. A standard has to be
"thou shalt" rather than "you should."
Business Intelligence
Business intelligence (BI) does not refer to a product you can
buy or to an application you can build. Instead, it is an
architecture and a collection of integrated decision-support
applications and databases that provide the business community
easy access to business data.
BI uses data as a
strategic asset of the company that, when turned into information
and applied as knowledge, gives the company an advantage over its
competitors. For data to become a strategic asset, each unique
data element must be carefully managed and reused to provide a
maximum return on asset. If the data is not good enough, the
information coming out of BI will be substandard and should
probably not be acted on.
Data Warehousing
Different people have different and sometimes conflicting
definitions for the term "data warehouse" (DW). The Data
Warehousing Institute (TDWI) defines a DW as "a data structure
that is optimized for distribution."
Author Ralph
Kimball sees the DW as "nothing more than the union of all the
constituent data marts."
Other definitions include
the operational data store (ODS) and data marts. One thing all
agree on is that a data warehouse is an important building block
in BI; it is the "under-the-hood engine" of BI. Business
intelligence applications and products require consistent, clean,
and integrated data in order to generate accurate information
upon which business executives and managers can act. The
objective of a DW is to provide that basic pool of consistent,
clean, and integrated data.
Master Data Management
Master data is not transaction data. Master data is the
description of real-world entities, such as customer, product,
employee, supplier, and so on. Transaction data is the
description of real-world interactions among these entities, such
as order, sale, booking, and so on. Hyperion defines master data
as "a set of core data elements -- with their associated
hierarchies, attributes, properties, and dimensions -- that span
the enterprise IT systems." Phillip Russom from TDWI defines
master data management (MDM) as:
The practice of defining and maintaining consistent
definitions of business entities, then sharing them via
integration techniques across multiple IT systems within an
enterprise and sometimes beyond to partnering companies or
customers.
To that end, MDM incorporates business applications,
information management methods, and data management tools to
implement the policies, procedures, and infrastructures that
support the capture, integration, and subsequent shared use of
accurate, timely, consistent, and complete master
Enterprise Architecture
An enterprise architecture (EA) is a set of pictorial
representations (models) of the business, including business
functions, business processes, business data, supporting metadata
(business, technical, process, and usage), hardware, software,
databases, applications, job streams, organization charts,
resource matrices, and so on. An EA describes a set of business
actions performed on any real-world object in the course of
conducting business. Information architecture is a subset of EA.
It is an enterprise data model of master data, transaction data,
and the business activities that link them.
Every active organization has an EA by default, even if it is
not documented. When the architecture is not documented, the
business actions and business objects of the organization are
most likely duplicated many times and are not consistently
understood by everyone in the organization. The goal of
documenting the architecture is to avoid abusing, misusing, and
redundantly recreating unique processes or data, thereby losing
sight of the company's 360-degree enterprise view.
Enterprise Data Modeling
Enterprise data modeling is the most effective method for
discovering business rules and finding data collisions. An
enterprise data model (EDM) does not have to be constructed all
at once. Instead, the EDM evolves over time and may never be
completed. It does not need to be completed because the objective
of this process is not to produce a finished data model but to
discover and resolve data discrepancies among different views and
implementations of the same data. An EDM often starts out as a
high-level conceptual data model showing the core business
objects (entities) and their data relationships. But the biggest
benefit of enterprise data modeling occurs at the refined logical
level where normalization, cardinality, optionality, data quality
rules, and other data administration principles are applied.
Business Process Modeling
Business process models describe the current processes for all
business functions. These models show how the organization
performs its business functions. Like most models, there is a
conceptual level (i.e., context diagram) and detailed
decomposition of each major business process. Since business
processes use data, business process models (also known as "data
flow diagrams") show the data flows in and out of each process.
The data on the process models is further described on the data
models, and both are part of the enterprise architecture. Process
models are essential for business process reengineering as well
as for business process improvement initiatives.
