Hello, I'm Chris Ittner, the EY Professor of Accounting at Wharton where I specialize in management accounting and performance measurement issues. We will talk about today's linking non-financial metrics to financial performance. There's a lot of reasons why this is an important fact. This is probably one of the most important things we do as managers. We need to forecast future financial performance. But if you think about it, the actions you take are non-financials. Am I going to invest money in R&D, am I going to do a marketing issue, am I going to improve my quality, am I going to introduce a new product? So, what we want to know is if we take these non-financial actions, how do we think that's ultimately going to impact our financial performance. So we need to know this for forecasting future cash flows. We need to know what we're going to choose projects. Of the many choices I have of improving things, do I focus on improving customer satisfaction or I focus on improving the quality of my product. Well, both of those are non-financial. What I want to know is when I improve this non-financial dimension, what's the ultimate impact going to be on my financial performance? And obviously one of the big reasons we care about this is you're going to get evaluated on these performance measures. A lot of company has been putting in dashboards and scorecards. They're everywhere for profit, nonprofit. The idea there is a lot of the actions that you take as a manager, do not show up in short-term financial results for a while. There is a lag or even worse they actually make your short-term financial performance get worse. Because under accounting rules, if you spend money on research and development or building the brand of your company, it actually hurts your short-term financial performance, even if in the long run, this is a good thing for your organization. So what we need to know is given the many non-financial performance measures we could use, which one of those actually tell us something we care about which is which one of those, actually say, if we make this measure go up and down, it actually makes our financial performance in the future go up. So what we're going to do today is talk about, okay, how can we pick out which non-financial performance measures are actually informative, which ones actually tell us whether they're predictive financial results. How do we set targets? Well, targets are easy for financial numbers. More is better. I prefer more money. It's not so simple when you start talking about non-financials. How happy do you want your employees to be, how satisfied you are at your customers? I could have the most satisfied customers in the world. I know how to do that. Take a box, pack it full of money. The amount of money is less than what you pay for it. Guess what, you're completely satisfied and your company is out of business. So, that's what we're going to do is let's go through how do you pick these non-financial measures using analytics to say which ones when they move, actually improve financial performance, then how do I pick the targets for these, how happy do I want my employees, what kind of employee turnover do I want? And finally, we're going to get some comprehensive examples of how companies have used this to start coming up with financial models that allow you to say, if I invest money in a non-financial dimension, how do I expect that to pay off in the future? And finally, we're going to finish with some of the organizational issues you really have to worry about. This is not just statistics we're talking about here. We're trying to change the way you run your business and this got to have major implications including things like politics. If you tell somebody that they've been doing something wrong based on your predictive analytics, you better be prepared for pushback. So there's the outline of what we're going to actually do over this session. So here are some fundamental questions we need to ask when we start talking about non-financial measures and leading into financials. One, what should we be measuring? I don't have any problem coming up with non-financial measures. We could sit around the table here. In 15 minutes, we could come up with hundreds. What I want to know is of those hundreds of non-financial measures, which one of those when they move up and down actually tell me something I care about. It's not just a matter of I want non- financials. What we want are what are the key drivers of financial success? Now driver is a very important word here. That means when a manager actually takes his action, it actually leads to this thing I care about on the other side on this, and that's what we want to get with predictive analytics. As I want to predict based on these actions, what is going to drive future financial performance? So that's the first thing is, what are we going to measure? The next thing we have to worry about is how do I rank or weight these things when I have various financial measures? Now some of your companies have something called a balance scorecard. And with a balance scorecard, you have financial measures but you also have customer measures, you have operations or productivity measures, and you have what are called learning and growth measures. Well, that's great. So, you have this balance scorecard that may have dozens of performance measures. How do I know how to rank or weigh these various? Do I really want to spend more time on improving quality or improving customer satisfaction? It's really easy to rank financial numbers because they're all in currency, they're in a common denominator. The problem with non-financial measures is they're not in a common denominator. Some are in percentages, some are in weights, some are in time, they could be in anything. How do I combine those or figure out which one is more important? Now given that, I also have to make trade-offs because it could be that in the short run if I improve one measure, the other one is going to go down, and a good example would be the trade-off between short-term non-financial and short-term financial. Again under accounting rules, if you spend money on research and development, that could be a non-financial, are we doing the research projects we want? In the short run, all the money you spend on research and development makes your short run accounting profits go down. That's how the accounting rules work. Now in the long run obviously, if you think these are good research and development projects, ultimately, you're going to get better financial results. But in the short run there's a trade-off. So what we need to decide is, okay, when is this going to pay off and what kind of trade-off am I willing to make? And finally, again, probably one of the hardest things to do with non-financial measures, what performance targets do we want? More is not necessarily better with these. So, the steps we're going to have to go through again is what should we be measuring, how are we going to weigh these very, very different performance measures when I'm trying to analyze the potential financial results, how do I make trade-offs between short-term and long-term, and finally, how am I going to set these performance targets? Now there's lots of ways we could go about doing this. It's not like companies don't make these decisions. We make these decisions all the time. Now I could do it just based on intuition. I've been in this business a long time. I know this industry. I know our customer base. It must be true that if X goes up, Y is going to happen in the long run. Well, that may be right, but we live in a very dynamic world that maybe that was right last year, and it's not right this year, or maybe your intuition is 90 percent correct. It's that 10 percent that may not be correct that's really going to give you a