In today's lecture, we're going to focus on the critical topic of how to forecast demand. Now of course, we could literally spend weeks or months going through all kinds of books and concepts. So what I really want to give you is just the essence of some key ideas around why it's so important and some principles that I think are going to help you, as entrepreneurs, as you go through and try and figure out what demand looks like for your new products and services. So we're going to do things in two parts. We're going to start out, how do we sort of overcome some of the biases that causes us to mis-forecast demand? What's a conceptual framework, the ACCORD model, that allows us to think about what might be the barriers to realising demand. And then why is qualitative research so important in the early phases of understanding the demand process? So that'll be our first section of material today. And then secondly, we'll then get into some principles and examples from actual sort of forecasting context. We'll think about demand decomposition, why that's absolutely critical. A few forecasting rules and methods. And then we'll ask the question which is going to relate to a quote in a moment, what if Dwight Eisenhower were wrong? You'll understand what I mean by that in a second. So, the key idea we're going to think about here is how do we forecast demand for products and services where demand may be uncertain? And the reason I'm putting the little warning or danger sign there on the slide is that this is one of the most critical risks that the entrepreneur faces. The risk of misforecasting or misunderstanding what potential demand for the product will service could be. There are other risks that we might have in execution, but if we get this one wrong in the way that we think about the total addressable market. This is potentially catastrophic, not just for the business, but for valuation, for investors, for the whole kind of soup to nuts. So let's keep that in mind. I also want us to focus on an interesting quote by Dwight Eisenhower, president of the United States, is that often times, we think we're coming up with something completely new and there's no way we can really know what the demand will be. But actually, that's not aways true. Of course there are products and services that are totally new to the world, that are discussed in some of the other lectures as part of this course. But many times there'll be clues or signals about what demand might reasonably be, particularly if we approach the questions correctly. So Eisenhower said, things are more like they are now than they ever were before. So let's not be afraid to use existing analogies when we're forecast demand. Now with this in mind, the first thing I'd like you to think about when you're forecasting demand as an entrepreneur for your new product of service, is the problem of projection bias that's ubiquitous in demand forecasting. This is the idea that we're all thinking about demand from our own vantage point and our own experience and sometimes, we need to step back and step away from that. So, what I'm showing here on the screen is a picture of a gentleman outside McDonald's in China. McDonald's is obviously a very successful corporation. But when they went into china they completely misunderstood the way people would use McDonald's as a product and a service. In the United States we tend to come and go pretty quickly we like drive through, we like sort of getting in and getting out. In China the McDonald's was more a place for families to sit and relax and friends and colleagues to enjoy each others company. So by projecting the American experience onto what we thought was going to happen in China led to a large problem in terms of the McDonald's and so forth being way too small and not really conducive to what customers wanted. So one way we can debias or get around this idea of projection bias is to try and literally put our shoes, ourselves in the shoes of the customer. So the picture on the screen here is a lady who's working at Ford and she's trying to understand what it might be like to be someone who is 65 or 70. Someone who is perhaps a little bit larger and how would I be getting in and out of a new Ford product? A new minivan in this case. So literally taking on the physical characteristics of the intended customer to try and get away from this idea of projection bias. And so now I want to share one of my favorite frameworks for thinking about how to understand impediments to demand. This is not the forecast per se but think about all of the things that might facilitate or retard the potential demand that we're going to realize. This framework is called the ACCORD framework and let me lead you through it and then ask you to do a little exercise at the end. So, the first thing you should do as the entrepreneur is list as many things as you possibly can that are the relevant advantages of this product or service that you're bringing to market over the status quo. It has better functionality, it has better value, it looks better, it has better functional, emotional, symbolic value to customers, it's easier for customers to use. What are the things that give it relative advantage? The greater the relative advantage, the greater the potential to realize demand. Secondly, is the product or service that you're introducing, is it built-in and compatible with the existing behaviours of customers? So, if I'm introducing a product like Stevia, which is a sweetener for coffee, people are already used to putting sweeteners in their coffee. I just put another one out there. I'm not asking you to change your behavior. The third thing you should think about is is this an easy product to understand or is it a complex product to communicate in terms of the benefit. The easier it is to communicate what the product is doing, the better off you'll be. If you want to Google something somewhat amusing, you can go to Google and you can type in the phrase will it blend, and you'll find a gentleman there who's blending phones of various manufacturers, I won't name who they are. But what you can see there is, it's a very