Inventory Management Metrics and Classification
Eric Johnson
Eric Johnson
Introduction
Slide 01: Overview
Hi, this is Eric Johnson and I'm a professor at Tuck's School of Business at Dartmouth College. Today we're going to be talking about inventory management and metrics. And this is just really an important subject. In my experience in teaching inventory and supply chain topics now for over 10 years, I found metrics to be one of the most important ideas that you can master in supply chain curriculum. And a lot of the companies I work with just really struggle with supply chain metrics. My background is this, after finishing graduate school, I spent time working at Hewlett-Packard.But after working at HP for a while I moved to academia, taught at Vanderbilt for eight years and now teach at the Tuck School at Dartmouth. Been here for my fourth year here. And well I teach, MBA students who are getting a master's in business administration and I teach classes on supply chain management. And one of the things that's important when thinking about metrics and supply chain is always keeping track of that business focus.
And we're going to do that today also. So you'll see that while we get down into the details many times, we'll be working on focusing on the business problems, what are the problems that managers face, and how do they solve their problems, keeping that kind of business focus in mind. So that's what we're going to do today. Inventory metrics and a little bit about classification.
Slide 02: Session Outline
We'll start by talking a little bit about inventory classification, because that's one of those core subjects to almost all parts of supply chain curriculum and so you always want to be keeping that in mind any time you start thinking about supply chain inventory management. Then we're going to talk about measuring supply chain performance and - wow there's a lot of things to talk about. Fill rates and more is my bullet point there and of course there's a lot more to talk about than fill rates.
And you might ask yourself, well, why are we talking about metrics theory? Companies use lots of different metrics in supply chain and inventory management - many, many different metrics. Some of those metrics are great metrics and really appropriate. Some of them may be not so great and not so appropriate.
But in any case one of the things that's really important to realize about metrics is that the wrong metrics can create as much havoc as good metrics can create good. And there is an old saying: you get what you measure. And, well, sometimes when you start measuring something, you might get something you don't really want and you have to be thinking carefully when you design metrics out of metric into an organization. Will it really achieve my objective? Will it produce a good result or will I get some strange behavior, maybe something you don't quite expect.
So we'll talk about that and then we'll turn and talk a little bit about a few specific metrics, one in particular looking at order aging and order windows. And really that metric is designed to start thinking about the role of inventory and supply chain performance and its impact on inventory.
So we'll talk about that and along the way I hope to be able to provide you with some great practical examples and insights. Since leaving HP and going into academia, I have been working with a number of high tech companies, but more recently studying also some other very different industries. Studying the toy industry, studying the apparel industry, and one of the things that you might notice that these industries have in common is they all have very short product lifecycles which creates lots of supply chain challenges and every one of these industries and companies in them really struggle to put together a good set of metrics to monitor their inventory performance and their customer service, their supply chain performance. So we'll be talking about that and I hope to be able to give you some good practical examples and insights along the way.
Inventory Classification
Slide 03: Inventory Classification
All right, so let's look at this concept of inventory classification. Inventory classification is a really important topic to supply chain because there are so many things when we start talking about inventory where we really want to be clear about what inventory are we going to be measuring and why. So first of all, you have to ask yourself, why classify at all? Why would companies classify inventory? Why should they classify inventory. And the idea is very simple. The idea is that there's a lot of different things to manage - many, many different SKUs - stock keeping units - items that you might consider managing. But what you want to spend your time on are important items. That is, what are the most important items to focus your management attention on, to focus investments and information technology on. What are those most important items? You know, there's a lot of different names for this. A lot of companies will call it ABC analysis, and the idea is that "A" items are very important, "C" items are not very important at all, and "B" are somewhere in the middle. Well, how do we decide what's important? That's the whole issue about how to classify.
And in my experience, companies almost always classify in one of two ways. They either classify based on the value of the product or the inventory that's being managed, or they classify based on the volume or how quickly something is moving - the turns of a particular item. So, dollars or volume. Now these are both good reasons to classify so the idea is that if something is valuable, that is you have an item that is very expensive, it's an "A" item.
