Progressive Insurance: Fast experimentation and low risk with Hadoop
Progressive is one of the largest providers of car insurance in the United States. The company uses data in all areas of the company, from quotes and claims, to analyzing driver behavior. In video 1, Brian Durkin, Innovation Strategist, shares how Progressive chose Tableau over traditional tools, because it allows for rapid testing on large quantities of data stored in Excel, Microsoft SQL Server, Oracle, and Hortonworks Hadoop. With Tableau’s native data connectors, the data science team can bring all of this data into one view. In video 2, Brian talks about Progressive’s Snapshot® program, which allows drivers to save money through safe driving behavior. The data science team uses Tableau to analyze Snapshot data along with external sources—helping Progressive understand how different conditions affect driving patterns.
Exploring big data in Hadoop
Tableau: How would you describe Progressive? Brian Durkin, Innovation Strategist: We're a data-driven company. So all of our leaders, our senior leaders, and all our analysts, and everybody in the company, really, they want to use data to inform their decisions. And so we've been, traditionally, a very hands-on type of company where people are using the Desktop version. Tableau: Who views Tableau dashboards at Progressive? Brian: Tableau is used pretty much across the board within Progressive. Our customers are everybody from our individual internal analysts up through our senior leaders that use the reports, and they use Tableau itself in order to analyze the data. So it's pretty widespread across the company. Tableau: How exactly do you work with data? And what types of sources of data do you have? Brian: Our marketing department will bring in data, we'll compare that against data from our policy servicing department. We'll also look at claims, we'll look at quotes, we'll look at all these different areas to mash up data. We use Tableau to do a lot of that. The technology used to store these things is also pretty diverse. So there's a lot of Excel, there's a lot of SQL Server, there's Oracle, a lot of DB2, and then all the new big data stuff. We store a lot of data in [Hortonworks] Hadoop. And Tableau, with all the connectors that it has, allows us to bring this data together and look at it in new different ways. Tableau: What’s something Tableau makes possible or easier? Brian: Sometimes when you're using Hadoop to do data exploration, your datasets can be so extremely large that you need new tools in order to really analyze that data effectively. So we use Tableau and do subset sampling to do data exploration. A lot of times, we'll use it to analyze the aggregate data. But it provides us a way to rapidly test data and look for new things. With some of our traditional tools, it takes a little bit of work to set up your data analyses and figure out really what you're looking for there. Tableau allows us to rapidly iterate over that and do this rapid hypothesis testing to look for new things. And that's important when you're dealing with these really large datasets. You need to be able to change your views and the way you're analyzing it pretty quickly. Tableau allows us to do that. Tableau: What’s the main value of using Tableau? Brian: I think it's somewhat hard to quantify the savings or the value that Tableau brings. A lot of times there's things we couldn't even do previously, using our legacy tool set. Tableau allows us to do these things now. And so comparing what you couldn't do at all to what is possible now, it's huge. It's night and day. Tableau: How has Tableau changed the way you analyze data? Brian: Being able to have the technology to analyze data, to fiddle with the data, to experiment with it, to play with it and just look at it in different ways has been amazing. A lot of our traditional tools don't allow us to do this rapid experimentation and exploration. Tableau's a great tool that allows you to do that, just to see what you have and analyze it, and figure out what you can do with it.
Using data analysis to help drivers stay safe
Tableau: How important is data to Progressive? Brian: At Progressive, our company is built on data. That's our product. So we don't build widgets or cars or anything. Data is what we do. And so we're a data-centric company through and through. So finding the best data tools and technology to really leverage our key asset, what we have, anything there gives us a huge advantage. Tableau: What is Snapshot? Brian: Snapshot is a program at Progressive that allows our drivers to prove that they're safe drivers, and they can save money when they do it. Tableau: How do you analyze Snapshot data? Brian: Our data science team that's responsible for the new scoring algorithms and Snapshot is always looking for the next big thing, the next thing that they can use to score driving behavior. So they bring in all sorts of different external datasets, external to our company—weather, traffic, and even datasets within different parts of our organization—and join those up and look at them in new and interesting ways to see what shakes out of it. They don't always know what they're looking for in the first place. They want to improve our models and make things better. But sometimes there's the surprises when you just join this data and you start to do these data exploration experiments. And Tableau and some of our big data systems really allow us to do that rapid hypothesis testing, trying out new ideas and just look what shakes out of the data.
Learn more about using Hadoop with Tableau
Want to learn more about how Tableau and Hadoop can work together? To find more case studies, user stories, news and info, visit our Hadoop resource page.