Since the first chief data officer was appointed at Capital One in 2002, the role has been plagued by confusion about its purpose. Although surveys of large organizations by Randy’s firm NewVantage Partners show an overall increase in the prevalence of the role — climbing from 12% in 2012 to 68% in 2018, and falling somewhat in 2019 — CDOs’ responsibilities have remained unclear. In the most recent survey, only 28% of respondents agreed that the role was “successful and established.”

Early on, when most CDOs were in large financial institutions, their most common roles were defense-oriented, involving the security, privacy, quality, and regulatory compliance of key data. However, these functions proved difficult to evaluate except in their absence or breach, and companies have increasingly desired offense-oriented benefits from data — better decision-making, marketing, customer service, and monetization.

Today, CDOs typically have multiple objectives, but often there is little consensus about which are most important. In addition, achieving these different objectives may require substantially different types of experience and skills. Many CDOs will find it difficult to excel at everything they’re expected to do.

The most recent NewVantage survey illustrates this problem. It asks respondents — more than half of whom are CDOs themselves — what the primary attribute of a successful CDO is within their organization. Some say their organizations desire an external change agent (the most common and fastest-growing response), others a company veteran or insider. Some want a line-of-business executive who owns business results, others want a technology executive, and still others want the attributes of a data scientist or analytics leader. You’d think at least there would be agreement that the CDO is primarily responsible for data — but several years of surveys find that that’s true in only about 40% of organizations.

In part because of this lack of consensus, we have observed short CDO tenures — typically two to three years in many firms, particularly in industries like financial services, where data is the organization’s lifeblood. Lack of role clarity and unachievable expectations are certainly a factor in short job tenures.

Seven Distinct Jobs

Understanding what a given organization expects of its CDO, and then matching capabilities with those expectations, can be the key to longer, more effective, and happier CDO tenures.  In studying CDOs and their roles over the past decade, we’ve identified seven key types of CDO jobs, each distinct enough that it would be difficult or impossible for one person to perform all of them well. New CDO hires may not know which combination of the seven they are expected to perform, and formal performance evaluations may not make clear on which of the seven an incumbent CDO is being assessed. One CDO, after reviewing the seven job types, said in an email, “I am evaluated on all seven, but the only ones I get any credit for are the offense-oriented ones.”

The first three roles described below are primarily offense-oriented; the last four are more defense-related. Of course, some CDO jobs involve combinations of these roles.

1. The Chief Data and Analytics Officer

CDOs in this role manage the data science, analytics and, sometimes, the artificial intelligence functions for the enterprise. Companies across industries including General Motors, Walmart, Chase Bank, Partners Healthcare, and MetLife have CDOs with this job responsibility, a role that’s becoming increasingly common.

Oversight of data management, data science, and analytics in one job is a reasonable combination, since data is the key ingredient for successful analytics and AI. And analytics allows organizations to use data offensively; analytical models can help target customers, optimize supply chains, and make better human resource decisions, all of which allow CDOs to create value. One CDO who controlled analytics and other functions told us, “Some other aspects of the CDO job were impossible to make visible progress on. Thank God I had analytics to fall back on.”

Incumbents of this type need experience applying math, analytics and AI techniques to business problems. They should also be familiar with the types of information technology environments that support analytics and AI, such as open-source tools like Hadoop for storing big data. Particularly with the addition of AI, data science is an increasingly complex and dynamic area of the business, and CDOs focused on this role often have little bandwidth for additional responsibilities.

2. Data Entrepreneur

An increasingly common objective for CDOs is to monetize data — either by selling it directly, using it in data- and analytics-based products and services, or spinning out new businesses based on it. Auto companies, for example, are interested in monetizing connected vehicle data; General Motors has launched a substantial initiative in that area. Software companies are intently focusing on capturing and monetizing data from the business processes that their software supports. (See this HBR article for more.) Online businesses already monetize their data by selling it for the purpose of targeted advertising. Between 11% and 13% of companies we surveyed report that their CDO has some revenue responsibility.

But despite some success and the popularity of data monetization in the abstract, it can be challenging to do. Companies that sell tangible products find it difficult to switch their orientation to selling data. And both consumers and businesses are increasingly concerned about the ownership and use of data that they help to generate. CDOs in the entrepreneur role find that these issues become part of their purview.

While we are certainly supportive of CDOs taking on the development of innovative products and services related to data, we expect that for most this will be a second-order rather than primary focus. In any case, doing this job well requires the CDO to have a background in building businesses, monetizing assets, and developing new products and services. Such attributes are not often found in combination with some of the other skills needed by CDOs.

3. Data Developer

Given the need for CDOs to generate tangible value, some have taken to leading the development of key applications or infrastructural capabilities that will help them deliver it. Many of these, for example, have overseen the creation of enterprise data warehouses or data lakes for their firms. For those CDOs with an analytical or AI orientation, the application might reflect that focus (an AI-supported loan-approval process would be an example). For others, it may be more transactional in nature. At Citizens Bank, for example, the chief data officer (Ursula Cottone, now in the CDO job at Huntington Bank) led an implementation of process automation software to streamline the back office.