If it's an inexpensive item, you know, paper clips or something like that, then it's a "C" item and you're not going to spend much time measuring it, managing it, and all these kinds of things. You just buy enough paper clips so you never run out because it's something that's so inexpensive you don't want to really spend much time worrying about it.
An "A" item - that's valuable - of course we want to spend time on. The same thing is true behind the rationale for volume. That is, a high volume item may be something that is very important to your business because there's a lot of volume there and you want to make sure you don't run out of inventory of that item. But beyond that, there still are many other good reasons to call something important or to classify it as an "A" item. And this is something I think where you really want to challenge the assumptions of the company you're working with or in. That is, it's so easy to just fall down the same path that everyone has done to say - okay we're going to do some ABC analysis and we're going to do it on value - where there may be many other features of an item or some inventory that might make it very important and may make you want to spend more time managing it.
Let me give you an example. Think about something like storage cost. Think about that for a minute. Maybe you have two items in inventory. One is let's say - they're both about the same cost, but one of them requires freezing. Maybe it requires a lot of space in the warehouse. And suddenly it's very expensive to store that item because it has to be frozen or refrigerated or because it's so large and bulky that it takes up a lot of space. That might mean that that's an "A" item. That is, an item that you might want to spend more time managing and concentrating on, designing good metrics around. Those may be items that you're going to spend management attention on. And it's an "A" item. There's many, many others. Think about it for a moment. There are many ways that something may end up being rather important to you.
Let me give you a couple of other examples. What if the item is supplied by a supplier who's unreliable. That is - you request shipment, place a purchase order - but you're never sure exactly when the product's going to arrive. It might arrive in two weeks, it might take six weeks, it might take three weeks, it might take a week. They're just unreliable. Well, regardless of the value of that item, it may suddenly be that you're going to have to spend more time managing that item. And, in particular, managing the inventory of that item so you don't run out. Because if the supplier is late or takes more time than expected, you could end up in a shortage - and even if the item is not very valuable, it still could create lots of customer dissatisfaction or result in other costs in the supply chain.
Another reason something might be important is that it's just critical to manufacturing. It may be an item or some component that goes into many different products and if you're out of it, it's a show stopper. Suddenly you can't produce anything. That could be a critical item, even if it's not very expensive. And then you couple that with an unreliable supplier and now you have an item that may be a "AA" item. It's really important. I mean you want to be spending a lot of time thinking and managing it. I bet you could think of many others - many other reasons something might be critical or important. What if it has a short shelf-life? Oh, then it might be important to spend more time managing it. If it's perishing or it's becoming obsolete or losing value quickly. All those may be reasons to consider that to be an "A" item, one that you want to concentrate on.
Slide 05: Inventory Classification Examples - PCs
If you look at PC makers. A PC when it's first introduced may be state of the art PC -these days a couple thousand dollars state of the art PC that is introduced. But from the moment that new PC is introduced and pushed out into the supply chain, manufactured and pushed out into the distribution channels, it starts to lose value. So that two thousand dollar PC a few months later might be an eighteen hundred dollar PC, then a sixteen hundred dollar PC, then a fourteen, and twelve hundred dollar PC. But as it starts reaching the end of its life, it falls down to a thousand dollars, eight hundred.Oh, now we're getting down there, maybe a six hundred dollar PC. This is a very inexpensive PC and suddenly at the end of its life, it goes from being an eight hundred or six hundred dollar PC to really being worthless. You can't even give the things away. Well, towards the end of its product life, it may mean that you have to spend more time managing it so you don't end up with inventory when this thing becomes worthless. Just like anything, like fresh fish, it's literally rotting as it sits in the distribution channel, PCs are rotting. That is, they're losing value.
And so, a PC maker may spend more time managing and worrying about the inventory of an item at the very end of its product life when incidentally it has a lower value than at any other time, simply because the ramifications of being caught with that inventory at the end of product life are extreme. So it's an "A" item. It could be an "A" item. So you could see you could think of lots of - we've already named quite a few of them. The liability to the supplier may be its lead time, the cost to store, the shelflife of the product, how critical it is to manufacturing, and I'm sure you can think of others. The point is, almost in any organization you walk into, you'll find inventory classification happening, ABC analysis being performed, the way it's always done, either by volume or value, when in fact there may be other reasons that some item is important and you want to challenge those assumptions. At least think through and make sure, is this really an "A" item?
