All right, well this session is designed to provide you with an overview of the course and an introduction to some of the techniques that we're going to rely on when analyzing marketing data specifically using Excel to conduct that analysis. And if we think about what you should be getting out of this class, most businesses are dealing with data in some form that they're collecting internally, so part of this is understanding, how do we organize that data? How do we describe that data? What are the techniques that we can use, both numerical and visual techniques? How do we understand the relationships that exist between different pieces of data? And there's a process that we're going to go through and I'll try to draw the distinction as sharply as I can between data and information. When we're talking about data, that's the raw form that it's coming in. When were talking about information, there's been some analysis all ready applied to the raw data and it's the information that's going to allow us to extract insights and drive business decisions forward. And ultimately, that's what we're trying get to, as we start with the data, we apply the appropriate analysis, we relate the findings to the decisions that we have to make, and we start that process over again with our new findings. So in this class, what we're going to be looking at, using Excel as our primary tool. There's a lot of powerful capabilities built into Excel, we're not going to be getting into any visual basic or using macros in this course, but we're just going to be using the native capabilities of Excel to understand some common marketing decisions that you may be facing. All right, just to give you some examples of where data analysis is going to show up in marketing. More and more, we're seeing targeting happening of consumers and this has been going on for a while, nothing new compared to the direct marketing that was happening 20, 30, 40 years ago, the only difference is that the data that's available to us is a lot more detailed. Think of every time you click on a webpage, every time you go to a store and you use a loyalty card, your purchase behavior, that's being captured. Well, if we're capturing that data it allows us to learn about you as a customer. We can form customer segments, not just based on demographics, but based on your preferences and we can even go more detailed than that and get to the idea of one to one marketing. In some other context, whether it be higher education or the performing arts, we're dealing with developing our donor base. So who are the individuals who are likely to give to us based on what we know about them so far? Who are the individuals who have revealed themselves to not only have the means to give, but also the interest to give? And who are those individuals who may not have the means quite yet, but are individuals that we should keep an eye out for because down the road, they're on the path to building a strong relationship with an organization and may have the means to become strong contributors in the future. Same idea is going to apply if we're looking at a financial service provider. We might use analytics to understand where I'm going to be allocating my resources, we can use past data to try to predict future demand for products and services, so all of the decisions that a business may face. We're going to try to use the data that's available to us, that's being collected internally or acquired externally, apply the appropriate analytic tools to that, relate it to the business decision, and then go about figuring out what's the best decision that we can make based on the model that we've developed? And so I'm going to lay out just a general decision making framework that starts with the data collection process, all the way through the analysis. And it's one that we'll refer back to over the course of this class. So to start out, we need to understand what we're going to be working on. What's the problem that we're facing? What are we trying to solve? What's the data that's going to be available to us? Now in a lot of cases, the exercises that we're going to be going through, we're going to assume we know what the problem is and the data is as given to us. But I can't stress enough the importance of these first two steps because any analysis that we try to undertake is going to be limited by the quality of the data that we have and our understanding of the business problem. So a first step might be to, with the data that you have at your disposal, understand the relationships that exist among the variables in the data itself. And the terminology of variables, data, field, we'll come back to this a little bit later on in this session just to clarify and make sure that we're all on the same page. But a common example might be to say, let's understand the relationship between price and sales. Well, we would expect that the higher the price we charge typically sales go down. If we think about the relationship between advertising in sales, that's one where we might expect to see a positive relationship, but it may start to slow down at some point. Maybe the first couple of ads, I start to see a big response in sales, but after you've seen a commercial for maybe for the 100th time, it may not be as effective. So, what's the nature of the relationship? Not just is it increasing or decreasing, but what's the rate of the increase or what's the rate of that decrease? So that's trying to characterize the relationship, the other thing that we've got to understand is how much certainty do we have in that relationship? How much certainty do we have ultimately in the forecasts that we're going to be making? It's not just that we're going to be saying that here is the specific nature of the relationship, it's a single number that's going to be able