Why data stewards are the real heroes of AI

If your university is thinking about using AI tools, getting the basics of data stewardship right should be your first step. Paul Clark explains more.

Paul Clark is a strategy consultant - until recently he was Vice-President (Strategy) at University College London. He was formerly Chief Executive of HESA, and Director of Policy at Universities UK.

A few years ago – at what felt like the dawn of the data age – data scientists were labeled “unicorns” – in light of the vanishingly small chance of finding all the desired attributes in a single employee.

But as we stand on the brink of the age of artificial intelligence, it’s the humble data stewards who really hold the key to unlocking the benefits of AI for universities. University leaders who truly want to leverage the benefits of data and AI should focus their efforts on good data governance. This starts with the stewards.

As with many sectors, universities are currently at the foothills of exploring what advantages AI can bring. While the spotlight at the moment is on the impact on teaching and learning, assessment, and operational efficiency gains, the level of excitement around what AI might bring to university strategy and operations in the future is high. At a minimum, major advances are predicted in learner analytics, personalized student support, strategic analysis, dramatically accelerated research findings, and major operational efficiencies. But these benefits won’t be realized without a focus on good data governance.

The nirvana for data governance is to ensure there is a “single source of truth” that institutional leaders and decision-makers can draw on quickly when planning and implementing their strategies. But behind this simply-expressed aspiration lies a hinterland of connected activity.

The success of the iPod wasn’t down to the iPod

When Apple launched the iPod in 2001, the aim was to make music more accessible and portable, delivering the right song to the right person at the right time. But the handheld device wasn’t the real innovation – that came from the iTunes infrastructure that sat behind it. iTunes was itself a complex, inter-connected network of technology platforms, software, data rights management, copyright and contract agreements, enforcement policies, and market intelligence. Effectively coordinating these disparate components into the iTunes system was what really gave Apple its competitive advantage and made the iPod the market leading device for decades.

Similarly, the aim of a data strategy in a university should be to deliver the right data to the right people at the right time. But achieving this aspiration depends on the effective coordination of such disparate areas as IT infrastructure, software engineering, data management, user requirements, data protection and privacy policy, cyber security, and increased data capability. Managing this in turn depends on good data governance.

Put the people first

What does good data governance look like and how do you get there? As the title suggests, it starts with the data stewards – or, to put it another way, put the people first. The data stewards should work closely with designated data owners, who are managers with accountability for deciding how data is defined and managed in their domains, for example, HR colleagues who look after staff data, or research administration staff who look after research contract and performance data.

At the institutional level, data stewards and data owners should both feed into governance groups and senior-level bodies that are ultimately responsible for making organisation-wide decisions on data policy and management, and for ensuring that the right areas are prioritized.

Layered around this structure should be institutional processes and policies that facilitate efficient and effective data governance. These include data classifications and definitions, storage and integration tools, discovery tools, access and identity management, data protection and privacy policies, and cyber security infrastructure. At the most advanced level, a mature data governance system should also include a focus on automation, innovation, and plans to foster an organisation-wide culture of valuing high-quality data.

Principles for good data governance

Designing and implementing a data governance strategy will yield significant and sustainable benefits for institutions. This can seem overwhelming at first if there isn’t yet mature capability in this area. So where to begin?

You need to make data governance a priority at the senior leadership level. Leaders rightly want and expect data to provide quick and reliable answers when they need them, and AI holds the promise of vastly accelerating this. Achieving this requires a focus on data governance, championed from the top. This approach should include ensuring that the ‘people’ aspects of effective data usage are put first, and a commitment is made to fostering a long-term culture of valuing data as a strategic institutional asset.

There’s a need build a cross-functional group to lead the data governance effort. This should comprise membership from all the relevant areas, including users of the data output. The size and composition will vary from institution to institution, but as a minimum should include colleagues covering planning, IT, data protection, cyber security, compliance, relevant professional functions acting as data owners (e.g., student systems, Finance, HR, Estates), and representative data users.

It is hugely important to prioritise initial areas of focus according to business needs. It may be tempting to take on the whole data ecosystem at once to improve governance and output. However, this could be an overwhelming challenge that leads to loss of momentum and lack of impact. A better approach would be to pick off those issues that are uppermost in the minds of senior institutional leaders, and work to improve the governance of the data in those areas first. For example, an institution may want to look closer at diversity and equality, in which case the data governance strategy should focus on HR in the first instance. Or there may be an issue with student recruitment and retention, in which case the first priority should be student data. Building data governance around user needs is essential to its success.

The benefits available to universities from data and AI are increasing exponentially as technology and practice evolve at pace. Those institutions that get data governance right will be best placed to maximise the coming opportunities.

Leave a Reply