Open data in Scotland: a blueprint for unlocking innovation, collaboration and impact

4. Recommendations

4.1 Set out a vision that makes the case for open data

What the research found

• Widespread excitement about the potential for open data in Scotland.
• Weariness, wariness and uncertainty about how to win the argument for open data and to turn commitments into reality.
• A desire to see open data publication efforts translate into real impact for a diverse set of stakeholders, held back by the marginalisation of open data as a technical project.
• A potential conflict between recent research suggesting that targeted effort is the optimal route to impactful open data and the ‘open by default’ principle underlying Scotland’s existing open data strategy.

Our first recommendation is to establish a clear and compelling vision for open data that positions open data as enabling infrastructure that supports people and organisations across Scotland to access high-quality information. This vision should be aligned with wider digital and policy priorities and explicitly connect the publication of open data to mechanisms of change and improvement. The vision should be shared, understood and promoted by senior stakeholders to counteract the perception that open data is a ‘nice to have’ rather than a priority or obligation and to give implementers the confidence to publish.

Much of the groundwork for this is already in place, and the consultation on data standards in Scotland’s public sector identified a very similar need. Any vision should align easily with the principles and priorities outlined in the 2021 digital strategy, A Changing Nation, which already sets out how open data initiatives can complement and contribute to broader digital objectives. Both the digital strategy and Scotland's National Performance Framework point towards an alignment with broader policy objectives. Scotland’s Open Government Partnership commitments situate data as an enabler for technical and non-technical audiences who need high-quality information to advance policy objectives across a number of themes.

The following sub-recommendations are designed around producing a compelling and achievable vision.

4.1.1 Broaden the discussion beyond technical outcomes

The research showed an underlying frustration with an apparent focus on technical outputs (i.e. the existence of a particular dataset in a particular format) to evaluate the success of open data publishing, rather than the outcomes that stem from the availability of high-quality information to anyone who needs it. Thinking and talking about information-availability rather than data may be helpful to:

• Reduce the risk of open data being pigeon-holed as a technical project serving a niche technical audience.
• Highlight the gaps between data availability and information access and use.
• Identify user needs around information access and translate these into technical activities in open data programs.

Widely-available, high-quality information will of course continue to rely on solid technical implementation, as later recommendations discuss in detail, but broadening the discussion on purpose will make it easier to justify technical work.

4.1.2 Take a purpose-directed approach to open data publication

Our research suggested that the idea of open data and the principle of open by default are both widely supported. But these abstract and all-encompassing concepts can be overwhelming for publishers (who need to decide what to focus on and in what order) and for open data advocates (who are less likely to make focused arguments to open up specific datasets for specific purposes).

We would therefore recommend that this vision for open data in Scotland reflects the more targeted and attainable approach to publication seen in the so-called third wave of open data, which ‘seeks not simply to open data, but to do so in a way that focuses on impactful reuse, especially through inter-sectoral collaborations and partnerships’.^[60] This should not be seen as a rolling back of ambition from ‘open by default’ but, instead, as a strategy to maximise impact in an environment where public sector publishers face capability and capacity constraints and where data users may benefit more from targeted and concerted effort in specific areas.

4.1.3 Agree thematic priorities and be explicit about how impact is expected to happen

A purpose-directed approach to data publication will require choices about where to focus effort, based on an understanding of how these efforts are expected to deliver impact. We therefore recommend that a vision for open data includes some high-level priorities for open data (while leaving some space for emergent needs) and is explicit about the mechanisms that will connect publication to impact. Together these should allow more effective consultation, planning and implementation.

The digital strategy, the 2015 open data strategy and our literature review identify a number of mechanisms via which open data can have a positive impact. These can be summarised as: contributing to growth and innovation; delivering social value; building trust in government and public services; improving public sector delivery and coordination; and, enabling collaboration between government and other stakeholders. Scotland’s Open Government Action Plan, meanwhile, includes a strong thematic focus on specific data areas, including financial transparency and the environment. Explicitly combining these approaches to document the expected impact of open data publication may be helpful in defining priorities, discovering neglected or high potential areas, and taking a purpose-directed approach to data publication.