Slide 06: Inventory Classification Examples - Automotive Seats
Let me give you one more example just to make sure you understand what I mean. Think about the auto industry for a moment. The auto industry, well from the outside there's one thing they procure that costs more than anything else and that's seats. Regardless of the company, you'll find throughout the world -auto makers procure seats. That's true for German auto makers, American auto makers, Japanese auto makers. Almost across the board, everyone procures seats. And it turns out there's really just two large seat suppliers in the world. Almost everybody uses one or the other. Now, you think about that. If you were managing the inventory of those seats, well here it's the most expensive item you'll procure from the outside so that would make it an "A" item by many standards. Secondly, seats go into every car. That's high volume, right? So high volume, high value, this is certainly an "A" item by most traditional ways of thinking about ABC analysis. But let me tell you another part of this story.
It turns out that in almost every case, whether it's a Japanese transplant in the U.S., European auto maker in Europe, a Japanese auto maker in Japan, in almost every case what you'll find is the seat supplier will end up building a plant literally two or three kilometers, four or five kilometers from the assembly line. And they deliver just in sequence to the line. What that means is when the car starts down the final assembly and it's being assembled into the final car, a signal is sent from the manufacturer to the seat supplier. At that point the seat supplier will assemble a specific set of seats for that car.
So, for example, if they need a pair of red leather seats, a signal goes from the assembly operation to the supplier: build us some red leather seats. At that moment the seat supplier will jump into action, build those seats - they have a much shorter assembly time than building an entire car so in a couple hours or less, they can have the seats built, loaded onto a truck, and shipped over to the manufacturer.
In the meantime, back at the automotive manufacturer the car is slowly going down the assembly line being assembled and it will be in there for hours. So the seats will easily show up in time for the car. They're typically installed at the very end of the assembly process because seats can be easily damaged. And so to put those seats in right towards the end of the assembly process, the seat supplier ships in sequence to the line, so just when that car that needs those red leather seats pulls up, the next pair of seats waiting there is that red leather pair. So, very little inventory, and once it's up and running, very easy to manage. It really may not be an "A" item. Once the suppliers really learn how to do this and the manufacturer works together and they've got it going, there may be no more than a couple hours of inventory between them and it runs very smoothly. But on the other hand, say you're a Japanese automaker transplanted in the U.S. and you're still importing some things from Japan. Well, long lead times - maybe some unreliabilities.
Maybe there's a port closure because of a strike and your items are slowed down. Maybe there's just unreliabilities of transportation. In any case, even though it may be an inexpensive item, even if it's a simple nut or bolt, think about it - if you don't have that, it may shut the assembly line down. Suddenly that could be a very critical item based on the fact that there's long lead times and unreliabilities. So you may spend more time managing that item. The point is to challenge the assumptions and think through carefully, what would make an "A" item. Now really for the rest of what we're going to be talking about today with metrics, when we talk about inventory metrics we're going to be considering primarily "A" items. I mean that is when you spend a lot of time measuring inventory and order fulfillment and all these kinds of things, typically you're measuring them around important items.
Business Metrics
Slide 07: Business Metrics
So let's turn to the next slide, and talk a little bit about metrics. Now, of course, there are many, many different type of business metrics that managers at all levels in an organization deal with on a regular basis. At the highest level, there are corporate financial metrics. These are metrics that Wall Street and financial districts in London and Taipei and Shanghai and throughout the world, Munich, these are the things that markets really focus on and drive stock price in many cases. Things like ROA return on assets or the price to earnings ratio of a stock - very, very important kinds of corporate financial metrics. Now we're not going to spend a lot of time thinking about corporate financial metrics today because we're going to drill down and think about inventory and supply chain. But it's important to always remember that in the end what we really are driving for is shareholder value and we're driving for business performance. And so whenever we start talking about inventory and start talking about order fulfillment, we always want to be thinking about what does this mean in terms of the overall corporate financial success? For example, if I'm reducing inventory, that's a good thing for return on assets, right? Because inventory is an asset. If we reduce our inventory, potentially we can improve our return on assets, Wall Street may welcome us with an increase in share price. That's a happy result. So we always want to be thinking about - okay, what am I doing here? If I'm improving customer service for the customer - well then again that might drive my revenues up as customers are happy with my product and again improve things like my return on assets, because I now have more revenue. Or my earnings may be improving and so the price to earnings ratio, which is the price of the stock divided by the earnings, will also potentially improve. So in every case, we want to be thinking about what is the link back to the corporate financial metrics? But typically inside the organization we focus more on functional metrics. Lots of different functional metrics, whether you're in manufacturing, engineering and sales.
