Over the last years, there has been a growing recognition that research data are of significant value. The importance of transparency, reuse, and verifiability of research data is often emphasised. Many funders encourage or demand researchers to critically plan their Research Data Management (RDM) at the start of a project, and to retain and possibly publish datasets once the research has been completed (e.g., European Commission, 2016). Publishers, too, are showing awareness of the value of data, asking or sometimes requiring authors to share the datasets underlying their publications (e.g., PLOS, 2017). Researchers recognise the value of RDM too, since it brings incentives of various kinds, such as efficiency of research, more societal impact, and increased chances of getting funding (Hoetink, Broekhoven, & van den Hoogen, 2016). Overall, the scholarly landscape reflects a growing concern with making data FAIR: Findable, Accessible, Interoperable, and Reusable (Wilkinson et al., 2016).

To ensure that they meet the above-discussed requirements, and that their research data assets are safely guarded and fully exploited, more and more universities are implementing research data management policies. While data policies have been most prolific in the UK and Australia (Shearer, 2015), they have also been found to exist at 44% and 41% of North American and European universities, respectively (Briney, Goben, & Zilinski, 2015; Tenopir et al., 2017).1 Where policies are not yet in place, they are often planned or in the process of being established. Overviews of implemented data policies are provided and maintained by the Digital Curation Centre (DCC) for the UK (Horton & DCC, 2016), the Australian National Data Service (ANDS) for Australia (ANDS, 2017a), and the National Coordination Point Research Data Management (LCRDM) for the Netherlands (LCRDM, 2017).

Data policy development has often been supported or even led by university libraries. Tenopir et al. (2017) report that two-thirds of the European libraries they surveyed is involved in the development or planning of policy related to research data. This key role of the library is not surprising, as libraries tend to already provide services on data-areas such as metadata and archiving. Moreover, libraries often have a central role in the organisation, collaborating with other departments such as IT Services, as well as being close to researchers (Erway, 2013).

Although there are some common themes that most data policies seem to cover (Berchum & Grootveld, 2016; Shearer, 2015), there is great variation between the policies of different universities, and universities have taken different approaches to defining their policies (Jones, 2011). To provide policy makers with a starting point, several templates have been developed (ANDS, 2010; Budroni, Sánchez Solís, & Traub, 2017; Hall, Corey, Mann, & Wilson, n.d.). These model policy guidelines cover a set of important research data themes, such as data retention, ownership, and the responsibilities of stakeholders. Institutions can select and adapt these themes to suit their own context.

While such formats can certainly provide a useful basis, a final policy needs to meet the data practices and requirements of an organisation. Context-specific information such as available storage and archiving services need to be taken into account, together with the actual data management needs of researchers. There are tools available to help institutes with such assessments, such as the RDM Readiness Survey (LEARN, 2017), the RDM Capability Maturity Guide (ANDS, 2017b), and the Data Asset Framework developed by the University of Glasgow and the DCC (http://www.data-audit.eu/). These resources help to identify which services are in place and which ones might need further development. While these tools are not developed to inform data management policies per se, they can provide information that is key to RDM policy development, such as the data assets present and the storage/archival practices used.

To get such information first-hand, universities often approach researchers directly. One example is the Monash University in Australia, which engaged their researchers in the framing of an overall RDM strategy, including a policy. This engagement gave the policy makers valuable information about data practices, and simultaneously allowed them to communicate the benefits of good research data management, minimising the sense of risk and compliance that often comes with RDM policies (Jones, 2013). The University of Leiden in The Netherlands also actively involved research staff in the creation of their policy (Verhaar, Schoots, Sesink, & Frederiks, 2017). They did so not only to ensure that researchers could comment on the policy, but also to create a sense of support amongst them. In the UK, the Universities of Southampton and Surrey have used surveys and interviews with researchers to set up their RDM strategies. As argued by Rans and Jones (2013), such engagement offers a complete and accurate view of the data practices and needs in an organisation:

