Get your hands dirty!

OR: "Why analytics without story is worthless"


It has been amply repeated in textbooks and real life that analytics needs to be hypothesis-driven to be successful. That any analytical project without clear guidance of one or several questions is bound to get lost in endless amounts of data and quasi relevant relations. Still, many risk analysts that seek to unveil another insight from data start with what at best can be described as a vague understanding of what they are looking for. The whole problem often comes down to the lack of interest of the small details of each case, and the actual desires, wants and motives behind the behavior. Yes, I'm talking about the fact that it is hard, tedious, low-tech, and tiresome to actually sit down and look at a critical amount of observations/individuals/transactions, but nonetheless necessary. The start of an analysis is not to figure out the statistical/analytical method, before you've tried figure out the human subject.


The knowledge that is amassed among fraud agents, customer care agents or transaction analysts tends to be low valued, since their conclusion is based on experience and cannot easily be reconstructed by someone else (academics) with just knowledge of a certain method. But it's rarely, not to say very, very rare, that any analyst uncover the truth from a data set that makes the agents go. Really? No Way?! We had noooo idea! If analyst succeed in doing so, it is more likely that the analyst did not clean the data enough, made an unrealistic assumption along the way, or any other methodological mistake.


Therefore, true insights about risk and fraud behaviors must always contain the width between micro and macro, i.e. the experience of looking at observations to the more sophisticated quantitative analytical exercises. Ideally, both sets of knowledge exists in the same individual, but in the lack thereof, both should at least exist on the same team. One very simple and effective way of achieving this is to make sure that your analysts spend time with agents and actually analyze cases together as well as their own. And like any good habit like exercising, eating whole-grain and brushing your teeth, it needs to be done regularly to actually be worth something.


On the other side of all that tedious work is not only deeper insights, shorter projects, and stronger results, but also better legitimacy when presenting conclusions and anchoring change initiatives with agents and the rest of the organization.


Get your hands dirty, and everybody wins. Except the fraudster.

Popular Posts