A melting pot for “data junkies”

There was a pattern showing in yesterday's many sessions and debates that goes a little further than my thesis that we are in the midst of the moment when data analysis is going mainstream and that it helps enterprises find key differentiators in more and more markets. Erik Brynjolfsson has shown in a study that "data-driven enterprises" generally fare significantly better than the rest. And it was him who more or less spelled out the idea that I have seen validated at this event for a few times by now: when data analysts start working in a new field, they tend to come up with good ideas, even when they have little experience or expertise in the particular industry. Sometimes these ideas lead to little improvements, sometimes they have the potential to transform the whole business. My point is, though, that many of those ideas can inspire similar solutions in other industries. This is why I said yesterday that I want our Universe conference to be the number one event for "data junkies" where our customers get the opportunity to learn from other customers.

Just to give you a striking example from Brynjolfsson's keynote: He and his colleagues figured that house buyers reveal their purchasing interest by surfing the internet, which usually begins with searching Google or other engines. The analysts took these data and succeeded in predicting house sales more precisely than official experts – without having much knowledge of the real-estate business. Brynjolfsson also reported how Android GPS data can be used to predict traffic jams (even if they are made of pedestrians) or spot the latest in-place at the local bar scene. This is an experience shared among many customers: once you've got the visibility, you get the agility, too. For example, Southern California Edison (SCE) was able to respond faster to blackouts caused by a heavy storm, because they could quickly identify households that were cut off thanks to the smart meters they have deployed in millions of households.

One more example from our media roundtable: Stephen Brobst mentioned an insurance company that explored new ways to turn its data into value and found that they provided formidable input for credit risk ratings – after all, someone with loose driving habits is more likely to have loose spending habits as well. The debate during this event was quite lively. Most panellists including Michio Kaku and Mike Breitenbeker of Overstock.com agreed that statistical skills were key for careers in the age of big data, with the exception of data warehousing veteran Barry Devlin who predicted that data growth was actually approaching an inflection point simply because "it has to change". Everyone agreed that "tools make excellent servants and poor masters", as Martin Willcox put it, meaning that the genuine human element in analysis was indispensible. But what exactly defines this human element? No two panellists seemed to agree here. Theoretical physicist Kaku, for example, listed intuition, emotion and humour. Brobst pointed out that science without art leads to stagnation, whereas Devlin said, "we can be fascinated by moving bars – but it's the intent that counts. How I want to run my business, how I want to live!" Maybe, in the end, it's the diversity that creates the human added value.

Hermann Wimmer

"