Every Chief Data Officer I meet has a view on where their office should really sit — especially now AI is in the picture. Far fewer have worked out the right answer.
Most government data leaders I speak to are somewhere between curious and certain about the right position for their team within their organisation. If the data office reported somewhere else – directly to the Perm Sec, say, or at least out from under Digital – would they finally have the clout to make change happen. So the question they bring me, more than almost any other, is this: where should the Chief Data Office really be operating from? My answer tends to be more challenging than they would like – there’s no right answer here.
The instinct behind the question is understandable: it doesn’t feel right where I am – so we must look across government, find whoever has is best, and copy them. But that may not be the right answer. The truth is, there is no single model that works everywhere. The best reporting line depends on the role of the organisation, its data maturity, its current pain points and the kind of change the CDO is being asked to lead.
At present, the government picture splits roughly three ways. Around a third of chief data offices sit within Digital. Another third sit under the Chief Analyst or Chief Scientist, alongside the analytical professions. The rest sit somewhere else entirely, often reporting directly to the Chief Operating Officer, the CFO, or elsewhere within the ‘corporate’ sphere.
It is not an unreasonable question to ask. The CDO role is still relatively new in government and has not yet settled into the executive architecture in the way finance, HR or digital roles have. The role often sits at deputy director level, expected to drive organisation-wide transformation without the authority, budget or access to do so. That tension is also not unique to government. In the private sector too, the role has evolved through different phases as organisations have worked out whether data is primarily a technology issue, an analytics issue, a governance issue or a business transformation issue. The trouble is – it’s all of the above.
So when I’m asked the question, my honest answer is: it depends. Each reporting line gives you something and takes something away. The right question is less “where should the CDO sit?” and more “what mandate does the CDO need at this point in the department’s maturity?”
Put the data office inside Digital and it can get close to the technology: platforms, systems, architecture, standards, the plumbing. That can be powerful. Data platforms and tooling are more likely to be treated as part of the wider digital estate, designed properly from the outset rather than bolted on afterwards. The data engineers and architects often feel more at home there too. But data can also become the poorer cousin to digital delivery. Worse, the rest of the business can start to see data as someone else’s problem, rather than something they own, steward and use every day. I know one Whitehall department that celebrated moving out from under Digital because it gave data more space to breathe, and another that celebrated moving in because it meant the technology it needed could finally move faster. Both were right in their own context.
A data office sitting under the Chief Analyst or Chief Scientist may benefit from a committed senior sponsor who already understands the value of good data. Analysts are often among the biggest and most vocal clients of the data office, so the relationship can be productive and energising. But there is a risk here too. Data can become framed too narrowly as something that exists for analysis, evidence and reporting, rather than as an operational asset that shapes services, decisions and delivery. I have seen an organisation deliberately move its data team out from under the analytical function for exactly this reason: the problem it needed to solve was no longer just better analysis, but better data ownership across the whole department.
Situate the data office under the Chief Operating Officer and it can connect more directly to transformation, performance and the running of the organisation. That can be particularly helpful where the COO understands that data change is cultural and operational, not just technical. It can bring data ownership closer to the parts of the organisation that create and rely on the data every day. But it can also mean data is not recognised as function which needs skilled technical capability, and longer-term capability building can lose out to immediate operational pressures.
Lately a new question has been layered on top: who takes responsibility for AI? Some public sector organisations are standing up a separate Chief AI Officer. A few are folding AI into the CDO’s remit; others give it a home in transformation, or none at all. Which of those is right to drive progress, again, depends on maturity. Where data still needs sustained investment and senior attention, hitching it to the AI agenda can create urgency and sponsorship that were hard to find before.
But it is worth being clear about what AI does and does not do. AI is a technology tool. Data is the fuel that AI runs on. No amount of AI ambition removes the need for someone to own the quality, governance and strategy of that data; if anything, it raises the stakes. And as with all technology tools – the real transformation with AI will come from changing processes, incentives, ownership and organisational habits, and that is a different job again from stewarding the department’s data. Neither of those jobs is solved by a new title. Whether they sit with one leader or two, they still have to be done, and done well.
So my advice to existing CDOs is this: do not spend too long waiting for the perfect box on the organisation chart. It probably does not exist. Wherever you sit, the work is to build mandate and sponsorship deliberately. Educate senior leaders on what data leadership is for. Demonstrate value in terms they care about. Make the connection between better data, driving AI, and the outcomes the department is already trying to deliver. Build alliances across Digital, Analysis, Operations, Finance, Policy and the frontline. The reporting line matters, but it will never do the whole job for you.