I just finished reading a great blog titled ‘How to Add Value as a Data Analyst’ by Cassie Kozyrkov, Chief Data Scientist at Google. She discusses what the role of a data analyst is and is not.
It's familiar ground, so I almost skipped over it, but then as I read two things occurred to me. First, it's easy to forget the fundamentals and get lost in our complex day-to-day, so it’s a good ‘saw sharpening’ exercise. Second, my mind leapt to how we use or don’t use data analysts as part of our governance programs.
As governance leaders and advisors, we are often outwardly focused on how we add value and mitigate risk for others. In that context, we think of serving data analysts in their role of helping the business. My focus is not on that, but on having our own governance data analysts.
I suppose what triggered this thought was my recent work on governance policies and their associated target metrics, measurements, and thresholds. I consider both policies related to data and metadata to be equally important. For example the quality of data in a column and the business title, description, and security classification of the column.
This dual orientation and the associated measurements lead to the question of who is responsible for them. I think many of you might have the same reflex response as me, it's the data steward. Data Stewards have a scope of responsibility within a given domain and they are therefore responsible for ensuring that target measurements for both data and metadata are maintained to a standard recorded in the policy.
From a reactive and maintenance-oriented perspective, I think that is right but it leaves a lot of unanswered questions. Who looks across all the measures and analyzes how the entire governance program is doing? Who goes beyond simple descriptive and diagnostic reporting to use higher-order predictive or prescriptive analytics for program optimization?
Again, I think our reflexive response is that it's the governance leaders' responsibility. After all, they are leading the program so shouldn’t they be the ones who understand how it's doing and think about how to optimize it?
The answer is obviously yes, but referencing Cassie Kozyrkov’s blog we need to be careful that we are not incorrectly conflating the role of an analyst and decision-maker. My experience is that we have been doing that unconsciously for a long time. There are a lot of reasons for that but one thing is very clear, there is a new, heightened regulatory and business demand for governance to play a pervasive and intrinsic role in how data is used throughout the enterprise. In response, the scope of governance programs is expanding, requiring increased discipline in the measurement and management of the program.
This expansion is increasing the complexity of measuring the program. Consider a simple example of reporting on sensitive data. We need to know of its existence, steward coverage, policy association, usage, and compliance. Now throw on top of that geography, system, and organizational department. Already we have a multi-dimensional rubrics cube that needs to be analyzed. Of course, this is only one of many, many governance topics that each have their own rubrics cubes of measurements.
The conclusion I’ve reached is that the increasing sophistication of these programs requires a formal separation of duties where professional data analysts serve and report to governance leaders in staff positions. In that role, they will be required to apply their exploratory talent and ability to provide actionable insight for leaders just as data analysts do for all other critical business functions.
The next step for me is to formalize this by revising the templated role descriptions I use for organization design. I would recommend you think about this and if you reach the same conclusion, start to do the same.