As will be discussed, the RDM policy at Wageningen University & Research was also set up with the help of involvement with researchers. This approach was chosen not only to gain understanding of the existing data management practices, but mainly to identify certain ‘best-practices’ in the organisation: use cases that can be considered exemplary in how they manage their research data. Such best-practices are highly informative in the definition of a RDM policy, for two reasons: (1) they outline how data should be managed, thus providing a basis for policy guidelines, and (2) they relate to the data management workflows of researchers, thereby providing concrete examples of how the established policy guidelines might work in practice. As will be discussed, these best-practices were identified through the assessment with certain criteria, which were set up with the help of existing regulations and frameworks on data management.

While the above section has focused mainly on the factual information extracted during the interviews, the engagement with researchers has proven fruitful in other ways. For one, it has provided some insight into researchers’ familiarity with the data support services at the institution. This has indicated what aspects of our services are known (e.g. data storage), and which might need a little push in terms of visibility (e.g. data registration). When informing researchers about the policy in the future, the less familiar services might need highlighting. Another valuable experience has been to learn more about the general data management practices in different research groups. Besides the preset questions, various issues were discussed during the interviews, such as file synchronization, version control, and the use of laboratory notebooks. Engaging with researchers about these issues is key to staying up to date on their daily data management practices and, in turn, to provide them with the necessary support.

What truly helped in setting up the guidelines was to consider the diversity of the organisation from the start. Use cases differed in what data they kept in terms of format, size and confidentiality level, and in where they stored and archived their data. This indicated that the guidelines should be sufficiently general. The final guidelines therefore outline which solutions are allowed, while also approving other (group- or domain-specific) solutions if these meet the M-criteria.

Another decision made was to take the institutionally available storage and archiving solutions as a starting point. The use cases indicated that the available solutions largely met their needs: researchers either already managed their data using the available solutions, or they could if they were to transfer their data. The guidelines could therefore be set up quickly, as the infrastructure was in place. Simultaneously, the interviews indicated for which data types the existing solutions were not appropriate (secret and very large datasets). A follow-up project explores what solutions might be appropriate for these data types in particular. Once identified and implemented, these solutions will be added to the guidelines.

This article has discussed how use cases have helped us define a policy, but their contribution is certainly also valuable in the follow-up steps. At the moment, the use cases are interviewed again to assess if the established guidelines are achievable. The policy guidelines are compared to their current data management practices, and any gaps encountered are bridged (e.g. datasets are registered to meet the guidelines). This is a highly useful follow-up step. It allows researchers to prepare themselves for the policy at an early stage, while it helps Research Data Management Support to focus and possibly expand our support services where needed. In doing so, the library’s central role in the organisation is used to its advantage (Erway, 2013), working together with other departments such as IT and Legal services to ensure that the necessary support is established. Moreover, continuing the conversation is important for creating support among researchers and for ensuring that the guidelines are credible (Rans & Jones, 2013; Verhaar et al., 2017). Credibility of the guidelines is likely enhanced due to them being built on best-practices in the organisation. Use cases that store their data on safe solutions, archive their data in trusted repositories following the FAIR principles, and/or consistently register their datasets, show other research groups how the guidelines can be followed in practice. In fact, if these best-practice use cases do not object, their data management practices will be used in communicating the new policy, serving as exemplars to other researchers.

Needless to say, the establishment of RDM policies takes time. It is a lengthy process of which defining the guidelines is only an initial step. Once in place, policies need to be integrated into researchers’ behaviour, accepted as part of their culture, and continuously reviewed and audited (ANDS, 2017b). However, as found at various universities (Rans & Jones, 2013; Verhaar et al., 2017) engagement with researchers is helpful in the further roll-out of a policy. Thanks to having involved use cases in this study, Data Management Support has an established network to engage with, which should be beneficial in the next steps of policy communication, implementation, and support.