Here I am, seated next to Hamilton, the pig, here at Bam Bams restaurant, a fine barbecue restaurant. In fact, one of the finest west of the Mississippi, at least in my opinion. It's a relatively small company, but it faces issues that apply to every company, large or small. Using point of sale data as well as data that they gathered manually, they used data analytic tools to make tactical and strategic decisions. Let's learn more from a member of their finance team. Cardone: I'm obviously knowing where sales are at, drives our whole weekly decision-making here. You can get super finite to the number of napkins you need to place out, but I mean we're not really worried about that. But we do have to understand where sales are at in order for us to make proper prepping decisions. So, we have to know how much meat to prepare, how much brisket to prepare each day, how much ribs to cook, all those things. So, understanding or knowing where sales are at or expected to be per day drives a lot of preparation work that we do here. So, I mean to quickly answer the question, that's extremely helpful. So, we actually worked with the data that we had from our point of sale system and created an analytics dashboard for us to be able to help us predict and understand where sales are going to be at. So, Bam Bams has been around for six years now so we have a lot of data to really base some of those decisions off of. But so, that helps drive some of those decisions as far as predicting where sales are out, but we also have to understand and use our own intuition. Our managers have to know what today is really going to look like. So, we can predict basically what a Monday in August is going to look like, and sales are going to be about $3,000 or something like that, but the managers have to make their own decisions as far as well. What the data doesn't know is that today is a big football game here at the local university, and so sales are not going to be a typical Monday in August. They're going to be two or three times greater. So, you have to use the data that we have, but also apply the managerial intuition to know really what sales are going to look like. So, you have to bounce between both the data and intuition of the manager to correctly predict where sales are going to be at. So, an ideal customer for Bam Bams has evolved, really. When we first started and open the door six years ago, we had an assumption of what that might look like. We thought, okay it's barbecue, it's meats, we're definitely be strong like the male demographic. We're probably going to have a bunch of 18 to 45 year old man as our customers. Although that was true, that was just an assumption. We really didn't know what to expect. But what we've learned over the past couple of years and with the help of looking at our data is that although it's not necessarily demographically based, it's become more behavioral-based. Meaning that although our customer might look or fit that demographic, that's not really what's driving what an ideal customers for us. What's driving an ideal customer for us is someone who is what we call a loyal customer. That someone who's coming back frequently between two to three times every six months, and that they're spending on average higher than what the normal customer looks like on an average ticket. So, we thought that the business world, doing catering for businesses and stuff like that, and new customers was really going to be our ideal customer, and it's going to be big money, whatever. But what we've learned is that once we get a new customer, we really just have to treat them well, make sure they have great experience, and give them a reason to come back. Once we have converted them from the new customer segments to a loyal customer segment, that's when we win. We live and die by our cost of goods. Especially being a barbecue joint, meat is our number one biggest cost of goods. So, managing effectively sinks and can destroy us. So but watching those numbers is crucial for us. So, it was interesting when we first opened up, we knew exactly what our optimal cost of goods number should be, it was X. Then after so many months, we can see that our actual costs of goods number, the variance between those two numbers was like 10 percent, 10 to 12 percent. There's an acceptable level of variance there, that just naturally happens, and that's between three percent, give or take. So, we looked at the disparity there between three percent and 12 where we actually were, and we knew something has gone wrong. So, the first thing that we did actually was we have scales where we measure how much meat is cut, and what we should be selling, and we charged by the half pounds. What we're realizing is that our cutters were cutting a certain way and they were eyeballing it, and then the cashier would just bringing it up as, "Hey, this is just a half pound." So, what we did was we started using the scales that we actually had. They were just there, but we weren't really using them. So, first thing was get the scales out and start measuring accurately. So, when we started cutting and it was, "Okay, this isn't 0.50 exactly, this is 0.53." We started charging people what it actually was and we immediately saw our variance plummet. The next level for us in measuring our variance was actually looking at our analytics dashboard. So, we're pulling all the point-of-sale data, and so we knew how much meat we would start preparing every day, how much was actually cut and sold, and what the difference was. So, what we are able to see was the difference in variances by each employee. So, we overlay that with which employees were working, which ones were the cutters, which ones were the cashiers, and we could see where the problematic employees were, essentially. So we knew that, "Hey, Annie, she's great and we love her, but there's a lot of variance in her meat cutting, so maybe we need to put her in a different position." We don't want to lose her, but we need to put her in a different position because we just can't deal with that variance. Well, what's funny is we actually found that Cam, who was the main pit boss and owner. He had the greatest variance of everybody here, and we showed him that. He's like, "Yeah, that actually makes sense." He's like, "That's why I don't work the line anymore, because I'm just too generous with people." So, but allowing to see that small variance made a huge impact on our cost of goods numbers in managing those. Guymon: In this module, I'll guide you through a mini case study that will illustrate the first three parts of the Fact Model, but I'll focus on the calculations part of the Fact Model. First, you will perform a correlation analysis to identify two-way relationships, and analyze correlations using a correlation matrix and scatter plots. You will then build on your knowledge of correlations and learn how to perform regression analysis in Excel. Regression analysis is the workhorse of Machine Learning. It's used to estimate parameters for linear models. Those linear models can then be used for making forecasts and for evaluating complex relationships that are difficult to visualize. Finally, you will learn how to interpret and evaluate the diagnostic metrics and plots of a regression analysis. These diagnostics have many purposes. So I want to focus on explaining their basic nature, and then on showing you how they can be used to answer business questions. Recall the Gartner Analytic Ascendancy Model. It places more value on analytics that can predict what will happen. Regression analysis is a really useful tool for creating a way to model the future. Thus by the end of this module, you'll be able to start implementing some predictive analytic techniques. With that, let's go ahead and get started.
