The problem with data governance? It’s too elitist!

Well, at least that’s the perception out there. Some verticals have embraced data governance out of sheer necessity, think of banking and insurance for instance. Large corporations also seem to have grown into the idea that if business is going to be increasingly digital, maybe they need to organize how digital operations are run.

All too often though we hear from senior executives that data governance is about compliance: “It’s a must-do in order to make sure we don’t transgress the rules.”

Organizations with an in-house Chief Data Officer (CDO) usually have a more complete picture of data governance, but the vision of ‘governance = compliance’ has trickled downmarket, creating a perception of data governance as a cost, rather than an opportunity to leverage data’s value.

Let us first define what we mean by downmarket. This can include companies with say 100+ white-collar workers, private companies (who have easier reporting requirements than public ones), companies in less digitized verticals, etc.

Possibly due to the industries where data governance first launched (i.e. banking and finance) where regulatory forces have led the way towards data governance, efforts have so far mainly focused on compliance and/or data quality. In other markets though, the main need for data governance comes from a business requirement: efficiency.

Data (in the broad sense: datasets, reports, KPIs, etc.) is a business asset, and just like other business assets, companies carry inventories of data. Whoever remembers the advent of ERP or warehouse management and inventory optimization tools knows that up to that point, finding an inventory item typically required asking an inventory manager who’d translate your request for a part into a stock item code, which he’d then input in an AS/400-like system to hopefully find it was on stock.

Then, a warehouse employee would be able to locate and pick the part for you. Very often, the right part wasn’t immediately found either because the correct stocking code wasn’t known or a code was entered for a similar but not exactly the same part. In companies with multiple stocking locations, one often had to resort to calling the other warehouses to find out whether they carried the part.

Thankfully, those days are gone. However, when it comes to digital inventory, we unfortunately still see this in action. A manager needs a report and doesn’t immediately find it. Therefore she asks a colleague who doesn’t know either. Typically after a trip to the coffee machine someone will have what they’re looking for, or at least they think they do.

The main problem with digital is how easy it is to make copies. And copies tend to live their own lives. Therefore, how can we confirm that the report that was located is the correct one? Who made that report? Are the KPIs in the report built using the definition as (should be) recorded in the unified glossary?

The funny thing about that situation is that almost everybody we speak to, whether from an organization of 100 people or 10,000 people, all confirm this situation. Many CIOs will blame this on the availability of self-service BI tools. With the introduction of PowerPoint, it became easy to have people build their own slide decks. Corporate styles were lost overnight and even today companies struggle to stick to a standard style.

For BI, some companies have set up central analytics teams who build reports on request. Once reports have been delivered, questions keep trickling in in the order of — “Where can I find report such and such?”, “How is that KPI calculated in this report?” or “Where is the data in this report coming from?” And they’re honest enough not to blame the end-user.

It’s just that they lack easy inventory management and documentation. Meanwhile, those many questions take up a lot of time from the analytics team, who in turn, build up an ever-growing backlog of new report requests. On the opposite side, companies that empower people with self-service reporting tools face challenges of inefficiencies in finding relevant data, lots of duplicate report building, and confusion around how to correctly interpret data fields.

And that’s where the view that data governance is elitist and “only for very large corporations,” hurts many businesses. They keep running their data and reporting inventories ad hoc even though they could easily be fully digitized.

That’s why we advocate that discussions about data governance shift from “must do” to “can do.” Data governance can be done in steps and first and foremost, it should focus on making business easier and more efficient. The latter speaks for itself but the “easier” part is equally important.

CDOs and by extension, anyone responsible for data governance within their organization should focus on putting processes and tools in place that are easy: easy to understand, easy to deploy and maintain, and easy to use. The success of data governance initiatives arguably depends principally on adoption.

If your data governance initiative takes 6 months to show the first results, chances are pretty high that you’ll have as many detractors as supporters even before you go live. Our advice: focus on quick wins; heavy lifting can be done later or in parallel. The key to success is user adoption.

About the author:

Combining engineering and MBA degrees, Carl Fransman has spent most of his career in analytics and AI. As the Chief Strategy Officer and Co-founder of dScribe Data, he now focuses on advocating efficient data usage. Data is everyone’s business; we must therefore ensure we guarantee correct comprehension and provide tools for easy, but managed, access to data and reports.