It can help governments to be more efficient and effective - open data provides policymakers with access to other departments’ data which can support the design of policies and programmes which don’t fit neatly within departmental boundaries.

It can stimulate innovation and drive growth - a McKinsey report from 2013 identified that more than $3 trillion in economic value globally could be generated each year through enhanced use of open data.

In addition, OGD acts as a means of holding government accountable to citizens and the community. OGD offers the public the opportunity to act as “armchair auditors”, by making available information through which the public can review and interrogate government decisions.

An example of how open data has been used recently is the various visualisations and analyses which were created using the data from the Revoke Article 50 petition.

However, while there are clearly some enthusiasts taking advantage of the government data that is being published, OGD use is not yet mainstream. A review of 101 academic studies from all over the world identified that in practice, OGD is barely used.

This blog will explore why, despite the widespread availability of OGD, it is not being widely used. I will focus on two countries - the UK and Australia.

In the UK, low use of OGD is highlighted as a serious challenge, with a recent academic paper highlighting that, “while thousands of datasets are released by government agencies, the majority is used by no one.”

Similarly, a blog from 2017 highlights that almost one-third of the datasets on the UK OGD portal were not downloaded.

In Australia, 78,798 datasets have been published on the OGD portal (at 1 April 2019). However, despite the availability of data, it appears that not many Australians are using it. We know this because:

• Despite the vast quantities of data available, the data.gov.au site lists only 43 use-cases.
• Site analytics for the data.gov.au site record only 30,462 unique visitors since its establishment in 2013.
• The Open Data 500 project, which studies private company use of OGD in Australia lists only 66 use-cases.
• In a report by The Economist Intelligence Unit, 17% of Australians reported never using OGD – the highest number among the countries surveyed (The Economist Intelligence Unit, 2017).

So, despite Australia and the UK being classified as OGD leaders, the data that is being published isn’t being widely used. The first question to ask is, does this matter?

Many academics make the case that, in order for the objectives of OGD to be realised, it is important that it is being used. For example, Martin argues that “unused data will remain a mere artefact” and Janssen et al. argue that “open data has no value itself; it only becomes valuable when used.”

However, it is important to acknowledge that use matters more or less depending on which OGD objective is being prioritised.

If innovation and economic growth is the primary objective of OGD, it seems clear that use is critical.

If, on the other hand, the primary objective of OGD is to drive democratic accountability, use might be less important, because one could imagine government departments holding themselves to a higher standard simply because they know that their actions and decisions will be publicly available for scrutiny.

Nevertheless, while it is important to acknowledge this caveat, it seems well accepted that in order to realise many of the benefits of OGD, it is important that the data is being used. This blog will proceed on that basis.

In recent years, a significant academic literature has emerged, which explores barriers to OGD use. What this literature demonstrates is that there are many barriers, which are often interrelated.

Given this, it is impossible to attribute low use levels to one barrier alone. For example, there are many technical barriers to use - for example, the publication of data which is not machine readable and reusable; or raw data being published which is very difficult to use by anyone who is not skilled in data use.

However my research has led me to conclude that a key reason why data isn’t being widely used is very simple - it’s because the data that is published is not the data that feels relevant or important to most people. Let’s call this “the relevance barrier.”

To very crudely summarise: the majority of OGD being published in Australia and the UK is technical and scientific, while the data that people want to see is data on social services and tax.

Let’s begin by looking at Australia. The table below shows us that three scientific agencies below publish 31% of the total datasets available on the Australian portal - 600 times more data than the Department of Human Services, the Australian Tax Office and the Australian Securities and Investment Commission combined. Yet the social service, tax and finance data attracts 13 times as many views as the scientific data.

The data.gov.uk site shows similar patterns. Of the 49,591 published datasets:

• 21% of the total number of datasets relate to the environment - weather, flooding, rivers, air quality, geology and agriculture (10,806)
• 15% relate to mapping - addresses, boundaries, land ownership, aerial photographs, seabed and land terrain (7,589)

Whereas only…

• 5% relate to society - employment, benefits, household finances, poverty and population (2,672)
• 3% relate to health - smoking, drugs, alcohol, medicine performance and hospitals (1,924)
• 3% relate to government spending (1,836)
• 0.008% of datasets are published by HMRC (434).

Why is this happening? Why is technical and scientific data being published in spades, while barely any data is being published by human services and tax agencies? What is driving the relevance barrier?

Again, the contributing factors are no doubt complex and numerous. However, I will focus on two key contributors to the relevance barrier:

1. A culture of risk aversion within key departments.
2. Insufficient engagement of the public to understand what they want and need.

A key driver of the relevance barrier is bureaucratic culture, which is characterised by a fear of failure and a culture of risk aversion. In an article on barriers to OGD, Janssen et. al (2012) liken the government to “an oyster that automatically closes up when approached”. Many bureaucrats default to keeping data closed because they are worried that, through publishing data, they might reveal or make mistakes. And, for many bureaucrats, this is something to be avoided at all costs.

This culture of risk aversion stands in direct tension with the culture of openness and transparency required to support the OGD agenda.

However, this cultural challenge does not appear to be universal across all agencies. As I explored above, scientific agencies are publishing vast quantities of data, while other agencies are much more closed. How might this difference be explained?

Firstly, scientific agencies’ data is low risk because it is generally non-personal, non-sensitive data. These agencies are consequently (rightly) less fearful about exposing or making mistakes in releasing data.

In addition, there are natural incentives for scientific agencies to share their data because open sharing is more likely to link them to research and communities to help them understand and solve problems.

