Data governance and operating models
Governance that lives in a document changes nothing. Governance that lives in the platform is what keeps it trustworthy on the day someone questions a number.
Most governance programs produce artefacts. The useful ones produce decisions: who owns this dataset, who may see it, what happens when it breaks, and how anyone can tell where a number came from.
Aurora Data designs governance that is implemented in the platform itself, in lineage, access, sensitivity and monitoring, together with an operating model a real team can sustain after the consultants have gone.
What the work covers
-
Ownership and stewardship
Named accountability for domains and datasets, at a level of detail people can actually act on.
-
Lineage and cataloguing
Being able to answer where a number came from, without a workshop and a week of archaeology.
-
Access and sensitivity
Classification and access design that holds under audit and does not quietly block the people who need the data.
-
Quality and monitoring
The checks that surface a broken pipeline before a report does, and who is expected to respond when they fire.
-
Day-two operating model
Change, release and support responsibilities for after go-live, when the project team has moved on.
How an engagement runs
-
Governance review
An assessment of what governance exists in practice rather than on paper, and where the gaps carry real risk.
-
Operating model design
Roles, decision rights and forums sized for your organisation instead of borrowed from a much larger one.
-
Implementation support
Putting the model into the platform, and coaching the people who have to run it afterwards.
Questions people ask
Is this a policy exercise?
No. Policy is the easy part and rarely the constraint. The work here is the ownership, lineage, access and monitoring implemented in the platform, plus the operating model that keeps them current.
Do we need a data catalogue product first?
Not necessarily. Tooling follows the operating model. Buying a catalogue before deciding who owns what usually produces an expensive inventory nobody maintains.
Who should own data governance?
It depends on the organisation, but it cannot sit entirely with the platform team. Part of the engagement is establishing accountability that survives a reorganisation.
Talk it through
Engagements start with a conversation about what you have and what is actually in the way. If that sounds useful, send an email and describe the situation in a paragraph or two.
Background on the practice is on the home page, and on the architect leading it on the profile page.