valid and very visceral way of representing the core value proposition of his product, the Blendtec blender. Because it's so, has such high efficacy in blending things that you normally wouldn't put in to a blender. And the fourth point on the slide relates to my favorite part of the ACCORD process because this is really interesting for the entrepreneur. What is it about the product or service that's observable in terms of consumption that might lead people just to pick up the product by social observation or even word of mouth. So the packaging can be critical here, the unboxing can be critical here. If it's a product that other people can see and use, like those headsets, the Beats headphones? You might remember back even further to when Apple introduced the first iPod. There were the iconic billboards that you could see in the United States and other countries around the world too, of a black silhouette and then a person with white earbuds and headphones coming down to something that's socially observable can be very important. Another point that I want to relate here that goes also with the example I'm going to get you to think about is if you're selling a product or service that's not observable in the act of consumption, like wearing a headset. Then you might want to think about, can I create content that people might be willing to share and that thing creates visibility or observability into the product? You can go to Google and look at this one on your own but the link there is dollarshaveclub. If you go to the website for the Dollar Shave Club you see a very humorous piece of video content for a product that people don't actually observe others using, which is guys shaving their face with a razor. That's not socially observable. I don't have people come to my house and then watch me shave, but because of that funny content I might then share it with others. The next point is low risk of failure, can I adopt your product or service without incurring too much functional risk, this thing hopefully is going to work. Not too much social risk, it's not embarrassing for me to do so. And not too much financial risk. If I buy into your product or service, its not going to be some unintended or negative financial consequence. And then finally, this is particularly important for those of you who are addressing perhaps less developed markets around the world. Can the product be tried or consumed in a relatively low cost or small units. We call that divisibility. So if you can think through this framework and you can check the boxes, really for all seven, that's going to give you a bit of sense of where the demand is going to expand rapidly or whether or not you're going to face impediments, which is, of course, a critical part of forecasting the ultimate demand for your product or service. And so now an example, what I'd like you to think about, to really try and apply this accord model. Just go online and go to the website of Dollar Shave Club. And I'd like you to apply the accord model to that business. So does this product have a relative advantage? Is it compatible with existing behaviors? And so on down and try to come up with a sense of whether or not you think this is a product that potentially can unleash a lot of demand or whether or not there are some negatives there. I'll let you do that on your own but that will give you a sense of how you can apply this framework to your own business. And now, just to conclude this piece about thinking around how we understand whether or not demand is going to be large or small or what the impediments might be I can't emphasize enough, based on my experience with working with many, many entrepreneurs, the importance in the beginning of the demand forecasting process of qualitative low cost research. What do I mean by that? I mean by doing a few simple interviews with potential and existing customers. Looking into online forums about what people are saying about the product category, overall what the level of sentiment is. Running some very early focus groups, even if they're just selected by convenience, where you have people physically experience the product and kind of talk and communicate about it. Are we able to do any simple surveys through the Internet? Very, very low cost to administer via email and other channels. And then finally, can we go onto online communities and just look and see what is being discussed and what's being said about the pain point and the existing products that are already out there. So this requires really a minimal investment in terms of the financial outlay. It'll require you to spend a little bit of time. A lot of what you learn won't necessarily be generalizable on the same way that quantitative research is but very, very important for you to do this early on. Just as part of your sensing process as the entrepreneur whether or not there are impediments or things that we should be looking at to really unleash the total demand potential for a new product or a new service. So the first thing I want to mention here is what we call the Golden Rule of forecasting. And the Golden Rule of forecasting is to be conservative. The entrepreneurial landscape is littered with companies that just had wildly inaccurate projections about what demand was ultimately going to be. Now, of course, in this short video session or lecture together, we can't cover everything. So one resource I want to point you to is to my friend and colleague, Scott Armstrong. Scott has incredibly deep knowledge of forecasting principals and forecasting tools. You can look him up on the Warton website, that's Armstrong, A-R-M-S-T-R-O-N-G and utilize some of the free resources that he has around. Actually forecasting tools. So with that in mind, I want to give you the second most important concept or principle in forecasting which is to decompose the problem, decomposition. Very, very simple. It requires nothing more than an understanding of multiplication. I'm going to give two examples. One, demand for car sharing services Right here in Philadelphia and another one was used by a major package goods company to try and understand demand