Mathew McKie

Why analytics belongs with strategy, and data engineering belongs with technology.

There must be value in all that data we’re collecting. Anyone know where?

Hire a Chief Data Officer.

We need to be doing something with AI. Anyone know what?

Hire a Chief AI Officer.

It’s an understandable response. Give an important problem a senior owner, establish a function and expect progress to follow.

Because analytics uses data, we often put it in the same function as data engineering. It sounds sensible. Both disciplines work with data.

But they exist for different reasons. Data engineering makes information available and usable. Analytics helps us understand what is happening and decide what to do about it.

Both are essential. But should they necessarily sit under the same executive?

The result can be a strategic mandate deficit: organisations have people responsible for developing strategy and people capable of rigorously investigating business decisions, but no one explicitly accountable for bringing those capabilities together.

I would argue that, in many organisations, the more logical arrangement is to combine corporate strategy with enterprise analytics, and technology leadership with data engineering.

That might mean turning the Chief Strategy Officer into a Chief Strategy & Analytics Officer (CSAO), and the Chief Technology Officer into a Chief Technology & Data Officer (CTDO).

The value was in the decision

At Capital One, one of the more consequential decisions I helped lead was the closure of Direct Mail as a channel for acquiring new customers in the UK.

Direct Mail was an established part of the acquisition business, supported by substantial infrastructure and highly visible as part of the brand. Questioning its future meant challenging something that had been central to how the business acquired customers.

We had sophisticated analytical methods for evaluating acquisition economics. We could assess what it cost to acquire different types of customers against the value they were expected to generate over time.

Behind those assessments were net present value calculations, customer segmentation, propensity models and detailed acquisition economics. The models incorporated the costs of operating the channel, including the expensive infrastructure required to purchase and maintain prospecting universes. That infrastructure was also coming under increasing pressure from marketing opt-out registers.

The economics had become marginal.

But identifying that conclusion wasn’t sufficient to close the channel. Direct Mail was too significant to the business and too visible in the brand for the decision to be made in isolation.

I worked with senior colleagues, including brand leadership, to examine the evidence, explain the commercial implications and build agreement around a different direction.

The experience has stayed with me because the analysis was only one part of the work.

We needed to understand the commercial problem, investigate the economics, determine what the evidence meant for the business and coordinate a consequential decision.

A model could identify the economics. A data team could provide the information. But neither would, by itself, change the organisation’s direction.

The value was in the decision we made using that evidence.

The specialism organisations misunderstand

Data engineering and analytics are different professional disciplines.

Data engineering is concerned with collecting, integrating, transforming and reliably making data available. It requires expertise in architecture, databases, software engineering, integration and technical controls.

Analytics starts with a different question: what are we trying to understand, and what decision might we make differently as a result?

That may require statistical inference, experimentation, forecasting, optimisation, financial modelling or machine learning. But analytical methods alone are insufficient. Effective analytics requires commercial understanding, judgement about uncertainty and the ability to connect findings to decisions.

Professional frameworks such as SFIA recognise data engineering and analytics as distinct disciplines. Thomas Davenport and Jeanne Harris, in Competing on Analytics, caution against treating analytics as synonymous with technology, emphasising organisational and human capabilities alongside technical ones.

Yet these disciplines are frequently bundled into a single data function.

Consider two specialists.

A talented strategy consultant can frame an ambiguous problem, develop options, construct a persuasive argument and negotiate agreement across an executive team. But they may lack the analytical expertise to establish whether their conclusions are supported by evidence.

An exceptional data scientist may be able to investigate the problem, test competing explanations and distinguish reliable conclusions from misleading patterns. But they may struggle to frame the wider strategic question or translate findings into organisational action.

Both are valuable. Neither automatically possesses the other’s capabilities.

The opportunity is to bring these disciplines together under one formal leadership mandate.

At Capital One, I experienced an organisation where analytical reasoning was deeply embedded in commercial management. Analytically trained leaders could hold direct responsibility for budgets and business outcomes.

Later, at VistaPrint, I held a combined Head of Strategy & Analytics role for the European business.

I coordinated strategic planning while leading analytical work across pricing, experimentation and commercial performance. Strategy and analytics were not separate activities in which one function developed a plan and another supplied evidence to support it.

