For years, creating and fostering a culture, where business users make increasingly better decisions, through the capture and analysis of meaningful data, has been the holy grail of enterprises. The pursuit of this goal has seen us create sophisticated systems that capture, store, process and analyze every data point we can get our hands on, with increasing accuracy and speed.
However, for all the wonderful work done by technology vendors and their buyers, only the last part of the statement I described above has been resolved: the capture and analysis of meaningful data. There are, however, two more parts to this equation, and no one seems to be reasonably addressing those: the creating and fostering culture bit and the making increasingly better decisions bit. In this piece, I’m going to take a whack at why, despite all our time and money spent in implementing our systems and educating our business users, we still tend to turn off that analytics tool and go with the magic 8-ball of our heuristics and biases.
To be sure, I am the business user. I am the one who has been fed BI tools for my career and have summarily ignored them in favor of my gut feel—despite all the research, all the thought-leaders and all my bosses asking me to be increasingly data-literate. So why have I ignored their advice and spent time nurturing my instinct, instead of asking the all-knowing, mysterious Oracle (wink, wink) who apparently knows the answer to everything?
I found the answer to this while exploring a related area of interest: the study of how humans make decisions. After all, isn’t our goal to improve the quality of our human decisions?
An Analytical Problem
The result of several studies into human decision-making all come to different versions of the same conclusion: Most of us are apparently incapable of making consistently rational decisions. And the more external variables (different stresses, time, value, consequences) you throw in, the easier it becomes to throw rationality out the window. The weight of our experience, expectations, preferences and biases weighs heavily on us when we make critical business decisions.
And when I, the business user, open any analytics tool, this is what I think:
- Numbers/Graphs are not my natural language. Graphs were invented as a visual aid. They were supposed to add visual representation to a context set in natural language, not replace natural language altogether. Therefore, it’s kind of a big leap to expect me to derive meaning and insight from an X-Y axis without context.
- Are these results even relevant? When I open the BI tool, I do so to receive a specific answer to a specific question. Often, it is to answer a question my boss will likely ask me in my next meeting. All I want is an understanding of that question—How has my latest campaign performed? How many page visits converted into registrations? What is my forecast for the campaign performance going forward? When I am frantically looking for these specific answers, it is borderline annoying to find that none of these complicated charts easily relays the results I am looking for.
The fallout of these factors is a paralysis in our decision-making abilities.
- Too Many Stimuli. In most circumstances, being presented with multiple, often conflicting, options confounds our ability to arrive at a decision. In 2000, a study conducted by Sheena Iyengar and Mark Lepper showcased that. In the study, one group of buyers was presented with a choice between multiple variants of jam (24 in total). The abundance of options stymied their ability to purchase one. Another group, given a much smaller selection of 6 jams, saw a dramatic increase in the probability of a purchase.
- Bias for Status Quo. When presented with too many, too conflicting or too varied options/inputs, the human brain tends to prefer the status quo. A study by Samuelson and Zeckhauser suggests that we naturally tend to prefer to continue doing the same or nothing at all. Our inherent bias of loss aversion causes us to prefer the status quo over taking a decision that might have potentially good or bad outcomes.
- Elimination Over Selection. The other unsavory response elicited by a lack of clear direction is our preference to eliminate options, rather than select options. The problem is that we usually tend to eliminate based on our biases, using statements such as “this doesn’t feel right.” Our criteria for elimination is often informed by a heuristic sense of how we see the world rather than how the world really works.
And I haven’t even reached the point where exhaustion kicks in. Making important decisions all day, every day is hard. So, while “data-driven” went out the window a long time ago, now, so has heuristics.
With the specter of cognitive dissonance and the constant fear of having taken the wrong decision looming large, even when all the decisive tools were supposedly available to you, it is not inconceivable that after a point, we literally start flipping coins in our heads and making decisions.
This is what I’m calling “para-lytics”—the sense of complete helplessness caused by existing analytics systems, that were supposed to help us in the first place.
Is There a Solution?
The short answer is yes, and it starts with improving the way in which analytics applications deliver information to us.
- Alphanumeric Presentation, Not Graphs. We are most accustomed to consuming information through alphanumeric text fields. There’s an argument to be made that without numbers and charts, text seems much less credible, but that doesn’t make such reports less valuable. I propose we put the numbers and charts at the back—a drilldown option if you will—for users to go explore the credibility of the information presented rather than as the default mode of presentation.
- Answers, Not Options. Often, when we go snooping around for answers, we hit a roadblock when we are presented with a cut the red or the blue wire situation. At this point, if the system were smarter it would naturally whisper in our ear “cut the blue one.” Seems crazy? Well, take the example of Google whose Google Answers tool now actually answers a question you ask it rather than just presenting a series of search results. That’s where AI and Machine Learning are helping: to develop systems that have a sense of the answer based on the data, rather than simply doing the math and leaving the rest to us.
- Nudges, Not Lists. Enough psychology researchers have stressed the importance of a nudge— literally a slight push in the right direction. The current style of information presentation is a never-ending series of lists, but nudges serve as an additional aid to decision-making. What kind of nudge, you ask? Simple things like – Hey John, your sales have fallen 8% MoM. Would you like to do X?
We can go on about the need for us business users to be data-driven because that’s the rational thing to do, but the fact is we aren’t rational! We are always going to be limited by our human constraints. The way out, then, is to develop systems that complement our thought processes, rather than compound our irrationalities.