I've listed a few there just to get you thinking about them. You know, in manufacturing, you're often thinking about labor productivity or utilization of your facilities or your equipment.
If you're in sales, you're thinking about market share and sales growth or if you're in engineering and R&D, you know, you're thinking about the unit cost of the item that you're producing or how long did it take you to bring a new product to market. These are all different types of functional metrics.
But the question you have to ask yourself is - how many of these metrics apply to more than one functional area? That is, how many of them cut across a number of functional areas? Really the answer is, not very many, right? But you know, if you are in manufacturing, you're not measuring market share. That's a marketing and sales function. Likewise if you're in engineering, you may not be worrying about the utilization of your manufacturing equipment. But typically in supply chain - one of the key features of supply chain is that many of the metrics we're going to measure start cutting across many different functional areas.
And there are, of course, many different types of supply chain metrics. But supply chain, by its very nature, includes manufacturing. It includes sales. It includes engineering. It includes suppliers. It includes distributors. And so many times when we're thinking about supply chain metrics and inventory metrics, we're often thinking about metrics that may span across more than one organization, more than one functional area within an organization. So, what are the things that we're thinking about when we start thinking about measuring inventory performance?
Inventory Measures
So let's think a little bit about measuring it. Now inventory measures themselves are pretty straightforward. That is, there are many simple and effective ways to measure inventory. Inventory can be represented by demand. That is, you can be thinking about how many days or weeks or months of supply you have of a certain product or finished goods or raw materials. You can represent it in terms of demand. You can represent it in terms of inventory turns.Inventory turns is simply the sales of that item divided by how much inventory you have. So if you sold ten thousand dollars of inventory that year of your product, ten thousand dollars of the product in revenue, and the average inventory was about a thousand dollars in average inventory that year, you would have 10 inventory turns. What that says is that the inventory turned over or was used up and resupplied by new inventory 10 times during that year. Now there are many different ratios like that.
For example, inventory to sales ratio is simply the inverse of inventory turns. So if you took that inventory turns and just turn it upside down, turn that fraction upside down, and divide inventory, where is that inventory coming from, well it's coming right off the balance sheet in most cases that's what people would use, the dollars of inventory on the balance sheet, divided by sales and you have the inventory by sales ratio.
So it and inventory terms are very similar, measuring basically the same thing. Many, many companies at a financial level will measure one of those two things. Or they'll just measure inventory dollars. Now the one thing that you want to be very careful about in measuring inventory itself is how is that inventory valued? How is the inventory valued? You know, on a financial summary of the firm, on the balance sheet, inventory is often measured financially in dollars but how do they actually determine the value of the pieces of inventory throughout the year? You know, the prices may have been changing. Prices may have been going up, going down. If you have inflation, they're always going up. How is it measured? What if the inventory is deteriorating or perishing during that year? Maybe the inventory value is going down. Now you'll find at a financial level companies have to make a decision.
For financial accounting purposes they typically will measure it using some specific accounting rule, like first in/first out or last in/first out. First in first out means that whatever was purchased a long time ago, the oldest item, that's the one that will be taken off the balance sheet as being sold when you sell one, whether or not you really sold the oldest one or not. And so if it had a lower value, now it disappears out of the mix of inventory. Last in/first out does the opposite. The point is that, well, depending on whether you're first in or first out, last in or first out, the value of the inventory you have in dollars will change. Now the number of units will be the same.
