Each week we find a new topic for our readers to learn about in our AI Education column.
Many businesses can be oaks. It’s much harder to be a sequoia.
If the modern business enterprise is a tree, then data is the sap that it uses to feed all of its constituent parts. Data has always mattered to financial institutions. What has changed is how much of the modern enterprise now depends on it—and how quickly bad data can propagate through automated systems.
If we extend our metaphor, when that tree starts to become exceptionally large, and it has to move more sap farther distances, the complexity and volume of the data may become a limit on growth. There’s nothing bigger than a sequoia.
It doesn’t have to be so in the business world. Enterprises are wonderfully elastic, though, because they can take action to alter or eliminate caps on their growth—we talk about this all the time in our AI work, because AI has the potential to make workers more efficient and productive, but there are other ways that businesses can expand capacity, like hiring more workers—which is the rough equivalent to our sequoias growing more limbs, leaves and roots to gather sunlight, water and nutrients.
But businesses can also do something that has no analogue in our forests—they can better manage the resources they have. Imagine of our sequoia could develop an intelligent system to manage and distribute sap—nutrients—in an optimal and coherent manner. Today, on AI Education, we’re going to discuss how businesses can manage their sap, the data, via a chief data officer.
What is a Chief Data Officer?
At its simplest, a chief data officer is the senior executive responsible for an organization’s data as an enterprise asset: how it is collected, defined, governed, maintained, shared, protected and ultimately converted into business value. IBM’s overview of the chief data officer role defines the CDO as an executive responsible for extracting maximum business value from enterprise data, typically through responsibilities encompassing data strategy, governance, quality, analytics and security.
That definition sounds straightforward. The actual job is anything but. A modern CDO sits at the intersection of business strategy, technology, risk, regulation and organizational culture. The executive may be responsible for establishing an enterprise data strategy, defining data ownership, creating governance standards, improving data quality, eliminating or connecting data silos, overseeing data architecture or analytics, establishing metadata and lineage practices, improving data literacy and making sure data can safely be used for artificial intelligence. Depending on the company, the CDO may also oversee data scientists, engineers, analysts and stewards, or share some of those responsibilities with technology and analytics executives.
The distinction between managing data and merely storing it is important. Businesses have spent decades collecting enormous quantities of information through transactions, websites, customer relationships, enterprise software, financial markets, communications and operational systems. Possessing that information does not automatically make it useful. Two departments can maintain different addresses for the same customer. Two systems can define “revenue” differently. Records can be incomplete, duplicated or stale. A company may not know who owns a particular dataset, who is authorized to access it or where a particular number originated. A CDO attempts to impose order on that environment.
More Than Data Management
One way to understand the CDO is to think of the position as having both a defensive and an offensive mission. The defensive job involves protecting the enterprise from the consequences of poorly managed data. That includes governance, quality, privacy, regulatory compliance, security coordination, lineage, retention policies and access controls. The offensive mission is to make data useful—to improve decisions, identify opportunities, automate processes, understand customers, develop products, improve efficiency and enable analytics and AI.
Those missions increasingly overlap. Good governance is not simply a compliance exercise if trustworthy data enables employees to make decisions faster. Improving customer records can simultaneously reduce operational mistakes, improve marketing, strengthen fraud detection and provide better inputs for machine-learning models. That is why the CDO increasingly looks less like the custodian of a giant corporate database and more like a business executive.
IBM describes data strategy as a plan for using information to improve decisions, optimize business processes and accomplish business objectives. It encompasses collection, management, governance, analytics, quality and security. A successful strategy can improve products and services, customer satisfaction and competitive positioning. The CDO’s task is therefore not simply to make data cleaner. It is to answer a more consequential question: What does the organization want to accomplish with its data, and what people, processes, technology and governance are necessary to make that possible?
That distinction also helps explain why a CDO cannot succeed from inside a technological silo. Data flows through practically every business function. Sales creates it. Marketing consumes it. Finance reconciles it. Compliance monitors it. Technology stores and transports it. Cybersecurity protects it. Operations depends on it. Executives use it to make decisions. The CDO has to operate horizontally across those organizational boundaries. Then came artificial intelligence
AI Raises the Stakes
Artificial intelligence depends upon data, and organizations attempting to deploy AI are discovering that acquiring a powerful model is only part of the problem. The model must interact with enterprise information that is accurate, relevant, accessible, properly permissioned and understandable.
The 2025 CDO Agenda research published by Amazon Web Services reported that 52% of organizations surveyed rated their data foundations as inadequate for generative AI implementation. Improving data quality and integration were among organizations’ leading priorities for generative AI. That relationship makes the CDO potentially more important in the AI era, but it does not necessarily mean every company deploying AI needs to create the position.