Metadata Management
Anyone with a digital camera can find metadata on each
picture, which includes the brand and model of the camera, the
date and time the picture was taken, shutter speed, f-stop, ISO,
possibly the latitude and longitude of where the picture was
shot, and much more. We should ask no less of the organization's
important data, but we should be careful to only capture the
metadata that is useful or will be useful in the future. This
would include:
Business metadata, including business names, definitions,
and valid domain values
Technical metadata, including databases, data types,
lengths, processes, algorithms, and so on
Standards, incorporating the organization's policies and
procedures as they relate to data and processes
Sources of the data
Time/date of data capture
Transformations applied to the data
Excluded data (load statistics and error statistics)
Ownership (CFO, HR, etc.)
Security (public, company confidential, HR only, senior
management only)
Timeliness
Glossaries and abbreviations
Metadata is the DNA of all enterprise-class data
standardization and integration initiatives because it serves
three functions: documentation, navigation, and administration of
data assets. Metadata is relevant to both master data and
transaction data. Metadata practices have been evolving over
decades, but organizations are slow to adopt them, mainly because
they don't understand the importance of metadata. The value chain
of metadata reaches into the actual business processes in the
business community because business processes are reflected in
the data relationships (i.e., in the transactions between two or
more core business entities).
Unstructured and Big Data
Unstructured data includes textual data, such as social media,
emails, medical records and contracts, as well as pictures,
videos, and sensor data (e.g., intelligent bar codes and RFID).
Unstructured data has eclipsed structured data in terms of
volume, a condition that has become known as "Big Data" even
though its sizes are a constantly moving target, ranging from a
few hundred terabytes to many petabytes or even exabytes of data
in a single data set. Big Data is often defined in terms of the
three Vs: volume of data, variety of data, and velocity of data,
which present certain challenges with its capture, storage,
search, security, sharing, analysis, and visualization.
The primary job of the CDO is getting an inventory of the
organization's unstructured/Big Data, its format, security,
ownership, and quality. He or she must also identify special
processing and storage requirements as well as advanced analytics
opportunities for its use. The CDO needs to be cognizant of the
value of the exploding data volume. Just because data can be
captured, it doesn't mean it should be, particularly if it is not
cost-effective and provides little or no significant business
Data in the Cloud
The challenges of Big Data have made cloud storage an
attractive option, especially to startup companies. The obvious
benefit is not having to implement and manage an inhouse,
scalable, fault-tolerant storage system. However, the costs and
time it takes to upload large quantities of data into and out of
a cloud could mean that businesses that crunch data sets could be
hit by unforeseen costs.
Another issue that
information experts, computer scientists, and entrepreneurs
debate is the concept of data ownership. Who owns the data stored
in a cloud system? Does it belong to the client who originally
saved the data to the hardware? Does it belong to the company
that owns the physical equipment storing the data? What happens
if a client goes out of business? Can a cloud storage host delete
the former client's data? The CDO has to carefully consider what
data to put on the cloud, recognizing the cost and any risks to
exposing the data outside the company.
Business Performance Metrics
Every senior manager and LOB manager has measures on which
they are evaluated. These would include budget, revenue, costs,
employee turnover, price of the stock, cost of capital, customer
satisfaction, and so on. The CDO would know of all these business
performance metrics and would know all the data that supports
those metrics. He or she would be aware of any problems in the
data and would take steps to rectify the problems before they
appear on the performance dashboards. A performance dashboard is
really a performance management system. It provides timely
information and insights that enable businesspeople to improve
decisions, optimize processes and plans, and work
proactively.
Business performance metrics are an
important component of business intelligence.
Security and Privacy
Not all data has the same requirements for security and
privacy. For example, data that is available on the
organization's website is exposed to the whole world and would be
classified as "public." On the other hand, the financial
information with data on profit for the next quarter needs to be
secure to avoid insider trading and trading by others outside the
company who would profit from knowing about extraordinary profits
and losses. Further, if confidential patient information, which
is secured under HIPAA, is breached, it will result in regulatory
fines and public relation erosion. The CDO works closely with the
organization's security office and with the owners of data to
determine the required level of data security.
Intellectual Capital
This might be the most difficult to sell and the most
difficult to implement, but it has the promise of exceptional
value. Intellectual capital is:
How the business is run. The documented
portion is in policies and procedures (sometimes current,
often not), which are too often only partially documented and
are sometimes incorporated in training materials.
