THE big decision your organisation will make on your journey to being data-driven is what type of governance you will put in place for your data and content. Ideally your governance model will be secure, while also enabling your staff to use the data to make better decisions.
Tableau Blueprint describes three models of governance. We’re going to explain each in turn and offer some real-life examples of each model in practise. You might also be interested in the other articles in this 'becoming data-driven' series.
Centralised
Your data and analytics environment is owned by one central department - usually IT. They curate the data sources and reports and make them available to your analysts and wider business users. You may have a centralised data governance structure for the following reasons:
- Lack of data literacy or skills in the wider organisation.
- The data required is extremely sensitive and needs in-depth monitoring of who has access.
- You have an existing traditional top-down IT or data strategy that isn’t changing anytime soon.
Here are the potential pitfalls of a centralised strategy:
- The owners become a bottle-neck for keeping up with requests from across the business - resulting in business decisions taking longer or being made without the right information (exactly what we’re trying to avoid by becoming data-driven).
- Word never gets out to the rest of the business that the data or reports are available for them to use because they haven’t been involved in the process of curating it. Your investment is never fully utilised.
- The data and analytics skills-gap in your organisation never gets addressed.
Think Centralised is for you? Then read some real-life examples: Simplyhealth.
Delegated
In a delegated model ownership and responsibility for the data is given to people outside of the central team. This may include using Site Administrator or Project Leaders roles in Tableau Server for changing permissions. Delegated requires comprehensive processes in place to validate and certify data and content that is published. In some delegation models it may fall to the centralised team to certify finished content by the delegates.
You may have a delegated data governance structure for the following reasons:
- Data literacy is good in some areas, but still needs improvement in others.
- Some of the data is sensitive and still needs to be handled by a central team only.
- Your organisation is slowly moving towards self-governing or self-service.
- You still need to verify users content before certifying it as data skills are still being built.
- Reporting and data requests are outstripping a centralised teams ability to produce.
Here are the potential pitfalls of a delegated strategy:
- You require a comprehensive process of validating and certifying data and content that is signed off by users to say they understand it.
- A training scheme for users is required to enable them to create good content. Little or no training will only produce poor content or a mixture of poor and good content (where users train themselves).
- Site Administrators or Project Leaders require training to ensure they understand the nuances of the roles in Tableau Server.
Kris Curtis discusses Just Eat's delegated governance strategy.
Self-Governing
Content and data is created regularly either by Creators in Desktop or by Explorers in web edit. All users including Viewers have a good level of data-literacy. Ad-hoc or sandbox content vs certified content is differentiated and the process of promotion to certified is clear and well-defined. Analytical skills are good across the organisation.
You may have a self-governing governance structure for the following reasons:
- Data literacy is good across the organisation and users want to be able to answer their own questions with the data.
- Demand for reporting would fast out-strip the supply by a centralised team.
- You have an open data policy in your organisation - where all staff members are allowed to see most (non sensitive) data sources.
Here are the potential pitfalls of a self-governing strategy:
- Monitoring of your Tableau Server environment and the ability to scale up quickly is required.
- Creating custom admin reports based on the Tableau Server data to track who has accessed what may be necessary for regulation requirements.
- Requires appropriate and frequent training for all users at all levels (Creator, Explorer, Viewer, admins).
So, which governance model is for you?