AI does, however, magnify the consequences of weak data management. If a human analyst receives contradictory information from several systems, the analyst may recognize the discrepancy and investigate. An automated AI workflow could potentially ingest that same information, process it at enormous scale and distribute an incorrect conclusion throughout other workflows. The old computing maxim “garbage in, garbage out” becomes more consequential when the garbage can be processed automatically thousands or millions of times.
AI also creates new questions about which enterprise information can be used for which purposes. Can confidential customer information be supplied to a generative AI application? Which datasets can be used for training or retrieval? Who owns AI-generated data? How should organizations document data provenance? What happens when models combine internal information with third-party data? Which employees and AI agents should have access to which records? Those are not exclusively technology questions. They are governance questions.
The relationship is already visible in organizational structures. Deloitte’s 2025 Federal CDO Survey found that AI use among surveyed federal organizations increased from 67% in 2024 to 78% in 2025. Thirty percent of CDOs surveyed also served as chief AI officers, 96% collaborated with AI leadership at least monthly and 64% were very or completely involved in setting data-governance policies for AI.
Some businesses consequently are expanding the CDO into a chief data and analytics officer, or CDAO, while others are adding AI to the title and mandate. Whether data and AI ultimately belong under one executive will depend upon the organization. But separating AI strategy entirely from data strategy becomes increasingly difficult when AI systems depend so heavily upon the availability and quality of enterprise information.
CDO, CIO, CTO, CISO—and CAIO
Part of the confusion surrounding chief data officers results from an increasingly crowded technology C-suite. The chief information officer traditionally owns enterprise information technology: applications, infrastructure, technology operations and the systems employees use to run the business. The chief technology officer often has a more product- or technology-development-oriented mandate, although the division between CIO and CTO varies considerably among organizations.
The chief information security officer owns cybersecurity. The CISO concentrates on protecting systems and information from unauthorized access, disruption and attack. The chief analytics officer concentrates on extracting insights from data. And the emerging chief AI officer typically focuses on artificial intelligence strategy, adoption, governance and execution.
The CDO overlaps with all of them without being identical to any of them. A useful shorthand is that the CIO owns much of the technology through which data travels, the CISO owns much of the security protecting it and the CDO owns the enterprise strategy for making that data trustworthy, usable and valuable. IBM similarly distinguishes the CIO as responsible for technology infrastructure that processes information and the CDO as responsible for how data is used to generate business value.
The boundaries are rarely that clean in practice. Security and data governance intersect. Data architecture overlaps with technology architecture. Analytics overlaps with AI. Privacy may sit under legal, compliance, security or data leadership. Indeed, organizations sometimes deliberately combine the positions. The Office of the Comptroller of the Currency currently combines its CIO and CDO responsibilities under one executive.
What matters more than the title is accountability. Someone needs sufficient authority to establish enterprise-wide data standards and enough organizational influence to enforce them. Creating a CDO position without defining where the CDO’s authority begins and ends merely introduces another executive into an already complicated governance structure.
Who Is a Good CDO?
The ideal chief data officer is neither simply a technologist nor simply a business executive. Technical literacy is essential. A CDO should understand data architecture, governance, quality, analytics, cloud platforms, databases, metadata, lineage, privacy and increasingly machine learning and generative AI. But the CDO does not necessarily need to be the company’s best data scientist or engineer.
The more difficult skills may be organizational. A CDO needs to translate between engineers and executives. The executive must convince business units to surrender some autonomy over data definitions and practices for the benefit of enterprise consistency. The CDO must negotiate with the CIO over infrastructure, with the CISO over security, with legal and compliance teams over regulatory requirements and with business leaders over priorities. That makes communication, diplomacy and change management unusually important.
Tamr’s discussion of the evolving CDO role describes the executive as strategist, evangelist, technologist, change agent and communicator. Its hiring guidance emphasizes that communication, collaboration and change-management capabilities can be as important as technical data and AI expertise.
A strong CDO should also understand the economics of the business. Data governance can become an endless internal project unless somebody continually connects it with outcomes. The objective should not be to create the world’s most elaborate data catalog. It should be to solve business problems: faster reporting, better underwriting, reduced fraud, more accurate customer records, improved regulatory reporting, lower operating costs, more reliable AI or new revenue-generating products. That requires a leader capable of speaking simultaneously about data architecture and return on investment.
Does every company need one? No. Every organization needs somebody accountable for important data. That does not mean every organization needs a full-time C-suite executive carrying the title chief data officer.
A 25-person professional-services business with a straightforward technology environment probably does not need the same data leadership structure as a multinational bank processing millions of customer relationships across hundreds of applications and jurisdictions. In a smaller organization, data responsibilities might reasonably reside with a CIO, CTO, head of data, director of analytics or another senior executive. This is why employee count or revenue alone is a poor hiring rule.