This non-sensitive nature of scientific data, combined with the potential gains to be gained by publication, mean that, while a culture of risk aversion may exist within other domains of scientific agencies, it is not a major inhibitor in the context of data-sharing.

In contrast, agencies such as the Department of Human Services hold highly sensitive personal data, which is difficult (although not impossible) to publish safely. In addition, the gains associated with opening up this data aren’t as immediately obvious. As such, for these kinds of agencies, the risks seem to outweigh the benefits, leading to the “closing up”, which Janssen et. al describe.

OGD will be more widely used when the agencies who hold data which is most relevant and interesting to the public are publishing more datasets.

For this to happen, there needs to be a significant culture shift within key agencies - such as the Department of Human Services in Australia, and Her Majesty’s Revenue and Customs in the UK - who are, at present, not publishing much at all.

Driving culture change within organisations – particularly in large bureaucratic organisations – is a notoriously difficult task. Nevertheless, there are some steps which can be used to kick-start the process:

1. Introduce incentives to encourage data-sharing. In practical terms, this might mean building data-sharing practices into performance appraisals as a key performance metric. It could also include the introduction of more informal mechanisms like nominating and publicly celebrating a “data-sharing champion” in internal departmental communications. A great example of an incentive scheme is the Louisville Metro Badge Program, which rewards employees for (amongst other things) creating open datasets.
2. Lead from the top. The example set by leaders can have a profound impact on the willingness of employees to support or resist culture change. To drive culture change around data-sharing, all levels of leadership across the civil service should be speaking about, promoting and encouraging the opening up of data as an essential part of what it means to be a good civil servant. It is important for leaders to prioritise data-sharing as being an essential part of a team’s deliverables, rather than just a side project.
3. Invest in training. Civil servants should be offered training to develop their data skills and understand the value of OGD. Helping people to understand why they are being asked to share their data, and appreciate the value it is creating, is likely to result in a greater willingness to embrace data sharing.

Any culture change programme must also measure change to understand what is working and what isn’t. A great tool to support this is Nesta’s culture change impact framework.

Another reason published data lacks relevance is because the public is not being adequately engaged in the OGD process.

Many governments across the world concentrate on addressing barriers to data supply focusing, for example, on identifying technical barriers which inhibit data publication.

In contrast, there is far less attention paid to understanding what barriers to demand for data might be - in other words, understanding what people want and need from OGD - and where there might be a disconnect between the supply and demand sides.

This is a problem because pushing data out is not enough. As the title of this blog suggests, you can lead a person to data, but if it’s data that they don’t understand, want, or need, they’re not going to use it.

For people to be interested and engaged, data must feel relevant and important. This will only happen when government starts investing more time and energy into developing mechanisms which allow it to hear and respond to what the public wants and needs from OGD.

To be fair, governments are trying. The Australian OGD website offers users the option to “request a dataset” and has established a “Community of Practice.” The UK website similarly offers users the opportunity to request a dataset and they also seek feedback through a user survey.

However, while these efforts should be applauded, they are of limited utility.

Firstly, these user engagement initiatives will likely only reach those who are already aware of and engaged with the OGD portals - they will not build a broader audience base.

In addition, many people have no idea what data a given organisation has, so wouldn’t even know what datasets they could, in principle, request.

More fundamentally, many of the engagement features offered are one-off and uni-directional, rather than systematic and dynamic. A more radical approach to user engagement is needed.

The first thing that needs to change is the mindset around OGD. Governments need to conceive of OGD programmes not as “one-way streets”, but as “ecosystems” with cycles of feedback between data users and suppliers.

OGD systems are made up of many interdependent actors and components including open data providers, open data users, infrastructures and institutions, who must work productively and harmoniously in order for an open government data agenda to succeed.

Taking an ecosystem approach would mean that governments would see their role as being less about “data delivery” and more about cultivating the OGD ecosystem.

The ecosystem approach also encourages a more participatory model of OGD, whereby data is initially constructed by government, but then used, revised and edited by citizens.

A great example of this is the “Demand-Driven Open Data” (DDOD) system, which was adopted (now abandoned) by the US Department of Health and Human Services. DDOD is described as a “systematic, ongoing, and transparent mechanism to tell HHS and its agencies what data you need.”

Unlike the Australian or UK ‘request a dataset’ function, which sees the request sent to the relevant agencies and processed with no further input or engagement from the requester, the DDOD system involves data-users in an ongoing, dynamic dialogue, as illustrated by the diagram below.

Taking an ecosystem approach is likely to address the relevance barrier by engaging users and institutions in a conversation about what they want and need and creating systems and processes which are able respond to that.

Governments are still learning a lot about OGD, and what it takes to ensure that the potential benefits are being realised.

Governments all around the world are (hopefully) learning that pushing data out is not enough; for people to use data, it needs to feel relevant; but this is not yet the case in Australia, the UK, and many other countries around the world.

I have argued in this blog that, for the relevance barrier to be addressed, agency culture needs to shift, and governments need to begin seeing themselves as being a player in the broader OGD ecosystem; rather than the key player.

Happily, it seems that these ideas are starting to gain some traction.

In the UK, Kit Collingwood and Robin Linacre have put out a call to help them start a data revolution for government. They highlight as a key area of attention the need for a greater focus on user needs and an iterative development of data products.

In Australia, the Digital Transformation Agency is doing extensive user-research to understand people’s pain-points with OGD, and what might be done to address them.

These are clearly steps in the right direction.

It is important for all governments to engage in these kinds of conversations and initiatives to give OGD the best chance of becoming the well-used and well-loved resource that it can and should be.