But you may have a thousand cars in inventory finished goods and it's a thousand cars. But depending on how you value those cars, first in/first out, last in/first out, the dollar value of those cars could be very different. Sometimes firms may use the market value of the inventory or some other measure of its value. In any case, you'll get a very different result and, of course, a good CFO knows how to take those valuation techniques and show one that's the most favorable for the given circumstances.
For what we're doing today, many times we often like to think more in terms of physical inventory or like days of supply, because it's less easily manipulated, I should say, by how we value it in terms of dollars. But both are very important. And so they're both are very useful measures of inventory.
Service Measures
Slide 09: Service Measures - Fill Rates
Services measures often become a little bit more complex. That is, there are many different service measures. If you look on the slide, there are many different service measures and they measure slightly different things. So for example, very popular ones are things like fill rates, whether they're item fill rates, line item fill rates, dollar fill rates, order fill rates - these are all very popular in different companies. I've put a definition down for each of them for you to help you understand the difference. Item fill rates are the percentage of total items filled without delay. That is, if an item's demanded, can you fill it immediately? What percentage of the items were you able to fill immediately?
Line item is the same idea except now if you have an invoice from a customer order, it may be that the first line of the invoice asks for 100 of the same stock-keeping unit. Well, a line item fill rate measures what percentages of those lines on the invoice that were completely filled. So a line may have a demand for one item of that type or 100 items of that type. The question is - were you able to fill all of that line, all 100 items or all 1 item, and that would count as one line in your line item fill rate. So you'd simply take all the lines of the invoices that you were able to fill during a time period of a week or a month or a day, and divide that by the total number of lines demanded during that time. So you can see how it's slightly different than item fill rate. Both very popular.
A dollar fill rate sometimes measure the percentage of dollar value of the item filled without delay.
Or complete order fill rate. Sometimes companies call this a perfect order. The percentage of complete orders filled without delay.
Each one of these is a great way to think about it. And a great way to think about service. But they each have their good points and bad points. And we're going to talk about that in a little bit. And of course there's many other ways to measure service.
Availability is a very common measure. Particularly in retail supply chains. Availability is the percentage of days without a stockout, usually a percentage of time. Many times days, but it could be weeks or months. It's a time measure. And so for a retailer, they may be asking themselves, what percentage of the time did I actually have something on the shelf. What percentage of the time was something available? That's an important measure for them. If you think about it, many times fill rates are very common in distribution. So if you work in distribution, you find people using fill rate measures. Whereas availability is often much more common at the retail level. And you might ask yourself why. Why would one measure one and one the other? Let me give you an example to help you think that through.
Say for example, I'm going to go out and buy myself a new tie. Now there's a couple of different ways I could do that. One way I could do it is I could head out to my favorite department store and browse the aisles and find a tie and buy it. Another way now, of course, is I could jump online. I could jump online at my favorite e-tailer, find a tie that I like and order it. Now what's different about those two experiences from a customer service/order fulfillment standpoint? I mean there's lot of things different about buying something online and buying something in a department store. In a department store, I get to touch it and feel it, and online I see pictures of it. In a department store, I physically go get it, and online they ship it to me. But think about it a little bit more.
What's different from an order fulfillment standpoint? What happens when there's a stockout? So I go to the department store and say that day I was kind of just looking for a red tie and it turned out that they might have had red ties at one time but they're stocked out. Well, I go to the department store and I'm looking around and I don't see a red tie. I don't even know if they ever had a red tie. I just don't find one. I don't know that one ever even existed. If I don't find it, I just leave. Maybe I just thought they never even had one. The point is even to generate the demand at the department store, they have to have the red tie. If the red tie is not there, I'm not going to try to buy anything. I mean, maybe if I'm really aggressive, I might go ask a clerk, "Hey, do you have any red ties around?" But if I'm not really quite sure what I'm looking for and I'm just browsing, don't see, you know, something that catches my eye, I just leave. Lost demand.