Glossaries. This includes organizational
definitions, industry definitions, and meanings of acronyms
and abbreviations.
Notes, cheat sheets, private glossaries, name and
contact rosters. These documents provide assistance
with recalling facts and figures that otherwise are not
documented anywhere else, as well as with knowing the right
people to contact inside and outside of the organization for
information, approvals, help, support, and so on.
Knowledge in employees' heads. This
knowledge includes how to run the business, how to deal with
problems, and much more.
ORGANIZATIONAL STRUCTURE
The first question to ask is, "To whom does a CDO report?" and
the second question is, "Who reports to a CDO?"
To Whom Does a CDO Report?
The most optimal reporting structure is to the CEO as a peer
to the other chief officers in the organization. A good
alternative is for the CDO to report to the CFO. Less optimal is
a reporting structure to the COO. However, this reporting
structure can be effective if the COO is also a "data person," as
illustrated in our case study (see sidebar). Clearly, the CDO
position is not a technical position and should therefore not
report to a CIO or CTO in IT.
Q: What authority do you have?
A: Our emphasis is on collaboration. We
have a center of excellence approach and a staff compliment
that acts as facilitators. We also set policies and standards
for data governance, and we enforce data standards. We
communicate the benefits of treating data as an enterprise
asset -- for example, maximizing growth opportunities,
achieving efficiencies, achieving customer intimacy, and so
Q: What approach are you taking to exercise
that authority?
A: My approach is two-fold: show the
business benefits first and deliver some of them, and then
close the loop with governance and standards. Analytics is the
carrot. But in order to have sustained success with analytics,
we need governance and standards. First we want to get a couple
of sponsors in the business who will champion the initial
efforts. Our COO has a strong influence in the organization,
and I rely on him for support as well. I believe we will
incrementally influence the enterprise.
Q: What skills, capabilities, and
experiences should a CDO have?
A: The most important question when hiring
a CDO is: "Are you a data person?" In other words, does the
candidate understand what is required to manage data as a
business asset? Clearly, that person must have had some
exposure to data management activities like data governance or
enterprise information management to be able to organize this
new function. And the person needs to understand change
management and collaboration models.
Q: Where should a CDO come from?
A: At the MIT CDO Forum, some came from the
business and some from IT. It was about a 50/50 split. From the
eight CDOs attending the conference, four were from IT, and the
other four were from the business. The other attendees who
worked in the role of CDO came from varied backgrounds. One was
a CIO at a law firm. One did the work of a CDO at the US Navy.
One came from the business competency center of the British
Army. Others were CIOs, research guys, standards and
integration managers. Again, what matters is that you are a
"data person" who understands that data is a critical resource,
and that it needs to be controlled and governed.
Q: How do you measure if you are
successful?
A: We are still defining success, measures,
and metrics. They are based on a two-year roadmap with business
benefits realization timeboxes every six months.
Q: Thank you very much for sharing your
personal experience as a CDO in a major US investment firm. Are
there any parting thoughts for our readers?
A: Companies will not be able to afford to
ignore their data problems much longer. I suspect that there
will be many different ways companies will implement the role
of CDO. It will take patience and endurance to "turn the ship
around." It is always best to achieve success by using carrots
instead of sticks. And, you need to take a holistic approach in
the organization.
Derek Strauss is CDO at TD Ameritrade Holding
Corporation (NYSE: AMTD). He has 30 years of IT industry
experience, 25 years of which were in the information resource
management (IRM) and business intelligence/data warehousing
fields. Mr. Strauss established data resource management,
architecture, and IRM functions in several large corporations
using the Zachman Framework as a basis. He coauthored the
book DW 2.0: The Architecture for the Next Generation of
Data Warehousing with William H. Inmon and Genia Neushloss
(Morgan Kaufmann, 2 July 2008).
Who Reports to a CDO?
The answer to this question is not as straightforward as it
may seem because "reporting to a CDO" takes on two flavors:
direct reports (solid lines) and indirect reports (dashed lines),
as illustrated in Figure 1.
