Data Integration Blog -

You may have noticed the words “Data Governance” cropping up every now and then in articles about BI, and particularly articles about MDM. But what is it? And Who cares?

We have gone through decade after decade of IT without anyone entertaining the notion that data should be governed in some way, never mind what form such government should take. But those days have passed. Data Governance has arrived. Nowadays data has rights.

Really?

Well yes, but only by proxy. Data has users, and some of those users are sentimental. They like to be sure that their data isn’t abused, mistreated and injured by those who don’t respect it for what it is. It’s those pesky users that would like their data governed and there is a plethora of them. In fact there are even the data carers, data social workers if you like, in various regulatory bodies that really care that data is looked after responsibly.

Mr. Sarbanes and Mr. Oxley, for example really don’t want just any Tom, Dick or Harry taking a peak at corporate financial data. And Mr. HIPAA, bless his heart, doesn’t want anyone’s health data being tweeted about on Twitter. And there are government-backed data protection rackets in Europe that just hate it when your personal data is incorrect. If you complain to them, they might just send the boys round to any business owner who is failing to show the proper respect.

Now you know as well as I do, that with advocacy groups of that kind, data is getting a lot more love and affection from people in high places with check signing capabilities that it ever used to. And that’s all grist to the mill.

But what’s even better is that the data glutens in many organizations are getting pretty sick of bad data. Now I know what you’re thinking, “they’re lucky to get any data at all.” But it’s worse than you might imagine. It’s not so much that some of the data isn’t as pristine as it should be, but before you know it, the bad data has left the production systems, infected the data warehouse, infested all those data marts and it’s sitting in spreadsheets or whizzy BI tools assisting the data users in making less than perfect decisions. And that’s not so easy to fix.

Truth be told, we rarely hold audit trails of where data actually came from. So once bad data gets included in an average or an aggregate, well, the damage may never be noticed and, anyway, it is pretty difficult to undo. One solution to this conundrum is to fix a sign to all replicated data, which says “caveat usitatus,” (let the user beware) but that’s probably not going to work because even though, etymologically, data is a Latin word, the usitati rarely speak Latin.

So there’s nothing for it, if you want to do something about it, but to seek a big check from the big cheese and embark on a data quality project, hopefully with the assistance of some resumé-enhancing software.

Why Not Go the Whole Hog?

But wait a minute. If we’re talking about sparkling up your resumé, you don’t want to simply have Data Quality Project Manager written against your current Job Title, you want the words Data Governance Overlord to appear there. Here’s why:

Data Governance is not going to go away. It’s not just about regulatory compliance and data security and data quality and data management. It’s also about Data Architecture and Master Data Management. In fact a really creative Data Governor can gets his finger into most pies. Also the data just keeps on growing and growing. As a consequence, your data domain keeps on growing with it.

OK, I know what you’re thinking: “Damn this is going to be a difficult job.” But think positively about this. On your side is the fact that Data Governance is new, so nobody in most organizations has any idea what the job is, and that gives you the right to define it yourself.

Going The Whole Hog

So before you even think about what responsibilities you’d like to assume, make absolutely sure that everyone understands that this is a strategic job.

Secondly, for that reason and because of regulatory pressure, this job is going to be a very political job. So identify the political heavyweights in your organization and make friends with them immediately, assuring them that their requirements are the ones you intend to meet first.

Next and make a plan. Don’t promise any early deliverables. We know what kind of trouble that gets you into. First paint a vision of the future where all data is clean, all data is secure, all data is well managed and those vermin from the regulatory bodies harass other businesses and only bother you once a year when they turn up to claim their obligatory free meal and declare your data to be in robust health.

Now take a data inventory. Yes, you are going to need to take an inventory of all the data heaps within the organization so that you know exactly what you are trying to govern. And from then on, if possible, you’re going have to keep tabs on them to see how they’re being used and to see how fast they’re growing.

Next assign someone the job of estimating the value of the organization’s data. This is a hopeless job in truth, because nobody really knows the value of data beyond the fact that the business will die if it gets destroyed. But that just gives you an opportunity, with the help of Wikipedia and various learned papers, to blind everyone with data science. You might even like to survey the users and ask them lots of embarrassing questions about how damaged they think the data is and whether they get the data they need – and then slip in a question at the end which asks them how necessary it is to their job. That will serve you well.

Next, declare yourself to be a risk manager. No one is going to give you the budget to completely solve the data problems, so you’ll have to blind the rest of senior management with risk assessments on the current “state of the data.” You know; the risks of poor data quality and the risks of poor data security. The whole point here is to cover your ass.

Finally, produce regular reports which not only keep everyone informed about the progress of your data initiatives, but actually produce figures on risk levels associated with bad data and insecure data.

I fully confess that this may not work, in respect of actually doing a good job. But if everything is heading south, you’ll know long before anyone else does. So think of it as a learning experience. You can just get a better paid Data Governor’s job elsewhere long before your bold initiatives crumble and the usitati want to throw you to the lions.