The e-tailer, on the other hand, they're in a different situation. I see some cool guy wearing this tie, this red tie, and I go, "Man, that's a great tie. I want to look like that. Right? And I click on that and say I want to buy it. Now at that point they may be honest with me and say, "I don't have this in inventory but we can ship it to you in two days." In any case, I've tried to buy it. In fact, they know that I've tried to buy it. And so the point is they don't need the inventory to generate demand. I mean, they still need inventory, but they don't need it to generate demand and they may not need the inventory right away. I might be willing to wait a day or two. I mean, obviously if I was going to buy it at the etailer, I was going to wait for some shipping delay anyway, maybe another day. It really wouldn't make any difference.
Well, if you think about those two stories, you begin to understand really the difference between this metric availability and the metric fill rate. Availability is the percentage of time without a stockout. What's important there for a retailer is, they use availability because not having something on the shelf is in effect a defect for them in terms of service. That is, if I walk into a store, even if I'm not looking for anything, if a lot of the shelves are kind of half empty, I'm probably going to have not a very good impression of that store to begin with. And so, having items on the shelf affects customer satisfaction. And it helps generate demand. If there's nothing on the shelf, people aren't buying. So they need to have very high availability of their products on the shelf. Fill rate, on the other hand, is more common in distribution. And it's more common because you don't really need inventories that generate the order.
Slide 11: Service Measures - Availability
The order, it's an order based system and the order will probably come to you even without the inventory. And fill rate is actually in many ways a less stringent measure of customer service because it provides what I call partial credit. I teach here Dartmouth College at the Tuck School of Business and my students take exams at the end of the class, right? And sometimes there'll be a question on the exam where they might get the final answer wrong, but they have a lot of good ideas along the way and were close to getting it right and maybe they'll get some good partial credit. Maybe if the question was worth 20 points, they might be 18 or 19 points out of 20. Even though the final answer at the very end, maybe they calculated the wrong number or made a mistake along the way, kind of partial credit. Well, fill rate is like that. Because you think about it during a week, say for example you had demand for a thousand items, and you were using item fill rate as your measure. Well, if you were able to deliver 950 of them immediately without any delay, fill rate would say that you're fill rate for that week is 95%. Nine hundred fifty out of 1000 items were delivered right away, 95% of the items. That's pretty good. But availability would say, wait a minute. You had a stockout. If you were measuring how many weeks had a stockout, well, you had a stockout that week. That counts against you in the availability calculation.
In fact, you think about that - take that example a little further, maybe if demand was 1000 that week, as I said, but you actually satisfied all 1000 - you had 1000 units of inventory and you satisfied all of them. Of course you used up all your inventory doing that, but 100% of your customers had their orders filled, their items filled.
Item fill rate was 100%. Availability, on the other hand, would say, well wait a minute though, you used up all your inventory. You had a stockout. Even if no customer came in and ordered anything later on or wanted something, you had a stockout. That reduces your availability for that calculation.
That's pretty strict, but for a retailer, that's important because they really don't know if somewhere during that week a customer, the 1001th customer didn't walk in looking for another item. They don't know that because typically they're just going to walk away if they don't get it. So they want very high availability.
Slide 12: Service Measures - Other
Now there's many other measures of service. One - a very good measure is a turnaround time. How long did it take you to process the order? When you start thinking about these service metrics, you have to start thinking about, okay - what kind of an environment am I in? Am I in an order based environment? Am I in a stock based environment? Retailers in a stock based environment, they like availability. In order based environment, you may be thinking about something else. If you are in manufacturing, and you're making items to order, then you never really have any inventory. You just make the item when the customer demands it. You may be much more focused on how long did it take me to get that customer order completed than what percentage of the time did I have a piece of inventory waiting for the customer when they arrived. So you have to be thinking about what is the environment and where in the supply chain are you when you start thinking about these service metrics. But there's still lots of other things to be thinking about.Pitfalls with Metrics
Slide 13: Pitfalls with Metrics