Figure 1 -- Suggested direct (solid lines) and indirect (dashed lines) relationships to a CDO.
Direct Reports
In order to carry out responsibilities and exercise authority,
the CDO needs to have the following positions report to him or
her directly:
Enterprise architects -- modelers who
create and maintain documentation of all components of an
enterprise architecture, such as hardware, firmware,
databases, programs, job streams, database schemas, business
models, organizational reporting hierarchy charts, HR
matrices, and so on.
Information architects -- data modelers
who create and maintain the project-specific logical data
models (business data models) and the enterprise data
Data analysts -- business analysts who
inspect, cleanse, transform, and participate in modeling data
with the goal of highlighting useful information, suggesting
conclusions, and supporting decision making.
Business modelers -- process modelers who
have an in-depth understanding of the industry and their
organization, and who create and maintain the business
process models.
Metadata administrators -- technicians
who are knowledgeable about metadata repositories, industry
metadata standards, and who create, populate, and maintain
the company's metadata repository.
EIM staff -- also known as data
administrators or data resource management professionals who
work with data stewards on the business side to create,
apply, and enforce data standards across the
organization.
Database architects -- also known as
DBAs, database designers, or database modelers who are
responsible for creating and maintaining the physical data
models for database schemas of enterprise-class data
integration databases, such as an enterprise data warehouse
and collectively architected data marts.
Data integrators/data developers --
developers who code the front-end BI applications, such as
performance dashboards, and developers who code the back-end
extract/transform/load (ETL) programs in an enterprise-class
data integration environment, such as an enterprise data
warehouse and collectively architected data marts.
Indirect Reports
As Derek Strauss points out in our case study, it is important
for the CDO to have strong relationships with individuals who
have other data roles on the business side as well as in IT (see
). The following data roles are all
possible indirect reports to the CDO:
Data scientists -- high-ranking
professionals with the training and curiosity to make
discoveries in the world of Big Data analytics. Their skill
set includes the ability to find and interpret rich data
manage large amounts of data despite hardware,
software and b merge data sources
ensure cons create
visualizations to aid in un and building
rich tools that enable others to work effectively.
Data miners -- statisticians who utilize
methods of artificial intelligence, machine learning,
statistics, and database systems to discover patterns in
large data sets. Their skill set includes database and data
management aspects, data preprocessing, model and inference
considerations, "interestingness" metrics, complexity
considerations, postprocessing of discovered structures, and
visualization.
Business analysts -- SMEs who understand
the industry and their organization and who participate in
requirements modeling, prototyping, data modeling, dispute
resolution, testing, and training.
Data owners -- business executives
(frequently LOB managers who are also data originators) who
have the authority to make policy and establish business
rules for their set of enterprise data.
Data stewards -- SMEs who are accountable
for the data under their control. They are responsible for
defining their data, auditing their data, enforcing data
quality rules, and working with the EIM staff.
Data custodians -- any IT technician who
touches data, such as a database architect or DBA, who is
responsible for the physical databases, or a developer who
creates, updates, or deletes data from files and databases. A
data custodian must be mindful not to introduce errors into
his or her processes that could corrupt the data in the files
and databases.
MEASURES OF SUCCESS
How will the CDO know whether he or she has been successful
and, more importantly, how will the CDO represent that success
(or failure) to executive management? Some areas of
responsibility easily lend themselves to metrics, while others do
Data Governance and Data Quality
By profiling the organization's important data, you will
establish multiple baselines for comparisons as the data is
improved. Some of the comparisons will be on missing data,
invalid values, incorrect data types, and aberrant values. Are
these data errors declining? Is the company moving up on the
information quality management maturity grid?
Have standards been documented, published, and accepted? Are
they complete, and are they being followed? A good deal of this
measure is subjective, and it will require surveying those
employing the standards and those affected by the standards.
Business Intelligence and Data Warehousing
Is the data in the DW sufficiently easy to access? Is it
trusted by those running the queries and reports and trusted by
the recipients of the information? Does the data in the data
marts reconcile to the DW and to the source systems? How much
effort is spent on validating the results of the queries and
Master Data Management
Does the organization have a gold copy of master data in an
MDM engine or in a homegrown database? Is master data
successfully synchronized across all IT systems? Is master data
successfully integrated into transaction systems and the DW? Can
master data be removed from the transaction systems?