4.1.4 Collaborate and communicate on the development of priorities

Our research uncovered a certain level of distrust and cynicism around the Scottish Government’s attitudes to open data. There is a risk, therefore, that a more focused vision for open data is misinterpreted as a turn away from openness, and leads to disengagement and fatalism among stakeholders. Building a coalition around a purpose-directed approach will be one of the key challenges, and requires building trust and dialogue with those outside the public sector. That dialogue will also be crucial to the success of purpose-directed publication, which requires a deeper understanding of user needs and priorities in order to direct resources where they will be most impactful.

4.1.5 Make space for high-ambition goals and initiatives

Our research uncovered a degree of frustration that potentially significant pieces of open data infrastructure may not be available or accessible (e.g. a land information system or an openly-licensed address dataset).^[61] Such potential big ticket open data projects are particularly complex because they rely on significant cross-departmental collaboration, a change in operating model from a data holder, regulatory changes, or digitalisation. They therefore tend to fall down the agenda in short term planning cycles. It is crucial to keep big ideas on the radar, however, as these may become crucial open data assets in the future - and an inspiring and trusted vision for open data in Scotland may create favourable conditions to realise such projects.

4.2 Embrace long-term collaboration and commitment

What the research found

• A disconnect between the high expectations for open data publication and the government’s ability to increase the quantity and quality of data, while making publication simple and routine.
• The effort involved in publishing data sustainably and at high quality is rarely visible, particularly in terms of new and complex datasets.
• Publishers are often disconnected from users and potential users of data, creating a producer-consumer dynamic at odds with the need for collaborative approaches to maximise the impact of open data.

Our research found a disconnect between the complex reality of how open data gets published and the expectation that the government has simple levers that could broaden publication and make publication simple, routine, and high quality - and do so quickly.^[62] Setting expectations too high, and failing to celebrate success, creates a risk of disillusionment among publishers and of disempowering those stakeholders outside government whose input and cooperation is needed to make open data genuinely impactful.

It may therefore be helpful to reframe open data as a collective and collaborative commitment to a long term project in order to change the way the public sector thinks about open data and the way that those outside government think about open data production by the public sector. A more realistic narrative on open data publication is likely to contain several elements:

• Impactful open data publication should be seen as achievable at reasonable cost, with low risks and significant benefits for publishers and external stakeholders.
• The complexity of open data publication, including the policy, governance and technical work involved, should be visible both within and outside government.
• Open data publication should be viewed as an ongoing collaborative process between data publishers and users, rather than a one-off event, and as a way to iterate towards impact.
• Rather than an isolated project, open data is a component of several broader trends in government, including towards openness, participation, and digital maturity - and, while open data can act as a catalyst, it is likely to gain the most traction where these trends are strongest.

Recommendation 3 sets out how this narrative can be made concrete and credible through a realistic and transparent implementation strategy.

4.3 Create an implementation strategy that is intentional, realistic, transparently monitored, and celebrates success

What the research found

• Uncertainty about what the Scottish Government was doing about open data and what it expected others to do.
• An uneven pattern of progress based on organisational factors.
• Negativity about the state of open data in Scotland, particularly from civil society, and a downplaying of progress and achievements in open data.
• The difficulty of finding consistent documentation of plans and progress relating to open data.
• Low awareness of success stories in open data and access to information.

Our research has unveiled several challenges in translating commitments on open data into reality, combined with a low awareness of the significant progress that has been made even among very engaged stakeholder groups. Experience in other countries suggests that the development of concrete and measurable plans has been helpful in motivating and monitoring data publication. In addition, any shift to purpose-directed publication and long-term commitments to open data needs to be balanced with mechanisms to keep up momentum on publication and data quality, and to unlock impact and value as soon as possible.

We therefore recommend the creation of an implementation strategy for open data in Scotland that is realistic, transparently monitored, regularly updated and designed to celebrate success. The following sub-recommendations are in support of such a strategy.

4.3.1 Targeted and intentional data production

Targeted production of high value datasets or data that serves current priorities, over the potentially scattergun 'open by default' approach. This should ensure that open data releases serve concrete purposes and meet defined needs and that resourcing can be directed where it will do the most good. Any data that cannot be made fully open should be justified and transparently communicated.

