The National Data Strategy Expert Roundtable

On the week of 16th November, a select group of experts – all data leaders in leading public, private and academic institutions – came together to discuss the National Data Strategy. This article summarises the key points of discussion and consideration for those concerned with the strategy.

The National Data Strategy sets out the UK Government’s ambition for data. It provides a framework for future data policy and outlines a commitment to action. It is a key part of the UK becoming a global data leader.

The UK needs a data strategy to support efficiency and growth across all sectors. In the public sector, it will support more effective delivery of policy and public services. It will also encourage scientific development, increasing the scope and speed of new research. In the private sector, it will support business development and stimulate new markets. How these benefits are realised will depend on the choices made in the content and execution of the strategy, which was largely the focus of the discussion.

Prioritisation and choice of approach are critical to operationalising the data strategy

The need to prioritise, and the challenges of doing so, was a recurring theme of our discussion. To effectively prioritise actions within a data strategy, we first need to understand three things:

1. What are our critical data assets?

2. What value are they currently creating and what could they create in the future?

3. How are they supporting our objectives?

The structure of funding is a significant barrier. Funding is often organised vertically, by department, teams and projects. However, the outcomes of investing in data are felt horizontally, across teams, departments and sectors. By saying investment decisions have to deliver returns in one silo only, data projects may not create enough value to any single department to justify its costs, leaving a lot of untapped potential.

One helpful analogy was the distinction between data that needs an engineering approach, versus that which need a gardening approach. ‘Engineering’ is where there’s more structure, centralised control and specifically designed actions. Alternatively, a gardening approach is gentle stewardship, encouraging autonomous data use within clear structures and systems. These approaches mirror the discussion of top-down and bottom-up approaches. Particularly how policy and regulation work with an entrepreneurial approach.

The discussion was a reminder that different contexts call for different tactics, finding a balance between these approaches to suit the situation.

Anmut reflections

Data assets typically support multiple use cases across different departments of an organisation, who in turn use that data to drive activities that create value for diverse stakeholders. Prioritising data assets then is a question of understanding which assets create how much value for stakeholders and what investment is required to improve value creation.

The uneven distribution of costs and benefits of developments in data needs consideration.

The benefits of investing in and treating data like an asset, don’t always accrue to those who make the investments. Nor do we see outcomes linearly over time, with some outcomes only coming into play in the long term.

Participants repeatedly highlighted the lack of coordination around projects to improve data, and how activities in organisations are often dispersed, resulting in duplicated efforts. This raised questions of how we incentivise actors (be they public, private or third sector bodies) to do the right thing when they don’t get the benefits directly, but rather they’re transferring the value to others? It also means having robust ways to understand, compare and trade off different stakeholders’ priorities, what type of value creation we prioritise, and whose interests we serve.

Lack of clarity is an issue for the costs associated with data. Studies have shown that approximately 5% of revenues in the private sector are spent on data, but much of this is hidden because the data is siloed. For example, one organisation had six analytics teams building similar solutions on top of flawed data.

Anmut reflections

The combined benefits of improving data assets can be extraordinary, for example, better managed roads can reduce journey times, improve economic output, people’s quality of life, reduce carbon emissions and avoid fatalities. In our experience, inside organisations, investments in data projects are made based on the return to the team initiating the work. This limits the benefits of improving data assets. By understanding the benefits that accrue to multiple stakeholders, and combining data projects together, it’s possible to better align both direct self-interest with broader stakeholder benefits. This may well be easier in the public sector than the private sector, unless a private sector organisation is following a stakeholder approach to capitalism.

The intersection of public and private data is complex

This is a particularly complex area, bringing together a lot of the points we have already addressed.

  • What is the right level of formalisation to create structure without encumbering innovation?
  • How do we translate the high-level strategic outcomes to prioritised solutions?
  • How do we future proof our processes so we don’t create legacies of bad data?

In addition to these questions, there is a fear in the public sector about the motives of the private sector. Is it right for the government to invest in producing and maintaining open-source data that is, in the end, used purely to pursue private profit? One interesting solution involving neutral data custodians was suggested. These kinds of practical solutions will allow us to navigate this space.

Anmut reflections

There is a risk when sharing public data with private companies that it only becomes accessible to large companies who can afford the capabilities needed to use it. If the value of data were clear, could sharing of data become a policy tool to enable the government to hit its targets? More broadly, we think this raises important questions around data ownership and stewardship. Where does responsibility for maintenance lie? Whose needs do we prioritise for?

Accessibility and fitness for purpose are important, from the beginning of the data lifecycle, and are often misrepresented through analytics

We talked about how, in the public sector more data could be captured in a better condition for different use cases. We also talked about how there are many opportunities in the public sector to collect better data, and make existing data more accessible. The private sector, in particular, may need further incentivisation, and reassurances about possible liabilities, to make more data accessible.

Practically, there was talk of embedding the idea of accessibility and fitness for purpose in procurement and with IT departments (when setting up new data bases) to avoid creating legacies of even more bad data.

For asset management organisations, fitness for purpose needs to be considered from the beginning of the data lifecycle, because they tend to procure data on the state of assets, often in the form of surveys by engineers that are then manually input into IT systems. Definitions and quality parameters often are not aligned across the organisation. When data gets into analytics, with easy to use tools like Power BI, the impression given is one of impressive results, but this often rests on data that is fundamentally flawed.

Anmut reflections

Making data openly available and accessible for as yet unknown use cases is expensive with uncertain return. While fully open data is an ideal outcome, the reality of organisations and governments operating in tough economic circumstances make getting there a question of prioritisation. Where to invest in fixing data problems and/or making data more accessible demands a decision about what purposes are being prioritised. In turn, this means making inevitable trade-offs between stakeholder groups and their respective outcomes. Something that requires clear and agreed information itself, and has inevitable political considerations.

Digital vs data must be considered in communication

The distinction between digital and data was highlighted. Communicating this difference will be an important challenge for the expert community to overcome as we aim to shift the conversation towards data.

Anmut reflections

Anyone working with non-technical people will know that data and digital are confused and conflated. This poses a risk to the NDS, in that flashy technology solutions are often preferred by organisations but won’t deliver because they don’t address the underlying data problems.
The National Data Strategy is a landmark in the UK’s journey towards harnessing the power of data to better navigate an increasingly complex world. From this discussion of experts, we can see many of the barriers to fulfilling the strategy relate to the unique nature of data as an asset. It’s through conversations, like the one that led to this article, that will reveal the best routes forward, to fulfil the aims of the strategy.