Enterprise Architecture, Including Data and Process
Do the application developers consult the enterprise
architecture? Do they reuse the information on the various EA
models in their work? Is the EA current enough that it can be
trusted? Is the EA current and accurate enough for maintenance
programmers to use?
Metadata Management
What type of metadata about the most important data has been
captured and published? Is it being used by IT and also by the
business? Is metadata being incorporated into queries and
applications?
Unstructured and Big Data
Can the unstructured data be easily understood, and can it be
integrated with the appropriate structured data? Is the
unstructured data supported by metadata? Is Big Data
appropriately filtered and sufficiently secured?
Data in the Cloud
The measure of success of having data in the cloud is totally
dependent on why the cloud was chosen in the first place. It may
have been chosen because the data volumes were expected to
explode and the organization's infrastructure was unable to
handle the volume. In that case, is the cloud able to accommodate
the large amounts of new data? Is that data readily accessible
and usable? Is the data in the cloud sufficiently secured? Maybe
the cloud was chosen because the cost of expanding the
organization's infrastructure to accommodate the new data was
prohibitive. If that cost was estimated, the value would be the
depreciated estimated cost minus what the organization is being
charged for the use of the cloud.
Business Performance Metrics
Business performance metrics allow management to track the
efficiency and effectiveness of their various business units. The
business performance metrics can be placed in three
categories:
Extremely useful. We couldn't live
we trust the metrics, and they are crucial to our
success and to our continuous growth and improvement.
Useful. We look at the metrics with some
concerns about their accuracy and timeliness, and we make
decisions based on them after validation.
Not useful. We don'
we use other information sources, usually from our own
departmental systems.
Security and Privacy
It's probably best to have an outside organization
specializing in security perform the data audits and provide
feedback and recommendations.
Intellectual Capital
Is there a repository, such as a knowledge management tool,
that is being used to store the organization's intellectual
capital? Is it well organized? Is it useful and is it being
Data Strategy
Does the data strategy support the strategic goals of the
organization and does it support the information strategy as
defined by the CDO? Does executive management (CEO, COO, and CFO)
buy into the strategy? Is there general acceptance of the data
strategy, or is there resistance to it?
Enterprise Organization
Has data ownership been established and accepted by the
appropriate LOB managers? Are the data owners fulfilling their
responsibilities? Has data stewardship been established? Are the
data stewards properly assigned? Are the stewards trained, and
are they doing what needs to be done? As for information
architects, data analysts, business modelers, metadata
administrators, and EIM staff, have they been properly recruited,
assigned, and trained? Are they being monitored for
effectiveness?
The CDO will want to include all of the above in a performance
plan that should highlight success and identify areas where more
work and attention is called for.
POLITICAL CHALLENGES
It might be hard to argue with the need for a CDO or for some
of the CDO roles and responsibilities, but there is bound to be
push back from some of the executives who might see the
introduction of a CDO as a threat, especially some managers in
IT. A CDO threatens existing empires, and the castles that
protect these empires do not easily surrender and are not easily
co-opted into cooperation. Let's discuss some reasons why the
introduction of a CDO position is bound to cause some political
Exposes Data Problems
A primary role of the CDO is to improve the quality of the
data, which includes exposing current data quality problems. The
problems may have been caused by inadequate editing rules where
invalid data would have been caught in data entry. The
departments of data entry are often measured by speed and not by
the quality of the work. This means that data entry clerks are
incented to ignore invalid data and just complete the work as
quickly as possible. It's not just the quantity of records that
there may be some incentives to complete the work
quickly so that operational and analytic work can begin on the
new data. The problems may also have been introduced as data was
brought from outside sources and not properly validated.
Validation is expensive and usually labor-intensive. This might
mean the manager responsible for data entry is open to criticism
as is the manager responsible for the edit checks as well as the
manager responsible for bringing in outside data. These managers
all have performance plans, and the exposure of bad quality data
is not in their economic or career best interests.