for toothbrushes in India. So, let's go on to those two examples, thinking about this idea of decomposition taking a larger demand problem and breaking it down into it's individual components. There are two reasons we want to do this. Reason number 1, when we decompose, any errors that we're making in the forecast should tend to cancel each other out, and we'll get an overall more accurate prediction. Academic research clearly shows that. And number 2, if we decompose the demand into its constituent pieces, that will show us the trigger points, all the things that we might want to influence. So let's go now into those two examples and that concept will become clear. So the first example I want to go through together is a car sharing example. Now if you're like me I recently sold my car. Many of you out there might now be thinking about how to share certain resources, like an automobile as opposed to owning one. So this was a very, very important practical problem for some entrepreneurs here in Philadelphia who wanted to start a car sharing service. And of course they wanted to know how large the market could be and how quickly it might grow. And so what they did is they went through a five step process that's incredibly simple but often overlooked by the entrepreneur. So first of all, what's the total addressable population, well it's residents of Philadelphia. Then secondly, of that total addressable population how many people need transport? Of that, how many people have a drivers license? Of that group now d, how many people don't have their own car or their own other way of getting around. And then on that, what fraction of those people might use Philadelphia car share. An important point of caution here when you get to point number E is often what people tell you in terms of their intention is not always what they follow through on. So again, remember the golden rule, be conservative. So if you ask people on, say, a five-point scale, how likely are you to use Philadelphia Car Share, extremely unlikely, slightly unlikely, I'm not sure, slightly likely, extremely likely. The research would suggest, only count those people who give the most extreme positive response, and maybe even cut that group down by half. We want to be as conservative as possible. Note, of course, that the answers that we get to these five questions that we've asked will also inform us about what actions we might want to take that influence the ultimate demand number. Now we could do this for the car sharing example. Maybe I'll get you to think about that. As a homework exercise, I'm going to do it for you in the toothbrush example from India, that one's coming up next. So now, we've just thought about car sharing in Philadelphia. Now I'd like you to put yourself in the shoes of a large multinational, manufacturer in toothbrushes and trying to sell them into India. So, I love going to India. I was there recently, so this is an example that I personally find very, very interesting. So if you look at the landscape of India, it's about 80-20 rural to urban. So that becomes important because people in urban environments might be more likely to use the toothbrush, which is more of a developed world kind of device. However, among people who might be willing to brush their teeth, there's also the issue of frequency. So on the chart there we can see some people might only brush once a week, 7%. Some people, almost a quarter, go through and they brush twice a day. So now let's combine those pieces of information to try and come up as the company in question, that I can't name, actually did in terms of the decomposition to develop the overall forecast. So in order to do the overall demand, or the total demand, addressable demand, for toothbrushes in India, we would start on the left-hand side by thinking about the entire population of the country, or maybe for practical reasons, we might limit it just to the urban population, for example. Then we would multiply that, i.e. decomposing, by the customers who, the percentage who currently brush. Now obviously the implication underneath that If that percentage were relatively small, we would want to engage in a marketing tactic or a campaign to persuade people to brush for the first time. Continuing on, with the demand forecast, we would then think about the percentage of people who would buy a brush in a store or our channel of distribution. Then we would think about how many times people might brush per month. Then we might think about how many times are they going to use the brush. So how many times do we use it before we actually want to replace the brush and get rid of it? Now if we went through all of that multiplication from the left to the right and kind of a waterfall, we would come up with an ultimate number of what brush sales might be. And we would also have four triggers that we could use to bend a forecast or to improve the overall amount of demand that we might realize. Of course count to much about the population but we could increase the number of people who are persuaded to use a brush. We might be able to persuade people to use brushes instead of the alternatives which a twig or our brand versus somebody else. We might also be able to run a campaign or run public health or something else to increase the incidence of brushing. And then finally, we might be able to encourage people to replace the brush once a month as opposed to once every two months. So this is a fairly simple example, but believe me was very, very powerful for the company that was concerned, a large multinational trying to introduce this product into India. And so now I just want to give a brief introduction to some of the kinds of rules and methods that we can use practically to forecast as entrepreneur. Again, earlier in the lesson I refered you to my colleague, Scott Armstrong who has a whole portfolio of tools that you can investigate that might be particularly appropriate for your own product or service. And so now let me give you a simple road map for thinking about demand forecasting methods and different kinds of rules. First of all, you want to try and surround yourself