to summarize everything for us. Well we want to understand, how much variation is there in that estimate? All right the next step, once I understand what those relationships are, we're going to try to build a model. So if I'm trying to build a model with sales I might say, well I know that price hazard is related to sales, I know that advertising is related to sales. So I might build a model where I control price, I control advertising, and I'm trying to forecast sales. So once that model's developed, then we're going to be in a position to test potential alternatives. What if I increase the price by 5%, what's going to happen to sales? What if I increase advertising by 1%, what is the impact of that going to be on sales? What if I change my advertising message? What happens if my competitors take particular actions? If we have that evaluation model where all of the factors that we think affect sales are in fact related to sales, then we can play these scenarios out and figure out, what's our best guess for what the response is going to be and then what action should we ultimately be taking? So let me use an example of where it's pure marketing, but it's a little bit different from the traditional context that you might typically think of. Let's take the case of political advertising. So I've pulled from the 2008 presidential election, just some of the data that's available as far as how much was spent on advertising. And you'll see, there's a lot of money going into political advertising, coming from a lot of different groups. If we focus on the candidates, what are they trying to accomplish? Well why are they out there raising all of this money? Well, they need the money out there to be able to conduct their campaigns, their ground operations, their media campaigns, whether it's television advertising, whether it's online advertising, what they're trying to accomplish is winning an election. Well how are they going to do that? They're going to put a lot of money into the advertising, and quote from the 2012 election where president Obama's compared to a top 100 advertiser, based on the amount of money that was spent on advertising. So this is becoming a space where a lot of resources are being poured into the political campaigns, so we're going to approach it from the same perspective as we would any sales problem. In this case, how do I get the sales or how do I win the votes? Well it's based on the electoral math. So just to give you an example, this was the 2010 electoral map. There were 538 votes available and you see them broken down by different states based on their population, any candidate's goal in this election is to say, well there are 538 votes available, if it's a two way race, I've only got to have more than my competitor, so I need to get the majority. So I need to get 270 votes to win this election. All right, well I want to spend money, I want to put my resources into those actions that are going to generate the biggest impact for me. So how do we go about winning that voter over? Well if we look at those actions that might have an impact on how people vote, maybe we're talking about TV advertising. Maybe we're talking about Internet, maybe we're talking about ground operations, having those local field offices, so having some feet on the ground. Now these are the factors that we can pour money into. Well let's take one of these actions, let's take TV advertising since that's where a lot of money ends up. Well it’s not just saying, I’m going to put money into TV advertising that we’ve got to understand, we’ve got to understand which programs is that money ultimately going to go into. And depending on what program I put that money into, there’s going to be a different cost associated with it. So in order to bear those costs, I need to have a war chest. I need to have funds available. If I want to advertise in college football games, what's the price to advertise in those college football games? How much is it going to cost me to do that? Is that the best program for me to be spending my money in? Well it depends, who's watching college football? Same logic is going to apply to the decision of how much should I be spending online, where should I be opening up these local offices? So there are the actions that we can take, there's a cost associated with those actions, and in order to cover the cost of my actions, I need to do fundraising. And ultimately, we're hoping that our marketing efforts are going to impact people's voting behaviors. Now here's the challenging part. We can have models that tell us, here's the expected impact of TV advertising in college football when I air commercials in battleground states, we've got our best guess and that's what our evaluation model is going to give us for the impact of an additional television commercial. But it's our best guess, it's not saying that this is a sure thing. So there's a lot of uncertainty that's going to come into play in terms of the outcome. What kind of impact does advertising have on voting behavior? Sure, there's that best guess, but there's that plus or minus. Think of it as the margin of error. So that's something that we've got to take into account. There are also going to be factors that we haven't controlled for, factors that we couldn't see coming because there was no reason for us to include those in our model. So the models we build, think of them as schematics. It's designed to provide a representation of what's going to affect in this particular example, voting behavior. The actions that we're taking are expected to have an impact. But there's no way that we can possibly hope to encompass every single variable that might affect an individual's decision to vote. That's going to get lumped into that measure of uncertainty that we're not going to be able to predict.