Exposes Holes That Should Have Been Addressed by IT
Management and Were Not
Besides the data problems that are likely to be exposed, the
activities of the CDO will make it apparent that the organization
may lack some basic IT policies and procedures that should have
been in place. Take, for example, a security and privacy policy
that would restrict access to employee salary information and
employee disability classifications. Another policy would limit
access to a publicly traded corporation's profit information
until the information became public. Even if it had not
explicitly been pointed out, senior management could not help but
conclude that IT management should have been doing a much better
Requires Authorization from Yet Another Place
In addition to authorizing budget and resources, managers want
the authority and ability to make all the decisions that cover
their areas of responsibility. They hate the idea that someone
else in the organization will be second-guessing their decisions
and keeping them from doing what they feel is their
responsibility. They also hate the idea that they are losing
control over what is rightfully theirs. Every time a manager
makes a request of another manager, something is being given up.
There is an implicit understanding that the requesting manager
has to be ready to support the person who honored the original
request. With every request comes an obligation -- something is
owed to, in this case, the CDO. Managers hate obligations that
must, at some time, be paid back.
Besides everything else, communicating, coordinating, and
getting authorization slows development and, in most
organizations, speed is among the most important determinants of
successful advancement.
Conveys Loss of Authority and Responsibility
While the CDO will have some roles not currently assigned to
anyone, there will be some roles that are currently the
responsibility of an existing manager. For example, the CIO now
has authority and responsibility over the enterprise architects.
We propose that the CDO will own enterprise architecture. With
the loss of responsibility, managers will be losing power,
prestige, people resources, budgets and the ability to
demonstrate their value, and probably the loss of future
promotions.
Threatens to Reduce Budgets
All executives and senior managers are aware of the total
budget available to them. They will be concerned that their
budgets are likely to be reduced, since the activities of the CDO
will cost money. If the CDO did not exist, those managers would
have more money to work with and would feel more comfortable
accomplishing what is in their performance plan. The excuse that
they were not given the budget they needed to do the job is
unpersuasive, and a smart manager would never use that excuse.
However, they know that stripping part of their budget will
affect them.
Potentially Poaches the Best Resources
The CDO will be recruiting, and a smart CDO will look for the
best data modelers, data administrators, database architects,
data analysts, SMEs, and so on. The CDO will be looking outside
the organization but will probably also try to poach the best
resources from the existing organization. The good people are
likely to be lured away with promotions, titles, more meaningful
work, and the opportunity to work in an exciting environment with
a terrific team. In addition, the people the CDO will bring in
from the outside are likely to be competitors for choice
positions and advancement.
Underscores Performance Acceptance
Some managers may think that they are already performing these
activities and that the company is doing OK now. This view is a
bit naive as it is quite evident how many pieces of the data
environment are being inadequately handled. If companies were
already performing these activities, they would not be mired in
their current data chaos. There may be a question of why the role
should be at the C-level when it could be performed by someone
two or three levels below. The answer is primarily authority that
would be unquestioned at the C-level. However, the danger exists
that the position could be stonewalled or rolled over at a lower
level in the organization, in which case it will be less
effective.
Countering the Challenges
First of all, it is important to recognize the political
impact of a CDO introduction and not pretend that politics are
unimportant. Second, there must be strong political support for
the role and authority of the CDO. This support should come from
one or more C-level officers, ideally from the CEO, the COO, the
CFO, and the CIO. The CIO should be able to see the introduction
of a CDO as beneficial to the CIO's role and as something that
will take pressure off the day-to-day CIO responsibilities.
The CDO must be seen as a permanent role, not just "let's see
if this works," and not as a position that will be terminated
once political pressure builds up against him or her. The
supporting C-level officers need to communicate their level of
support and their reasons for the support to the entire
organization. It may be helpful to seek advice from an outside
consultant who can provide the positioning, talking points, and
presentation for the C-level officers who would be the CDO
CONCLUSION
Organizations have struggled for decades with the value of
their data assets. In 1980, the first Data Management Association
(DAMA) group was formed in Los Angeles. Its members were mostly
data administrators and enterprise data modelers who completely
understood the value of data but were unsuccessful in convincing
and engaging the businesspeople in their data management
Until recently, managing data has been seen as an IT
responsibility. However, more and more companies are finally
realizing that managing data assets is a business responsibility.