competitive advantage and we want to understand that. Now I could go beyond that; get past one person's intuition and use management consensus. We get a whole bunch of really smart people to sit around the room and decide, these non-financial measures are going to lead to these financial results. Most of us have been in meetings where consensus really means who's loudest. Okay, whoever's loudest and talks the most, they win. Consensus does not really mean it's necessarily right. Now obviously it's a good place to start. We could use one of these measurement frameworks or benchmarking studies. Use the balance scorecard. You've got four categories. I have to have measures in each one of those that must be true that they lead to financial performance. I could use a benchmarking study. I go out. I hire a consultant. I ask them, "What is best practices?" Let me tell you, this is the worst question I ever get as a consultant. Best practices means you're doing exactly what the other guy is doing. No, what you want to do is, how can I figure out in my company? What non-financial measures I need to make my strategy work? Now, what are other guys using to make their strategy work? We could do informal data analysis. We kind of plot some stuff on Excel spreadsheet and they kind of look like they're correlated with each other. It kind of looks like and this goes up, that one goes up as well. What we're going to talk about today is that, let's get past that. Okay, let's try to do more rigorous predictive analytics methods. The tools are there. Most of our software packages have the capability. As companies say, "We have lots of data, but we have little information." What we're trying to do is turn the data that almost every company has into information or something that I can say, "When this measure goes up, what's going to happen on the other side?" Data is not the problem in companies. Coming up with measures is not a problem. The problem is coming up with data or measures that tell me something about future financial performance. Now there is a theory we can start use to structure this analysis, and it's a good way to think about this. The first thing we have to start out is developing something called a "causal" business model or a strategy map that describes what are the value drivers in your company and how are they linked to strategy. So, here's what you want to think about. What a causal model is is an if-then statement. If this happens then I expect something else to happen. Now what you want to do is start out with your strategy. Pull out your strategic plan. And for those of you who have never seen a strategic plan, they're actually extremely high-level. We're going to move into this market. We're going to go into this product line. We're going to divest. They actually don't tell you how you are going to get there. What a causal model is is start out with a strategy and say, "These are the steps I need to take from A to Z to make this strategy work." First, I've got to do that. Now when I do that, that should lead to this next thing, which should lead to the next thing. There's a cause and effect relationship between putting the strategy in and getting the ultimate financial results. So what you want to do is start out with this causal model. Here's a strategy. Here are the steps I need to take. Once you know those steps, then you can lay out what are called the value propositions or hypotheses. How is it that we as an organization think we're going to create value? Now value, in many for profit companies, is going to be financials. If you're a nonprofit, it doesn't have to be financials. Here's the ultimate goals I want to get. If you think about it a strategy is a hypothesis, here is how I think I will get there. The question is when you implement the strategy, will it work? Well, that's testing the hypotheses. I've got these steps I need to go through. I'm predicting that A happens, then B, then C. Well, let's see if that's true. When I implement A, let's come up with a performance measure. Does it predict B? If B happens, does it predict C? Because what you don't want to do is wait to see if financial results work or not. Am I taking the intermediate steps that I need to do to achieve my financial goals that ultimately lead to financial performance? Because financial performance could be a long way out there, I don't want to wait to see if this works or not. Now that you have that, here are the steps I need. I'm going to come up with performance measures to see if I've done the steps. Let's test those hypotheses. If your causal model again says I need to do A that's going to lead to B, that's going to lead to C, let's test it. When I implement A, let's come up with a measure that says, "Did I implement it well? " Then let's see if there's a relationship. When that measure related to A goes up and down, did the measure related to B go up and down? That's the hypothesis that if I do A, B will change. Now if I don't find that, I want to know why. Well, there's lots of reasons why. Your strategy is wrong. There's no reason strategy has to be right. Two, your measures are garbage. I can measure all kinds of stuff. That doesn't mean it actually predicts anything. You could ask me on a customer satisfaction survey whether I like the color on the wall. I'll answer it, but it doesn't tell you whether I'm going to buy again. It could be there's an organizational barrier. Yes, A happened but B didn't because somebody wasn't doing something they were supposed to do. All of these I want to know. But to do that, first of all I want to see if is there a relationship between A and B. That's the role of analytics. Now based on these results, now I can start coming up with decision making models and performance evaluations. Okay, if the analytics says when A happens, B happens, and here's the magnitude. Well, I can start coming up with a little financial model that says, If I invest $100 in A and it moves that metrics by two, what's going to happen to B? Well, it may move that by five. When B happens and it goes up by five, that may change something else by three. So based on these relationships as you go through the causal model, all you're going to do is build a little mathematical model that says, if I do A, here's ultimately how it's going to translate to improve financial performance on the end. Because financial performance is an endgame. But I need to know these intermediate steps. I need to know how big. If I improve A, what is the impact on B? I could also use this for performance evaluations. Because obviously, if I evaluate you on a performance measure, guess what? you're going to improve it. People will go for whatever you're getting measured on. The question is, when you improve that measure, are you actually making your organization better off or not? Well, I don't know until I can figure out whether this metric, this measure I gave you and your bonus plan or performance evaluation, whether it's actually predictive. So once I find these predictive measures, that's what you want to evaluate employees and managers on, not just any non-financial. The ones that when they move around actually tell me something I care about. And finally, this should be a dynamic tool. Strategies are changing. Performance measures that were good before may not be good anymore. Competitors are coming in. How can we set up an ongoing mechanism that's going to allow us to use these analytics to refine our strategy over time, change or see if our value predictions actually work and keep our measures up to date? Because the last thing you want to be using is a measure that was real predictive two years ago, and the economy has changed so much now, it doesn't predict anything. So somehow we need to set in a mechanism that's going to allow us to do this on an ongoing basis.