A national strategy can identify high level priorities and particularly complex data publication projects but there should also be guidance for individual publishing organisations who need to plan their own publishing activities.

4.3.2 Sustainable open data publication

Adopting the principle of sustainable open data publication, with clear expectations set around the lifecycle of the data, including metadata about publication history (a changelog) and transparent communication and a roadmap about the future of the content or availability of datasets. Sustainability also requires ongoing demand, which requires that the data meets quality standards related to coverage, timeliness, accuracy, and interoperability (as discussed in Recommendation 4). The parallel development of publishers’ data maturity will be integral to this effort.

Take a long-term and iterative approach to open data work that emphasises ongoing improvement over one-off data publication. This approach allows for value to be realised early, while ensuring that data quality and other factors improve over time. A long-term approach will also require goals and targets that recognise the importance of improvement in data (discussed further in Recommendation 4).

4.3.3 Review, report, celebrate

Build momentum, trust and accountability through regular reviews and transparent reporting. These plans should be publicly available and easily discoverable. Transparency in progress and accomplishments should extend beyond official OGP plans, creating a comprehensive view of open data initiatives' impact and advancements in Scotland.

Apply standard assessments of the benefits and risks of publishing open data and standard techniques to reduce the risk of disclosing personal or confidential data. Risk aversion was highlighted as a key organisational barrier to publishing open data and risk reduction as a time-consuming and stressful process. Adopting standard methods for assessing and managing risk (for example, techniques for anonymisation, criteria for exclusion or redaction, and clear redress mechanisms) could help here. Making such assessments public could also help build trust in data. The data nutrition label is a useful example of how this could work.

Ensure that success is highlighted and celebrated, both in formal reporting and through other fora that connect publishers, users and other practitioners. Some success is likely to be measurable but it is equally important to identify and tell the human stories behind data publication and impact.

Recommendation 4 sets out a more granular view of how to support the technical activities needed to implement such a strategy.

4.4. Set the standards for open data publication

What the research found

• Publishers appreciate the existing technical guidance resources but are still left with many large and small implementation decisions.
• Less choice and more direction would help publishers and give data users more confidence.
• Users struggle to discover and understand the scope of open data.
• Use and reuse of data is held back by uncertainty over licensing, sustainability, and data quality - and a lack of specialist resources holds back improvement in these areas.

Our research uncovered significant challenges for open data publishers in Scotland, from the technical challenge, the often overwhelming number of choices publishers are faced with, the perceived risks of publication and uneven data governance processes. These challenges reduce the number of datasets that are published and reduce the utility for users.

We recommend that the government provides clear direction, specifying quality standards for open data publication that incorporate technical best practice, usability and findability, and clear data governance, in line with the UK Government Data Quality Principles.

The following sub-recommendations outline how to provide publishers with clearer direction and establish quality standards for open data publication, bolstering the confidence of data users and advancing the effectiveness and impact of open data in Scotland.

4.4.1 Tailor data quality principles to the context

The FAIR Principles (Findable, Accessible, Interoperable, and Reusable) are already integral to thinking on Scotland’s public sector data and the detailed guidance notes provide a useful starting point for building assessments of and targeted support for open data publication. As the needs around open data become clearer, and there is increased justification for practice to converge, it may be useful to consider a FAIR implementation profile (FIP) for the public sector.

Adopt new measures of data quality and coverage in place of or in addition to the target of 5-star open data. The 5-star open data system is useful when information is not published at all, but is an increasingly blunt instrument when trying to improve data quality to unlock specific use cases. There are several alternatives that, depending on context, may be more useful. These include macro-measures of data availability, quality and usability in particular jurisdiction, as used in the EU’s Open Data Maturity Report; functional assessments of data quality and availability on specific themes, as used in the Global Data Barometer; and, domain-specific or standard-specific assessments of individual datasets, as used in the Aid Transparency Index.

4.4.2 Mandate the use of technical standards.

As part of a drive to reduce the burden of publication and improve interoperability across datasets, it may be useful to adopt a more proactive stance on standards for publication and exchange formats, common data components, specialist data standards for complex data, and metadata standards and vocabularies. Adherence to such standards should be measured in any assessment of data quality and publishers should be able to easily test their implementations by validating against the standards. There is a detailed review of this area in Section 2.5.2 of this report.