This is evidenced by the data governance programs initiated in
many companies. Unfortunately, the organizations that have
implemented a data governance program have done so for only a
subset of the activities necessary to properly control their data
assets. This is evidenced by the fact that too many companies no
longer have the necessary EIM professionals to help their data
stewards with data governance activities. In addition, these
activities are not always coordinated or supported at the highest
To achieve maximum benefit, all enterprise-class activities,
including data warehousing, business intelligence, master data
management, customer relationship management, data governance,
data quality improvement initiatives, enterprise architecture,
and so on, should be lead by a new chief officer whose primary
responsibility is the standardization and management of data
assets in the organization. This new position is the chief data
1 Adelman, Sid, Larissa T. Moss, and Majid Abai.
Data Strategy. Addison-Wesley Professional, 2005.
2 Adelman et al. See 1.
3 Anne Marie Smith, PhD, is a highly acclaimed data
management professional, consultant, author, and speaker in the
fields of enterprise information management strategy, data
stewardship and governance, data warehousing, data modeling,
project management, and metadata management. She holds a
doctorate in management information systems and is a certified
data management professional (CDMP).
4 McGilvray, Danette. Executing Data Quality
Projects: Ten Steps to Quality Data and Trusted Information.
Morgan Kaufmann, 2008.
5 English, Larry P. Improving Data Warehouse
and Business Information Quality: Methods for Reducing Costs and
Increasing Profits. Wiley, 1999.
6 Moss, Larissa T., and Shaku Atre. Business
Intelligence Roadmap: The Complete Project Lifecycle for
Decision-Support Applications. Addison-Wesley Professional,
7 The Data Warehousing Institute. "TDWI Business
Intelligence Executive Briefing." January 2006.
8 Kimball, Ralph. The Data Warehouse Lifecycle
Toolkit. Wiley, 1998.
9 Russom, Phillip. "." TDWI, 23 October 2006
(http://tdwi.org/articles//master-data-management-consensusdriven-data-definitions-for-crossapplication-consistency-report-exce.aspx).
10 Loshin, David. Master Data Management.
Morgan Kauffman, 2008.
11 Clark, Jack. "." ZDNet, 7 December 2012
(/big-data-in-the-cloud-be-careful-what-you-pay-for-).
12 Eckerson, Wayne W. Performance Dashboards:
Measuring, Monitoring, and Managing Your Business. Wiley,
ABOUT THE AUTHORS
Sid Adelman is a Principal in Sid Adelman
& Associates, an organization specializing in developing data
strategies and data governance. Mr. Adelman's IT career spans
over 40 years. He has been involved with business intelligence
and data warehousing for the last two decades. Mr. Adelman has
published hundreds of articles, white papers, and columns. He
coauthored the books Data Warehouse Project Management,
Impossible Data Warehouse Situations and Solutions from the
Experts, and Data Strategy. Mr. Adelman was a
consulting systems engineer at IBM for 24 years, working as an
IMS and then a DB2 specialist. He holds an MBA in business
economics from UCLA and is a member of the IBM Gold Group and
Teradata 3rd Party Influencers. Most recently, Mr. Adelman
focuses on the wise and unwise data management decisions made by
individuals and organizations. He can be reached at
The Role of Chief Data Officer in the 21st Century
Did you enjoy this article?
Talk to Cutter today about membership, including one-on-one guidance from and interaction with Cutter's experts and with their peers, access to research, webinars, podcasts, white papers and more.
Become a Member
Research and inquiry privileges, plus regular strategy meetings with Cutter's Data Insight & Social BI experts are just some of the perks!
Want to learn more? Talk to Cutter today about trial membership, including access to research, webinars, podcasts, white papers and more.
Featured Webinar: Leveraging BI for Competitive Advantage
In this on-demand webinar, Nancy Williams reveals how successful business intelligence leaders leverage data assets for performance improvement and competitive advantage.