Analytical expertise was part of developing and challenging the strategy itself.

These were different organisational arrangements, but they shared an important characteristic: analytical capability was connected directly to business leadership.

That is the principle I would seek to institutionalise.

Why a Chief Strategy & Analytics Officer?

A Chief Analytics Officer can lead an excellent analytical function, establishing professional standards, developing models and providing valuable insight across the business.

But leadership of analytical capability is not necessarily the same as leadership of the process through which corporate strategy is developed.

Similarly, a Chief Strategy Officer may be highly accomplished at developing business strategy and coordinating executive priorities without possessing the analytical expertise required to rigorously investigate the choices under consideration.

My preferred arrangement is to combine those responsibilities.

A CSAO would have a formal mandate to coordinate the corporate strategy process and ensure that analytical investigation is integral to it.

I would expect that mandate to cover framing strategic questions, evaluating competing options, coordinating analytical investigations, establishing experimentation and analytical standards, and assessing whether major decisions produce their intended outcomes.

The difference from an analytics centre of excellence is important.

An analytical team can be exceptionally good at answering questions while having relatively little influence over which questions the organisation chooses to ask.

The CSAO combines leadership of analytical capability with responsibility for coordinating the strategic process in which it is applied.

This does not mean moving every analyst into a central strategy department. Commercial, financial and operational teams often benefit from having analytical specialists embedded close to their decisions. Enterprise analytical leadership can provide common methods, professional development and coordination of strategically important work while allowing much of that capability to remain distributed.

Nor must the CSAO personally be the organisation’s most accomplished strategist and its best practising data scientist. The requirement is to understand both disciplines sufficiently to lead specialists, challenge their work and bring their contributions together.

Why not the CFO?

Finance is an entirely credible alternative.

A commercially sophisticated CFO may already combine investment evaluation, analytical expertise and considerable strategic influence. In some organisations, Finance may be the most effective home for enterprise analytics.

My preference for the CSAO reflects the breadth of analytical investigation.

Enterprise analytics is not limited to financial forecasting, performance reporting or investment scrutiny. It includes understanding customer behaviour, investigating operational problems, designing experiments, evaluating interventions and discovering strategic opportunities that may not yet feature in the corporate plan.

The objective is not simply to evaluate choices already presented to the executive team. It is also to help determine which choices deserve consideration.

Where a CFO already holds that wider strategy mandate, the principle may be satisfied without a separate CSAO. Similarly, a CAO who genuinely coordinates strategic investigation and participates in framing corporate strategy may effectively be performing the combined role.

The central requirement is the formal integration of strategic leadership and analytical capability, not the creation of a particular job title.

The CEO and executive committee remain accountable for corporate strategy. Finance retains financial stewardship, and business leaders remain accountable for their commercial and operational outcomes.

The CSAO provides leadership of the process connecting strategic questions, rigorous evidence and executive decision-making.

From business strategy to AI requirements

The current enthusiasm for AI makes this distinction particularly important.

Organisations often begin by declaring that they need to do something with AI, then search for opportunities to apply it.

I would reverse that process.

Start with the business objective. Investigate the underlying problem, establish what outcome needs to change, evaluate possible interventions and determine what improvement would justify investment.

Consider customer retention.

A technology-led approach might begin by building a model to predict which customers are most likely to leave.

That could be a technically excellent model. But it doesn’t necessarily tell the business which customers it should try to retain.

A customer who is highly likely to leave may be almost impossible to persuade to stay. Another customer, with a lower predicted probability of leaving, may be far more responsive to an intervention.

Predicting what someone will do is not the same as establishing whether we can change what they do.

The second question requires different analytical thinking.

We need to establish the causal effect of an intervention, potentially through controlled experimentation. We need to understand which customers respond, what the intervention costs and whether the resulting improvement creates sufficient economic value.

The best answer might be a targeted retention intervention informed by experimentation, rather than the most accurate churn prediction model.

And it might not require AI at all.

This is what I mean when I suggest that analytics can become an advanced form of requirements definition.

