“Big Data” is all around us, and it seems everywhere you turn, data is continuously being collected about everything, from what all the machines and devices in our world are doing at any given point, to what each person on the planet is talking about.
We collect data to monitor for and resolve problems as soon as they surface. We collect data to understand trends from weather patterns, to traffic, to stock prices. And, we collect data in an effort to understand ourselves better, to lead a healthier lifestyle or figure out our likes and dislikes, so we can spend more time and money on the former, and less on the latter.
But, there is a fundamental difference between the collection of process data and human data. Data generated from measurement of physical processes, such as the outside temperature rising to 70 degrees, or a computer server fan overheating, offers few options for analysis. We must simply capture it as best we can and throw all the tools in our arsenal at it later to derive some meaning or predictive power. In this realm, most analysts would agree that the vast majority of data collected is garbage, or at least uninteresting, containing no actionable patterns. Even worse, in some cases, this garbage can actually obscure hidden gems scattered underneath. Yet, devoid of a better option, we still collect as much as we can, and clean it up later. In other words, with “Big Data” comes “Bigger Garbage.”
However, when we try to use data to understanding human beings and their behavior, it suddenly becomes clear that, unlike measurable processes, humans are not immutable. They can actually be influenced to alter their behavior over time. They respond to and adjust to their environment, a characteristic that is inseparably linked to our psychology. On the one hand, this wreaks havoc on statistical analysis, since the subject of study is changing as it is being studied.
But on the other hand, understanding this fundamental limitation can actually be liberating. What if we could leverage modern understanding of human psychology (and perhaps also sociology) to build products that enable clean self-identification or self-segmentation of the user base? What if they impose a natural timeline that users will follow without friction?
In studying human behavior, we generally aim to learn several things:
- How do human beings use a product or a service?
- What makes them use it more or less?
- Why do some users respond to a set of incentives, while others completely ignore them?
- Who are the users we should focus on, and who can we ignore?
- Who should we learn from, and who should we apply that knowledge to?
The world of human beings is a collection of distributions — no two are the same, but we share many attributes with others like us, while we have very little in common with others. And, any product out there is targeting some shared attributes of certain groups of people.
When you consider all the shared traits of all the various segments of people you could be targeting, and add in the reality that these can all change over time, an already complicated problem becomes almost impossible. We have segmentation and regression problems intertwined.
Enter gamification. The basic idea behind gamification is to leverage understanding of human psychology in a data-supported iterative design, to combine human desires for competition or collaboration, achievement, recognition, or exclusive privileges into a predefined natural flow. Gamification allows users to self-segment as they progress along a set of predetermined paths, while we as providers of the game use their data as feedback and deploy various rewards, loyalty or social mechanics as feedback to them.
In Gaming Systems, User Experiences Build Meaning
To do this, we must first identify the most actionable segments along the way at the appropriate time and in the appropriate context, whether it be a group of new users struggling with getting started, or a set of power users needing some extra attention. It’s a much simpler problem, if you already know who those users are, instead of having to deduce them from the data. This is easy to do if these people are awarded levels or rewards as they progress, or if their experience is incorporated into missions and tracks.
It’s a different paradigm in a lot of ways, and it seems ironic that a system which generates additional throngs of data can actually help resolve a data volume management challenge. But the key here is not just in managing “Big Data,” it’s in working with smart data. That’s because gamification does not simply add more data, it adds information to the data you already collected.
Yes, we are casting aside every scientist’s traditional and time-honored techniques for randomized samples, control groups and the performance of unbiased analysis. But, if we realize that human beings are not ruled by static laws of physics, it becomes obvious those concepts often do not apply anyway. If we abandon the notion of a fundamental truth or rationale behind human behavior, we are then free to embrace the idea of a constantly iterative and interactive guided experience.
After all, in the business world, what matters most is not whether we biased our analysis and got the result that we wanted, but whether it actually works in a practical application. Or more importantly, whether it will work next time. Gamification supports this predictive process when it comes to erratic human behavior better than any traditional scientific analysis ever could.
Tim Piatenko is manager of analytics at Badgeville, a gamification platform, where he works on user analytics for social gaming and engagement. He earned a Ph.D. in physics at the California Institute of Technology, and has held senior analyst positions at eBay, Like.com, and Zynga.