 So one of the things that's been really interesting for me lately is I've had a lot of leaders from company who aren't necessarily trained in analytics asking, how do I manage analytics problems? How do I get them started? And after being asked about this a number of times, we decided to write a book called Mining Your Own Business. And that is really trying to help executives and managers of analytic projects understand really what's behind analytics. And my premise in the book is that the gap that we experience between the promise of analytics transforming a company and results that fall below that expectation are really a failure of leadership and leadership understanding really what analytics is. This morning we had a great presentation by BMW and one of the things they talked about was the evolution of the car over a hundred years. You know, I think a hundred years ago they would have never imagined the I-8. But sometimes in analytics projects we want the whole Death Star from the very beginning. We want to try to define the end point when really what we need to do is just get started to find something initially and then as we go along we can evolve that into what we need for our business as we understand our business better because the thing is our businesses are changing around us, our consumers are changing around us as well. So we need to be flexible and adaptable to that. So that's one of the points I'd like to make. The second point I'd like to make is a lot of times managers because they have two executives, they let their intuition get ahead of the perception and the perception is what we might see in the data. Intuition is our experience behind that. There's a funny story we were working with a telco company trying to help them understand their churn. They were getting a lot of churn in their model and they believed it was because they hadn't offered the iPhone. Everybody needs to offer the iPhone. So we built a churn model for them and we deployed it into their call center and sure enough their churn came down as we expected it would based on the model. And everything was great. They were actually paid the model back within the first couple of months, the cost of actually implementing that model. But then they were able to offer the iPhone about a year later and as soon as they offered the iPhone for some unknown reason they quit using the churn model. And what happened? Churn came back. There was a belief at the executive level that the iPhone was the cause of their original churn. But if you think about it, we modeled that without having any knowledge of the iPhone and were able to reduce their churn. And so when they realized that the iPhone wasn't really the root cause of their churn, there were other reasons people were churning. They reinstituted the churn model and that brought back their churn down and helped the business. And so it's sometimes these things that we believe that aren't actually born out by the data that we have to let the data sort of influence our intuition instead of the other way around. And John Elder talked about that a little bit this morning. But finally, the first and foremost problem that we see is that people start out in the wrong way. We tend to want to see things getting done. And we believe getting things done is actually looking at the data and starting to build a model. But many times what we find is the executives or the managers haven't defined the business opportunity well enough yet. And if we don't define the business opportunity well enough, then the model we're building at the end of the day may not actually meet that business opportunity, may not actually cover what we thought it should cover because we hadn't taken the time to do that. And we hadn't taken the time to figure out if we implement this model, any time you implement a model, it's going to impact some business decision in the business. And to do that, that's going to affect some process in the business. So the idea of change management, how do we actually build this into a deployable thing that people use every day and make decisions on? Sometimes that thought isn't there either. People just get excited. They hear something like deep learning and they say, I got to have some deep learning without really truly understanding what deep learning is or how it might impact their business. And those key questions that you need to answer upfront can make all the difference in the success of the model and how it transforms your business.