While application development is traditionally more the responsibility of the chief information officer, some development projects may be particularly data-intensive and are thus appropriately a CDO role. At Morgan Stanley’s Wealth Management unit, for example, Jeff McMillan, the chief data and analytics officer, led a project to build a “next best action” system, an investment-recommendation tool for clients.

Developing systems of any type is a demanding role that requires expertise in various information technologies, agile and other development methods, working effectively with stakeholders, and integration with systems and processes.

4. Data Defender

Keeping data safe from breaches, fraud, and attacks, and persuading regulators that the firm was adhering to data safety rules, were the most common reasons CDO roles were created in financial services firms in the 1990s and 2000s. This defensive role has declined for CDOs over time — not because data safety became less important, but because other roles, like the chief information security officer (CISO), were created to address it. Still, in our most recent survey, 45% of data executives said that the primary role of the CDO is defensive.

There are still strong pressures toward data defense, particularly in financial services. One banking CDO told us, “I try to keep the balance of offense and defense at 50/50, but all the pressures are toward more defensive activity.” Auditors and boards of directors often push for defensive CDO roles. However, we’d argue that defending data is a full-time leadership role best left to CISOs. If the CDO focuses on it, he or she isn’t likely to deliver much competitive advantage or business benefits from data.

5. Data Architect

There is a long tradition of using architectural or engineering-oriented methods to create better data environments — that is, with less duplication and fragmentation — and some CDOs lead this activity for their companies. Today, these efforts largely focus on assuring that key enterprise data is aggregated, cleansed, consistently formatted, and readily available throughout the organization. However, these programs tend to be expensive and time-consuming, and many business executives don’t see sufficient value in them. When they are cancelled, CDO tenures may be ended along with them.

Some firms ask their CDOs to architect substantial data-modernization initiatives. These typically involve a move from enterprise data warehouses to data lakes, and may also include the use of AI-based approaches to data integration. We wrote about such an effort at GlaxoSmithKline’s R&D unit, which was led by Mark Ramsey, then the CDO. Still, only 6% of respondents in our latest survey say they have a “modern data architecture” — most have a mix of legacy and new systems and data formats.

6. Data Governor

Another key CDO role is to establish “data governance” programs. These typically involve enlisting middle and senior-level business managers to take responsibility for data domains that relate to their business processes, for example tasking the CMO with oversight of all marketing data, and supporting them in the effort. When it works, it can be very effective at aligning data programs to business strategies and goals. However, it is often difficult to persuade business-side managers to spend time and attention on data management issues. One CDO told us, “I still try to get the business side involved, but I don’t use the word ‘governance’ — it’s become somewhat toxic.”

CDOs focused on governance need the ability to build relationships and establish trust. They also must understand the relationship between the business’s data needs and business strategy. The combination of data orientation and organizational/business change is often difficult to find in a single person. Organizations whose CDOs have a strong focus on governance include the financial services firms Citigroup, Charles Schwab, UBS, and PNC.

7. Data Ethicist

Perhaps the least common focus of the CDO job, but one that is growing in popularity, is on the ethics of data management, specifically on how it’s collected, safeguarded and shared and who controls it. There is no doubt that consumers, regulators, and legislators are becoming more concerned about the misuse of data. If the CDO is charged with managing data, it is logical that the role should include a focus on ethics. Some companies, primarily IT vendors and a few financial services firms, are creating chief ethics officer roles that can absorb this CDO job.

One prominent example of an ethics-focused CDO is at Mastercard, where JoAnn Stonier holds the job. She was previously chief privacy officer there and at American Express, and data privacy and ethics are the primary component of her job. Mastercard has developed a “Data Responsibility Imperative” and Stonier plans to implement its principles internally. They include security and privacy, transparency and control, accountability, integrity, innovation, and positive social impact. She has a law degree, which might be expected for the ethics focus (though they’re not common among CDOs). Although the ethics orientation is somewhat rare now, we expect that it will increase over time.

Too Many Roles for One CDO

Clearly it would be difficult for one person to perform all of these diverse roles effectively. They require different backgrounds and capabilities. That suggests, of course, that organizations need to choose which CDO jobs make most sense in their case. The choice also dictates the optimal reporting relationship for the CDO; a data architect might well report best into the chief information officer, for example, while more analytically focused roles might best report elsewhere, for instance to the CAO if there is a separate one. All of these jobs need to be done by someone, so if your CDO isn’t doing them, they should be given to some other role within the organization. And since there are many different information-related jobs within firms today — chief information officer, chief data officer, chief digital officer, etc. — clarity on who is supposed to do what is a necessity. We expect that the number of CDO jobs will continue to grow in organizations, but CDOs will only succeed if their roles are clearly specified.

• This article was originally published by Harvard Business Review on February 7, 2020.

Thomas H. Davenport is the President’s Distinguished Professor in Management and Information Technology at Babson College, a research fellow at the MIT Initiative on the Digital Economy, and a senior adviser at Deloitte Analytics. He is the author of over a dozen management books, most recently Only Humans Need Apply: Winners and Losers in the Age of Smart Machines and The AI Advantage.

Randy Bean is CEO and managing partner of consultancy NewVantage Partners.  You can follow him at @RandyBeanNVP.