Building a FAIR culture

Data management and the FAIR data principles may feel very new to some R&D organizations, meaning that a significant amount of change effort might be needed to establish the ideas and make them stick. How can a data management-conscious culture be built?

Cultural change and change management are often challenging around any new business initiative, but they’re absolutely essential to establishing effective data management and are not something to take shortcuts around.

The two most important aspects of getting this right are: first, working with the existing culture to get a data management initiative up and running, and second, setting up some key roles to ensure that the initial momentum is built on and data management practices and habits become self-perpetuating.

Launching data management by capitalizing on existing culture

There are four key relationships around any data-producing unit in a typical R&D organization – the relationships the unit has with the IT department, with other units in the business, with senior management, and with data itself. Any of them could introduce resistance to getting a data management effort started.

IT relationship: Mistrust of IT departments is sadly not unusual among research scientists. The IT group often feels distant and impersonal, or like they are always trying to ‘do something to’ R&D rather than facilitate it. Data is owned by the business, of course, not IT, but technological enablement comes from IT and they will also have a lot of data expertise that, if not put to good use, represents a missed opportunity. One solution to this problem is to embed temporary, bridge-building teams directly in and alongside R&D teams. From there they can coach scientists in how to be more responsible data stewards and teach by example by tidying up and curating data sets.

Other departments: Low trust between peer groups can cause ‘not invented here’ syndrome, leading to undervaluing the potential of the work and data of other departments and missed opportunities. Data management itself is often an idea injected from outside. One way to address this is to find entry points into data management that build on activities that the group already believes in strongly. For example, in a quality- and metric-focused engineering department you could echo existing reporting conventions by showing progress on data management as a traffic-light quality metric. This might then encourage the group to attend to the underlying data management issues so that their quality dashboards remain green.

Senior management: Sometimes a research group might be persuaded of the value of data management but need help getting support from senior management to make time and resources available to properly address it. The best approach here is to build a strong business case for data management in terms that appeal to managers. For example, this could be by building evidence that well-managed data makes data handling and processing easier, resulting in cost reductions, or alternatively by finding costly negative examples of the consequences not managing data actively.

Data relationship: Very few people get into scientific R&D because they want to do data management. Successful R&D organizations are experts at something else entirely, e.g. discovering new medicines, developing new vehicles or optimizing new energy generation methods. Data of course powers these activities, but while the need to become a data-driven organization is often acknowledged, it’s not always acted on by researchers, who usually prefer to focus on their specialism. One way to increase the level of interest in data management is to start small, for example by sharing experimental metadata across global teams (scientists often love to know what other groups are doing). This often reveals data gaps (prompting an interest in data ownership) and lower quality data (prompting an appreciation of the role of a data steward).

Establishing dedicated data management roles

Once a data management initiative is up and running, it’s useful to define and establish some specific roles in the organization to accelerate and embed data management and FAIR principles further.

There are three roles that we typically recommend. They aren’t necessarily full-time roles, and they don’t even have to be held by different people – the important thing is to define the responsibilities and scale the roles in proportion to the overall scope of the data management effort.

Data policy lead: A policy and governance guru and advocate of data management. They are responsible for developing policy and guidance around data management principles covering industry and regulatory rules (such as GxP), data lifecycles and infrastructure for different data types.

Domain specialization steward: An expert within their scientific domain and again a data management advocate, the steward handles the day-to-day management of master data (e.g. vocabularies and ontologies) within their group and acts as a bridge between their research group and the other two roles.

Infrastructure steward: A data infrastructure expert who makes sure all technology supporting domain research both adheres to the policies created by the policy lead and meets domain users’ needs.

The three roles are designed to create a virtuous circle where the outputs of each role strengthens the work of the other two. Adoption and evolution of data management practices and technology then becomes continuous and self-perpetuating.

Conclusion

Sum up of the thoughts in the post. What did we learn, and what should we do next?
Overcoming cultural resistance to data management can be one of the toughest parts of a digital transformation initiative within an R&D organization. At Capgemini we have years of experience of identifying and removing these barriers with creative solutions. Take a look at the data management services we offer as part of our vision for Data-Driven R&D and get in touch if this is something you’re struggling with.