A 200-person AI company whose product depends entirely on enormous datasets may have a greater need for senior data leadership than a 2,000-person company with relatively simple information requirements. A rapidly growing fintech handling sensitive financial information may encounter sophisticated governance requirements long before it resembles a large bank by headcount. Instead, businesses should look for inflection points.
The case for a CDO strengthens when nobody can clearly identify who owns enterprise data; business units maintain conflicting definitions of critical information; executives no longer trust reporting; acquisitions have created incompatible data environments; privacy and regulatory obligations are becoming difficult to coordinate; data scientists spend disproportionate amounts of time finding and cleaning information; the organization is building data products; or AI has become strategically important but teams repeatedly discover that enterprise data is inaccessible, poorly documented or unreliable.
For smaller businesses reaching those points, an intermediate step may be a head of data, data-governance leader or fractional CDO rather than immediately creating another permanent C-suite position.
The threshold should therefore be expressed in terms of organizational complexity rather than a magic number of employees. A company should consider dedicated executive data leadership when the economic value and risk associated with its data become large enough that fragmented ownership costs more—or creates more risk—than centralized leadership.
Finance Is a Special Case
Banks, insurers, asset managers, broker-dealers, wealth managers and fintech companies are fundamentally information businesses. Account balances, transactions, securities prices, positions, credit histories, identities, communications, portfolio holdings and risk calculations are all data. A physical manufacturer can temporarily operate despite imperfections in a customer database. A financial institution cannot tolerate uncertainty about whether a customer’s account balance is correct.
Financial institutions also use data in decisions carrying unusually significant consequences: whether to extend credit, flag a transaction as fraudulent, execute a trade, calculate capital, detect money laundering, price insurance, allocate portfolios or provide financial advice. The regulatory environment reinforces that importance.
The Basel Committee on Banking Supervision reiterated in January 2026 that accurate, comprehensive and timely risk-data aggregation and reporting capabilities are critical for identifying and managing material risks that could create financial losses and affect institutional safety and soundness.
Meanwhile, revised model-risk-management guidance issued in April 2026 by the Federal Reserve, OCC and FDIC emphasizes governance, defined responsibilities, data quality and controls while tailoring expectations to the nature, scale and complexity of institutions and their model usage. The guidance is expected to be most relevant to banking organizations with more than $30 billion in assets, although regulators note that smaller institutions with significant model-risk exposure may also find it relevant. Importantly for the emerging AI governance discussion, the revised guidance says generative and agentic AI are outside its formal scope, while stressing that banks’ broader governance and risk-management practices should determine appropriate controls for those technologies. That $30 billion figure should not be interpreted as a CDO hiring threshold. It illustrates a more useful principle: governance should scale with complexity and risk.
A community bank with straightforward operations and limited analytics may be able to place data accountability within existing leadership. A larger regional bank integrating numerous systems, deploying machine learning, managing substantial regulatory reporting obligations and experimenting with generative AI has a much stronger argument for dedicated executive data leadership. At a global financial institution, allowing responsibility for enterprise data to remain indefinitely fragmented among technology, risk, compliance and individual business units becomes increasingly difficult to justify operationally. AI accelerates that progression.
A CDO As a Bridge to the AI Enterprise
The irony of artificial intelligence is that the more sophisticated models become, the more attention organizations may have to devote to something considerably less glamorous: their data.
Companies can purchase models, cloud infrastructure and AI applications relatively quickly. Creating consistent definitions across decades of legacy systems is harder. Establishing data ownership is harder. Resolving duplicated customer records is harder. Documenting lineage is harder. Changing organizational behavior is harder.
Those are precisely the problems the CDO was created to address.
The future of the position consequently may involve less distinction between “data strategy” and “AI strategy.” Organizations are already experimenting with titles such as chief data and analytics officer and chief data and AI officer because analytics and artificial intelligence increasingly represent the mechanisms through which companies extract value from data.
But the underlying question remains unchanged: Who is accountable for the organization’s data?
If the answer is “everyone,” the practical answer may be nobody.
For financial institutions especially, that ambiguity becomes increasingly difficult as more decisions migrate from spreadsheets and human workflows into models, copilots and autonomous agents. AI can accelerate analysis, automate processes and expand the amount of information an organization can use. It can also accelerate errors when the underlying data is incomplete, inconsistent, improperly governed or misunderstood.
A chief data officer cannot solve all of those problems alone. Nor does appointing one magically make an organization data-driven. The CDO still needs executive sponsorship, cooperation from business units, appropriate technology, effective governance and employees who understand their own responsibilities for data.
What the position can provide is something increasingly valuable in an AI-powered financial enterprise: a single senior executive charged with connecting data governance to data value.
That is ultimately the best test for whether an organization needs a CDO. The question is not how many employees it has. It is whether data has become sufficiently important, complicated and risky that it deserves an executive whose principal job is making sure the organization can trust it—and use it.
For a growing number of financial institutions deploying artificial intelligence, that threshold may already have been crossed.
