Data Governance

In the corporate world, you can pick any noun, attach the word “governance” to it, and you have now invented a new necessary function into your organization, which obviously requires resources to manage it. Corporate busybodies come up with interpretations for guidelines and “frameworks”, which were passed on to them by their public sector busybody counterparts and the entire self-perpetuating process has set tens of thousands of people up for a lifelong career of doing nothing much at all.

Governance is not doing its job if it’s not making your job harder. Governance does not exist to “streamline” or make anything “transparent”. Governance is there to make sure you write documents about the work you don’t have time to do because you’re writing all those important documents.

There is one blessing when it comes to governance, though, and that is that it’s completely detached from reality. It exists solely as fiction written on documents that no one will ever read, yet of which the authors are very proud. Thus, whenever a document or “guideline” is written up and handed over to an innocent analyst whose job it is to interpret and simplify it into a powerpoint presentation or - heaven help you - a process diagram, only those presentations and process diagrams are ever reviewed by auditors or governance specialists. No one is the least bit interested in whether or not the actual work that takes place follows this set of guidelines, but governance begets governance, so naturally a separate governance process is implemented to review these processes.

All that these reviews do is bring up the process document so that someone can check the box that says the process has been reviewed. Of course, the drawback is that if the review doesn’t find a problem with the imaginary process that no one follows, but which was signed off on a year ago, then the review has fallen short. So you will probably have to write more documentation about how your process was outdated and now it’s up to par with whatever someone got paid to say about it.

Data governance is a subset of governance that, clearly, concerns itself with governing data. Since data is the cool new thing, inventing governance for it was only a matter of time. We do need to put those mathematically illiterate elites to work somewhere, so why not in data governance?

Excluding a handful of social media corporations, data is a side product, not the actual product. If your business provides a service or a commodity, it makes little sense to treat your data as such, even though social media giants want to make you believe otherwise. I’ve been working as a data analyst for 15 years and I had never even thought that you could “govern” data in some way, nor that it would even be necessary to do so. Data gets stored in databases because you want to pull up customer or product information and if your business is clever, you can use that data to model customer segments and behaviors to increase revenue. That’s pretty much it.

One day, some years ago, completely out of the blue, someone said we will now be focusing on data governance and that someone has invented, as governance has a way of doing, a new job to do just that. This role came with the title of Data Steward. It takes months, even years, to push an actual efficiency increase through in a company (such as a shell script that automates some mundane manual task), or to hire a new data analyst, but it only took a week to hire dozens of data stewards and their managers.

Since nothing of this important and necessary function all these data stewards were fulfilling seemed to have any effect on what I was doing, although I was called a “stakeholder”, I needed to have the purpose and function of a data steward explained to me multiple times over 18 months." Initially, I didn’t think they had anything to do, because what about data is there that requires it to be governed? I was mistaken, of course, because like all governance, the data variety also comes with copious amounts of documentation. I haven’t read that documentation and as far as I can tell, neither has anyone else besides the small cabal of data governance professionals who occupy themselves in the mysterious craft of governing data.

The first time their function was explained to me went something like this: A data steward catalogues data, not only from the database in which we work, but also other places. In its most extreme case, I was told, if an analyst - such as myself - were to have some trash spreadsheet stored and forgotten on my hard drive, then data stewards should know about it.

For the longest time I thought that we had hired an army of people whose singular purpose was to scour everyone’s hard drives, searching for forgotten spreadsheets. After no one had come for my hard drive, I was forced to conclude, however, that they were up to something else, and I was determined to find out what.

One day my boredom got the best of me, so I decided to listen in on a discussion about data governance to understand what it is that these people actually do. My takeaway was that a junior data steward rewrites documents that have already been written, or which are unnecessary, and sends them to a senior data steward to examine. The senior steward says the documentation is lacking, basing his assessment on a rigid and well-defined metric instead of an arbitrary one, and then the junior data steward has to fix it. Why the senior steward doesn’t just write all these documents himself, I still don’t know. His job is only to read and complain about documents.

I found this nonsense quite fascinating, so I decided to understand more about it. I have to admit, figuring out this puzzle was infuriatingly addictive, and having nothing better to do, I investigated it thoroughly. I had to be careful though, as I was afraid I’d get a stern talking-to about why I’m not taking these people seriously and about how I ought to embrace the governing of data.

I then decided to call up our junior data steward, let’s call him Robert, and have him explain in detail what happens to these documents when he rewrites them as ordered. I was sure there had to be some kind of official process around this. So much effort goes into producing these documents, surely they land in the hands of a regulator or someone who bases their decisions on these documents, right?

Even though my expectations around corporate bullshit and nonsense are set low, the so-called end-to-end process was shocking even to me.

Robert spends his days typing up a document that explains all kinds of different data that exist in our main database. When he told me this, I just emailed him the technical specs of the database and asked if that’s the same thing, but apparently it wasn’t. I couldn’t tell the difference between the technical specs we already had and Robert’s document, but I guess his was more conversational.

Anyway, once Robert was done paraphrasing a document that already exists, it was ready to be signed off by the senior data steward. The senior steward’s job is simply to reject every first draft of this document and give a vague explanation of why it’s not good enough. Robert then changes the wording and resubmits the document to the senior steward, who is now satisfied with how everything looks.

Now, one thing to mention here is that the senior data steward is not Robert’s manager. Junior data stewards are placed in various teams around the organization, so Robert actually reported to my manager. This is important because once the senior steward had signed off on the document, he sends it to our manager for what Robert called “validation”. At the time, my manager’s name was Annie. She had no idea what “validation” actually means, and to be honest, she wasn’t particularly interested in reading about some obscure details about a database she’d never use. Anyway, once Annie had validated this document, she’d send it back to the senior data steward. At this stage, I believe it was possible for the senior data steward to still essentially veto the document and send it back to Robert to start the whole process over, but as far as I could tell, this didn’t happen. Still, the process allowed for it.

After all this validation and checking and double-checking of a non-issue, Robert gets the document back. Now it would be time to upload this document into some kind of data governance platform or repository of sorts. I asked why anyone would go into this platform and read these documents, and Robert delivered quite the surprise: All this procedure and protocol and verification and validation and typing was being done so that I, as a data analyst, could log into this platform and go read these documents.

It was all being done because of me. I asked why I would go and read these documents, considering that I had the technical specs which, from what I could tell, was the exact same content as this document that Robert had written, and he just shrugged his shoulders and said that if I wanted to, I could go and look at what he’d written.

Out of curiosity, I asked Robert for a link to this repository that I never knew I needed. He pasted it into our chat, I copied it into my browser and….

ACCESS DENIED

Robert’s job is to write documents that are proofread by two different people, only for his work to wind up locked away in a repository, and the people who ostensibly benefit from his documents can’t even access them.

Epilogue

I had a performance review with Annie some time after we were committed to governing data. We had some constructive and clear targets to meet, and Annie tasked me with “coming up with a way to utilize our data stewards.” Indeed, their jobs were so nebulous that even the people who hired them had no idea what to do with them, so managers just delegated coming up with some use for them to other people.

A friend of mine is often part of technical interviews at the company for which he works. He usually ends the interview with some kind of a trick question, just to gauge the reaction of an applicant to being completely stumped. I told him to ask about data governance, and it became his go-to trick question to end the interview, and he said it’s perfect. Applicants usually squirm and talk about “data lineage” and “data artifacts” and all kinds of pointless nonsense, and my friend just laughs it all off and tells them not to worry about it, turning the whole thing into a positive and fun moment.

In the corporate world, one firm’s trick question is another firm’s strategic focus for the year.