Traditional requirements often describe what a system should do. Analytical investigation can establish what business problem actually needs solving, which interventions are likely to improve it, what improvement would be worth and how success should be measured.

That provides a stronger foundation for technology investment.

The CSAO would coordinate this investigation with the executives accountable for the business outcome, helping turn strategic priorities into evidence-based opportunities and requirements.

Technology specialists would contribute throughout, assessing feasibility, cost, risks and potential solutions. Technical discovery may change which intervention is worthwhile, just as experimentation may change our understanding of the problem.

But the starting point is a business objective and an evidence-based reason to invest, not an instruction to find something to do with AI.

The complementary technology mandate

The same reasoning leads me to place data engineering alongside enterprise technology leadership.

Data infrastructure depends on source systems, architecture, software engineering, integration, infrastructure and security. These disciplines should be developed as part of a coherent enterprise technology capability.

I would therefore give the CTO explicit accountability for technical data engineering, architecture, integration and data platforms. Where useful, that combined remit could be recognised through the title Chief Technology & Data Officer.

Specialist technical data leaders would remain important, as would multidisciplinary teams containing analysts, data scientists, analytics engineers, ML engineers, software developers and business specialists.

Strategic analytical direction belongs with the CSAO. Technical engineering, deployment and operation belong with the CTDO. But effective delivery requires those disciplines to work together continuously.

The distinction concerns executive accountability, not separating people who need to collaborate.

It also clarifies AI leadership. Strategy & Analytics coordinates the assessment of opportunities and expected business value. Technology & Data leads technical delivery. Business executives retain accountability for the decisions and outcomes within their domains.

Information governance requires deliberate responsibility too. My default would be for technology leadership to own technical data-management standards and controls, while business leaders own the meaning, required quality and appropriate use of their information. Risk, legal, privacy and assurance functions retain their respective responsibilities.

A dedicated CDO or CAIO may still be appropriate where the scale, complexity or regulatory demands justify one. But those appointments should address a specific need rather than substitute for clear strategic and technical accountability.

Where this could go wrong

The strongest case for a CSAO arises where organisations make consequential strategic decisions under uncertainty but lack a disciplined process for connecting analytical investigation to those decisions.

The warning signs are familiar: strategy developed primarily through planning exercises, analytics commissioned after priorities have been agreed, business cases built on assumptions that are rarely tested, and analytical teams with limited influence over investment choices.

CTDO leadership becomes particularly valuable where technical data capabilities are fragmented, enterprise architecture lacks direction or operational IT responsibilities have crowded out longer-term technology planning.

Neither title is inherently necessary. What matters is whether the organisation already possesses the relevant capabilities and authority.

The model also introduces risks.

A strategy executive without analytical credibility will not become an effective CSAO through a new title. An annual planning function could isolate analytics from continuous experimentation. Centralising every analyst could weaken the relationship between analytical specialists and the business decisions they support.

Likewise, separating executive accountability without enabling close collaboration between strategy and engineering could create additional organisational barriers.

These are reasons to be precise about responsibilities and operating arrangements.

I would judge the proposed model by whether it improves strategic decisions, strengthens investment outcomes and enables the organisation to recognise and correct mistaken assumptions more quickly.

Where another arrangement achieves those outcomes more effectively, it should be preferred.

Before hiring another chief

For years, organisations have debated how Chief Data Officers can become more strategic. We ask them to demonstrate business value, develop stronger commercial skills and secure greater influence over executive decisions.

Perhaps the more useful question is whether responsibility for analytical business leadership has been placed in the right part of the organisation to begin with.

The distinction matters even more as AI makes increasingly sophisticated technological capabilities available.

We can build excellent models, platforms and agents without establishing whether they solve the problems that matter most to the business.

The opportunity is to connect the people who frame strategic questions, those who rigorously investigate them and those who engineer the capabilities needed to act on the answers.

So before appointing another Chief Data Officer or Chief AI Officer, I would encourage CEOs to ask two questions.

Who is accountable for turning evidence into better strategic decisions?

And who is accountable for engineering the technological capabilities those decisions require?

If the answers are unclear—or both lead to a function principally designed to manage data technology—I would question whether the organisation has designed its leadership around the value it actually wants to create.