Acknowledging that every organisation is different, and every use case is different, varying degrees of governance are required. The data and content governance models can be right-sized and applied to any kind of data regardless of where it falls in the governance spectrum. Establishing three primary governance models – centralised, delegated and self-governing – provides the flexibility to satisfy the governance needs of most organisations.
Like other Tableau platform management activities, an agile, iterative approach is needed to adapt to new business requirements as user adoption and engagement increase across your organisation. Processes will change over time as skills increase and responsibilities are delegated more broadly than the IT organisation. Establish governance review points twice a year to continue to evolve the models.
The Data and content governance tab in the will help you to define your organisation’s governance models based on information collected in the Tableau enterprise architecture survey and Tableau data and analytics survey. You should establish and document who is responsible and what processes support each area within each model: centralised, delegated and self-governing.
Using the matrix approach to separate data and content governance and segmenting by the three models, it is easy to mix and match across models. For example, data and content governance may be centralised at the start. Then, after user training, data governance areas may be centralised, but content governance is delegated or self-governing because the data is curated. Similarly, specific areas within data and content governance, such as delegated metadata management and centralised security and permissions, can be tailored to meet your unique requirements. As business users’ analytical capabilities grow, more responsibilities can be delegated over time. Once defined, the governance models should be communicated to the user community by publishing them to the enablement intranet. For more information, see Tableau enablement intranet.
In a centralised model, IT or another authority owns data access and produces data sources and dashboards for business consumption in a one-to-many manner by a small number of Creators and everyone else as Viewers. Centralised governance is required for maintaining control of highly sensitive data.
Addressing a skills gap among the target audience is another case where centralised management is necessary. You can still provide business users with prepared content to make data-driven decisions while they build their analytical capabilities.
If you are transitioning from a traditional, top-down analytics approach driven by IT or favour a phased delegation of responsibilities to governed self-service, it may be advantageous for IT or a centralised BI team to build the initial use cases across departments, including certified data sources and dashboards.
Over time, as users are encouraged to ask and answer their own questions, the domain of available trusted content will grow organically with the teams and departments, and users will have access to a wider range of analytical content for self-service. To avoid the risk of recreating a “report factory” delivery model, establish goals and dates to evolve beyond the centralised governance model and to begin delegating responsibilities.
In a delegated governance model, new roles are introduced outside of IT or a central authority. Site administrators and data stewards are identified and may have direct access to sources of data. Content authors have access to certified published data sources to ask and answer their own business questions, while some content consumers are given web authoring capabilities to save derivative content to sandbox projects. Processes for validating, promoting and certifying content are introduced but still may be limited. There is increasing collaboration between IT and business users as IT shifts from a provider of reports to an enabler of analytics.
In a self-governing model, there is strong collaboration between IT and business users. Certified content and data sources are available, and ad-hoc content is being created regularly by Creators and Explorers. Viewers understand the delineation between certified and ad-hoc, sandbox content states. The process of validation, promotion and certification is well defined and well understood by users of all skill levels. With increasing analytical skills across the organisation, the boundaries between the roles of the modern analytics workflow are fluid as users switch from consuming to creating to promoting content with the appropriate level of permissions.