4.4.3 Make discoverability an intrinsic part of publishing data

Users should be able to find data that they know about and discover data that may be of interest to them - and the experience should be relatively consistent across publishers. The difficulties reported by users in finding relevant data are reflected in the creation of an informal open data portal by Scotland’s open data community and in a government project to improve findability via a CivTech Challenge. Resolving this issue, and ensuring that users can take multiple routes to find a dataset, could solve a significant pain point. Potential approaches include: ensuring that metadata is present; adding markup to publication pages for search engine discoverability (for example, conforming to Google's dataset structured data requirements); requiring visibility at a central location via direct publication or by aggregation; and, ensuring that users can both browse and programmatically interact with the catalogue of published datasets.

Recommendation 5 sets out how to provide ongoing support to new and established data publishers and users.

4.5 Provide targeted technical guidance and support for publishers and users

What the research found

• Existing technical guidance resources are useful but could be expanded to cover more of the publication process and provide clear guidance at specific decision points.
• Data publishers can be isolated during planning and implementation.
• Users can struggle with the purpose, scope and subtleties of open datasets.

We recommend providing targeted technical guidance and support for data publishers and data users, and doing so across multiple modalities. The following sub-recommendations aim to entrench the idea that open data is a long-term, collaborative project, improve confidence in the data itself, and, over time, improve trust in the government’s strategic support of open data.

4.5.1 Self-service support for publishers

Expand and update the self-service toolkit for publishers. The open data resource pack within Scotland’s Open Data Strategy includes valuable reference material and guidance, and an update would be helpful to bring this in line with current priorities and the learning since publication. In particular the practical tools listed in Annex A of Scotland's Open Data Strategy could be expanded to address current pain points. This might include:

• An information mapping tool to assist with planning complex data releases. This could help identify, for example, the components of a planned dataset, any required cleaning or merging, current ownership of the data and data governance policies, any technical details of how the data is stored and accessed, and stakeholders involved.
• A standards catalogue to help identify existing formats for publishing data.
• A list of tools and libraries to assist with publication, including automation of data production and testing.
• Templates for data risk assessments, data sharing agreements and other data governance processes.

4.5.2 Direct support for publishers

Many publishers are starting from scratch with open data publication, in terms of institutional knowledge and capability. Some support provision could therefore be worthwhile to reduce the burden on individual publishers and to avoid reinventing the wheel and divergent practice. The appropriate type of support is context-dependent but could include:

• A helpdesk that answers queries and provides asynchronous support to data publishers. This is a common service model for large data initiatives where responsibility for publication is highly distributed.
• Drop-in sessions for publishers.
• Live and on-demand training on specific issues in data publication.
• Provision of technical infrastructure to assist with self-service publication.

4.5.3 Join up communities of practice

Join up communities of practice and bring publishers and users together. Scotland has several communities of practice in open data and adjacent areas, including Open Data Scotland, the OGP community, organisations in or graduated from the Data Maturity Program, and the government’s own data community of practice. These communities are valuable assets that could be leveraged more. For example, peer support and validation can be crucial for removing the fear factor from open data publication and helping to spread best practice, so providing more informal channels to discuss open data may be helpful. And there is currently no mechanism to bring producers and users of data together to design improvements, discuss priorities and uncover potential demand; establishing a multistakeholder data user group to address this.

4.5.4 Lightweight support for data users

Recommendations related to technical standards in Recommendation 4 inherently improve data comprehensibility and reusability when implemented well. Recommendations related to reporting and celebrating in Recommendation 3 will increase awareness of the availability and potential of open data, so additional support for data users at this stage can be lightweight.

Support for data users can take the form of context, explanation and points of contact for published data. Having data that is comprehensible to non-technical audiences is a strong marker of maturity in open data publication. Data can be difficult to understand if it is just a spreadsheet or other technical artefact. The barrier to entry can be reduced through good documentation, non-technical explanations of the dataset (e.g. ONS bulletins that accompany statistical releases), and demonstrations of data use.