as the entrepreneur with as many experts call them domain experts, as possible. So if you were launching a new kind of a business to educate people around the world in mathematics or something else, you might talk to people who were professors of mathematics that would be an example of a domain expert. The second thing you want to do when you set up the forecast is try and decompose all the customer behaviors that lead to demand. That means the search for information. That means the evaluation of alternatives. That means thinking about the criterion that people use before they make a buy versus no buy decision. So all of those drivers that ultimately lead to one individual making a decision in favor of your product or service, added up over all of those individuals. The next thing you should think about doing are what we call judgemental methods. So a judgemental method is exactly as it seems you simply ask experts what they're opinions are about the demand for your product or service and also most importantly get them to articulate by writing down or verbalizing what the reasons are their demand forecast. When you do this you should use structured analogies. Remember Eisenhower's quote? The future looks more or less like the past, that's my paraphrase. So when people are thinking about the demand for your new product and service, what kind of analogous products or services that already exist in the market are they referring to? And then, finally, very, very importantly, again my colleague Scott here at the Wharton School is very big on this. You should combine forecasts from different individuals, different experts to give some kind of a weighted average which will ground you in the good territory for the ultimate forecast for your product or service. And then finally, if you're a little bit more sophisticated or a little bit more inclined to quantitative analysis, you could do this yourself or you could outsource it. Running some kind of regression based models where you're trying to figure sales, market share or demand as a dependent variable. Model is a function of different cause or factors, that's a little bit beyond our lesson for now but there are companies out there that can help you do it. Probably most of you know somebody who's somewhat facile in the area of regression analysis as well. And so now in our final piece for forecasting demand let's go through situation where demand might be really quite uncertain. So a little bit of a counterpoint to president Eisenhower that things are not necessarily the same today as they were in the past. So, the first thing that you would want to do here with your team is sit down as an entrepreneur and try to identify all the potential users and also users of the product. So one of my favorite examples here that lead to a lot of expansion around demand is that little thing called baking soda. Baking soda used to be used of course just by people who cooked and in cooking but now that product appears in all kinds of things from toothpaste and detergent and so on. So try and identify as many users and uses as you can for this potentially new product or service. Same thing if we were to think about eyewear. Eyewear could be used as a functional product just because i need to be able to see but also one use could be as a fashion use that might have an implication for me buying more than one pair of sunglasses or one pair of eyewear. The second thing that you should do is really drill down into the different segments for the product. The component pieces of demand, group A versus B versus C, and so on. Thirdly, you should try and develop a point of view on what are the key forecast drivers or macro factors affecting the demand for your product or service. This will be something that you'll have to think about in the example that we're going to go on to in just a moment. For example, if we were thinking about eyewear, perhaps the fact that people now get LASIK surgery and permanently correct their eyes may have a very, very big impact on a market for a product that is something that we're currently wearing on our face. And then finally of course, all good forecasters taking the golden rule from Scott Armstrong, being as conservative as possible, but on top of that, can we identify sensitivities? Are there particularly extreme scenarios where demand might literally explode and or fall off a cliff? And if so what would be those things that would cause that dramatic change? Again, if you were to go through these four steps in detail. If you are faced, as an entrepreneur, with a product or service that was particularly new I think you will find this very instructive and very helpful. And now as the final part of our session, I'd like you to focus on an exercise. There is nothing quite like it as an entrepreneur to really have the rubber meet the road and really put these concepts into practice than to actually do them. So here's the example I'd like you to think about. This is a product that I've come across quite a lot recently at the gym and other places. Coconut water, no particular brand but if you've been living in the United States you might recognize this as a product that all of a sudden is appearing all over the place. So imagine that you are introducing a new high quality premium brand of coconut water. I'd like you to take two concepts from our lesson today. First of all the ACCORD model. What would be the various factors that might inhibit or retard demand for your new brand of coconut water. And then number two, could you do a decomposition that would give you an estimate of roughly how much coconut water might be drunk in a year in the country that you live in? Could be the United States or it could be elsewhere? So, take those two concepts, ACCORD and decomposition. Pretend that you're an entrepreneur, as there are many now coming up with new types of coconut water and try and go through that process of forecasting. Again all the best are applying these concepts from our lesson. Forecasting demand for your product or service, is absolutely make or break particularly in the early stage of the entrepreneurial journey.
