Recognizing the Downside of Data

Recognizing the Downside of Data

Ever since the concept of “big data” broke into the collective consciousness of business and culture in 2011, managers and decision makers of all types have felt compelled to have analysis available to inform their activities and actions.

However, despite the vast interest in having analytics to drive actions, the past six years have not necessarily seen an evolution toward a more rational and analytically driven populace or institutions. In fact, while some truly analytically driven ventures have emerged from the ether, it could be argued that over the last six years, many efforts at analysis have created as much disappointment as they have clear direction.

Recently, with Brexit and the 2016 US Presidential election, we have seen failures in very high profile analysis by experts and pundits. At the same time, in everyday discourse, it seems that the power of data (in the form of verified facts) is often diminished in exchange for opinions and misinformation (unverified or untrue statements presented and willfully perceived as fact) that better align with and thus further cement preconceived perspectives on the world.

To explore why we might be achieving less than expected returns from efforts at building data-driven understandings of the world (from good, verified data) while seeing expanding “filter bubbles” inflated through confirmation bias, it might make sense to consider the nature of this thing called “data” that we’ve been so fixated on for the past six years.

The fundamental unit of data is a bit, represented as ‘1’ (is) or ‘0’ (is not). What makes up “big data” is the number of bits (measured as bytes) we are able to maintain and scale while allowing high computational performance.

What has not been stated yet, because it is not really an aspect of data, is what the bits are about. The understanding of what the 0s and 1s mean, where they come from, and what they can be used for is a step beyond data, it is information.

In essence, data is the presence of a coin with one side facing up, while information is recognition of the side as a ‘head’ or a ‘tail’ along with any meaning assigned to each.

Lots can be done with data and information. It can be visualized in tables, charts and graphs to allow for the evaluation of patterns. But when pattern evaluation is conducted by humans, we can often read what we want into the data – or conversely miss certain patterns, so it is better to statistically evaluate patterns and transform them into predictions. But even statistically valid predictions are only estimations of what might happen based on models or theories of how the world works.

Sometimes, we do not have the right information to begin answering questions that explain our world and guide our decisions in it. Sometimes, we possess what seems like enough data or information, but its processing through modeling and analysis still does not yield answers that effectively address the questions we want answered. Or sometimes the data and analysis does give answers, but they’re not the answers that were wanted or expected given our prior understanding of how things work.

These are the situations that create disappointment and lead people to raise opinion to the level of fact; when analysis of the facts does not give us what we believe to be “actionable insights” to answer our questions and address our problems as we understand them.

This leads us to a third pillar of data analysis – the one that has been neglected through our infatuation with data itself. The first pillar is the data, and the second is information that gives meaning to the data.

The third pillar is knowledge. If disappointment with data/information/facts comes from not receiving what are felt to be actionable answers to what are considered key questions, then knowledge allows us to understand if the issue is with the data, or with the kinds of questions we’re willing to ask, or with the kinds of actions we are willing to take.  Knowledge informs the type of data we collect and the effort we put into its collection. Knowledge of how the world might work (via observation and theory) inform the models that process our data.  Self-knowledge lets us evaluate the patterns we already have established about the world (heuristics) that will either be confirmed or challenged by new information, which will in turn inform the answers we’re willing to seek and hear.

Data can augment knowledge with great impact, but data & analysis without knowledge will yield indifference at best, and at worst, may potentially be establishing a backlash against data-driven decisions and an inflated valuation of personal opinions or established beliefs about the workings of the world as being equal to verifiable facts.

Posted in CRM