How Would you Characterize your Relationship with Data?

How Would you Characterize your Relationship with Data?

Eugene Lee

Blogpost Image

“She is frequently kind And she's suddenly cruel She can do as she pleases She's nobody's fool But she can't be convicted She's earned her degree…”

  • Billy Joel, American singer/songwriter

Databases have been called “relational” but if data were a person, I would describe my relationship with her as “complicated.” On the one hand, I love data. Favorable data helps me justify my decisions and vindicates me in the face of doubt. On the other hand, she can be fickle or even capricious. Why does she only give me partial answers and seemingly disappear when I need her most? To some extent, I think we all share this love-hate relationship with data. Some days, she turns us into heroes. Other days, she betrays us with numbers taken out of context. This dichotomy of experience leads us to approach data cautiously: we are eager to harness her power but fearful of getting burned.

We probably should start by admitting that while we’re infatuated with data, we don’t really know her very well. We’re used to making our business decisions on a little data and a lot of instinct. To many of us, the world of data is foreign and exotic. For me, I didn’t own my first computer until I was 20. Today more data crosses the internet every second than was stored on the entire internet when I got that computer.[1] Since we didn’t grow up with data, it’s going take us time to get used to incorporating her effectively into our decision making.

Yet we find it hard to trust data because it seems she has no sense of loyalty. Everybody thinks they have her on their side. It doesn’t matter if the dispute is politics, sports, or business; we are always right and we have the data to prove it. Different constituents can even look at the exact same numbers and draw very different conclusions. We are inundated with examples of this phenomenon every election year.

Perhaps our unfamiliarity with data has led us to unrealistic expectations of what she can do for us. Data does not concern herself with right and wrong. She only answers questions of fact: “What is my loss ratio?”, “What is my policy retention rate?”, and “How many claims did I have last year if I exclude claims >$50K?” But these facts only get us halfway to our goal. Because we don’t just want to know what our loss ratio was, we want to know if that number was any good. And this is where data cannot help us. Sure, more data can give us more context, but in the end questions of good versus bad require human judgment. Expecting data to answer questions of morality is to be disappointed – which has happened to all of us.

We can take comfort that this path has been trodden before. Over five hundred years ago Gutenberg’s printing press significantly reduced the cost of producing books. Before the printing press, books were rare– the domain of monks and scribes. Fifty short years after Gutenberg, an estimated twenty million volumes had been printed[2], democratizing knowledge for a growing and literate middle class. The parallels to present day are fascinating to consider. What will the world look like fifty years from now? Nobody knows but it’s a good bet that data literacy will be a functional requirement.

Data literacy, in our industry-specific way, is what Guidewire Live is all about. The common technology platform and data model shared by all Guidewire customers makes community benchmarks and curated external content available in real time, all with no implementation project or upfront risk. We think it’s an exciting time to be working in insurance. At the same time, a change of this magnitude can also be the source of some anxiety. “Can my organization absorb all this data? How can I be sure it won’t be misinterpreted or misused?” These are valid concerns. As in any relationship, the process of getting to know data better is going to take time and will have some rough spots. But I think we can all agree: better to have data as my friend than as my enemy.

As published by the Insurance-Canada.ca blog, The Intersection, February 14, 2013.

http://hbr.org/2012/10/big-data-the-management-revolution/ar/1

Febvre, Lucien; Martin, Henri-Jean (1976): "The Coming of the Book: The Impact of Printing 1450-1800", London: New Left Book

Tags