You take the shiny new data analytics tool out from under the tree. You put it through its paces. The tool generates a report, likely driven by opaque algorithms hidden in a black box. Apparently, your company’s widgets have a 30% chance of causing debilitating migraines. Now what?
Well, sometime down the road -- maybe five years, maybe ten -- someone’s going to sue the company over the migraine widget. If the lawsuit gets far enough, both sides will have to exchange information relevant to the case in the discovery process.
That fateful analytics report, showing the widget’s propensity to cause migraines, is undoubtedly relevant by any layman’s definition. And with text messages, Facebook posts, and sundry other nontraditional (read: non-email) electronic information proving to be discoverable in civil litigation, is there any reason to doubt that the report will need to be produced to opposing counsel?
This is not necessarily a novel question, and lying at its root is the BYO(x) movement. BYOD (Bring Your Own Device) has gotten the most coverage in recent months, with widespread (and justified) alarm over how exactly organizations are supposed to satisfy their legal obligations with respect to data contained on their employees’ personal devices.
The stakes are undoubtedly raised with BYOA (Bring Your Own Analytics), a term that is just now making the rounds but will likely gain importance as do-it-yourself analytics platforms continue to proliferate. Here, the risks compound. An employee who, two years down the road, purges the migraine widget report without knowing any better may lead to the company facing accusations of spoliation in the future.
The key point: analytic output data, like all enterprise data, must be centrally managed. Knowledge workers who interface directly with data at its inception are normally experts at using that data for their immediate business purposes. To expect them to simultaneously account for long-term, enterprise-wide information governance needs is ultimately unrealistic.
The logic set forth here is, again, not new. These concerns arise each and every time a new type of data enters the corporate sphere, and we can all rest assured that further types will come to the forefront. Indeed, the Internet of Things will accelerate this novelty along dimensions that are not even fully understood yet.
A unified approach to information governance—where data is centrally stored and retention and disposition policies for all types of data are centrally set by teams whose mission it is to enact such policies for the corporation’s long-term benefit—is the only failsafe way to reliably accommodate and account for existing and future data. Indeed, a significant collateral gain from properly managed data is that it can, itself, be utilized for more meaningful, robust, broad analytics initiatives.
One might put it like so: manage once, leverage forever. Unlike that ugly sweater you never asked for, you can’t re-gift or exchange unfavorable analytics results; you need to plan to manage them instead.