There's really three or four major pitfalls that I've seen in the different companies I've worked with. The first one are metrics that are too focused, too myopic, too focused on some little local issue or problem. I'll give you an example. Machine utilization. Machine utilization is a great metric on the factory floor to help you understand how much you are using a machine. But it may not be a very good supply chain metric. Why? Because many times you start measuring utilization, you'll start doing strange things like producing inventory when you really don't need it simply to keep the machine busy. Many companies have fallen into that trap where by measuring utilization, they start driving themselves to build inventory and fill up their warehouses with inventory - thinking they're lowering their costs when in fact they're just driving their costs up by holding more inventory, renting more warehouse space, tying up more of their capital. It's a local focused, myopic metric that just sends people off in the wrong direction. Another major pitfall is what I affectionately call tinkering. Tinkering is something that we've all learned to do. That is, any metric that you have can be beaten. What do you mean, beaten? Well, you think about it yourself. Anyway you've been measured over time. You've probably learned ways to be able to make that metric look maybe even a little bit better than it really is. You learned how to beat the metric. One of the best examples, a good friend of mine who I work with quite a bit together - we were working on a job for a large grocery distributor, grocery store chain - and we were doing some consulting for them on order fulfillment and supply chain and part of that project we were looking at different distribution centers to see how they performed, how much inventory did they have, what were their fill rates like. And we found that one distribution center had a much higher levels of order fulfillment than any of the others. That is, they measured a line item fill rate like we were talking about earlier and their fill rate was much higher than everybody else's - and yet they didn't have any more inventory than any of the others so we thought - wow, how do they do this? So we got on a plane to visit them - it was in Texas and we came and visited them and sat down with the distribution center manager and asked him "How did you do this, much higher fill rates than anyone else? How did you do this?" And he leaned back in his chair and kind of smiled at us and said, "Well, it's easy." And we were thinking, all my goodness, this will be good. You know, we'll write this down and this will be best practice, I am sure if this fellow's figured something out.
And he says, "Well, this is what I do. Anytime I run out of something, say I run out of frozen peas, frozen carrots, I call up all the stores and tell them don't order any more frozen peas, I'm out." Well, think about that a minute. No orders means that there are no order fulfillment defects and if you're measuring line item fill rate, and nobody's ordering the carrots that aren't there, of course you're not having any fill rate problems, because there are no orders. And so by telling customers not to order, of course he was improving his metric.
Now what would happen if as soon as those carrots would come in stock, he'd call everybody up and tell them, "Hey, carrots are in." Everybody would place their orders, you'd have 100% fill rate and things would look great. Well, of course, that's tinkering and so really what he was doing was a good thing. He was providing information to his supply chain partners so at least they weren't ordering and not receiving anything and wondering where it was. But, of course, from a supply chain perspective he was also hiding a problem. That is, when he had stockouts, nobody knew it. And it may have been causing other problems. That is, the stores may have been buying from other distributors during that time or their shelves may have been going empty during that time and yet it would've looked like the distribution center wasn't part of the problem because they had a high fill rate. So he was tinkering. Classic metrics problem.
Then there are metrics that are measured only periodically. You think about these problems, I'm talking about myopic measures, tinkering, periodic measurement - what I mean by that is how frequently is the metric measured and when is it measured. You've probably seen this if you've ever worked around a sales organization that has end of the week or end of the month or end of the quarter quotas, suddenly you get a lot of sales on the last day because, well the measurement is going to be made tomorrow or that day. That causes a lot of problems there just like it will in a supply chain. I was once working with a manufacturer in Singapore; it was a division of a bigger company and at that particular plant they were trying to reduce factory inventory which was measured on a monthly basis. Once a month on a specific date they would measure how much inventory was out on the factory floor. And the factory manager would get a bonus if he could drive that number down. Well, that guy was smart and what did he do? He- about a week before they would make the measurement, stopped releasing new orders to the factory floor to kind of dry up some of the inventory and so when the time came to measure it, sure enough the inventory would be pretty low, nice and low. The next day though things would chaos. They'd dump a whole bunch of new orders out there and there'd be inventory all over the place. Suppliers would be arriving with raw material. Everybody would be expediting orders and they'd just be crazy. But yet for that manager, of course, he was tinkering and he was tinkering because of the periodic nature of the measurement, the fact that it was being measured on a monthly basis created strange behavior.
So you see these different pitfalls - myopic, periodic, tinkering. And there's plenty of others. Some metrics are too complex or there's just too many metrics and no one really knows what to concentrate on. Or sometimes you measure things and never take any action. You know, if you don't take action on a metric, pretty soon everybody realizes it and they don't pay any attention to it anymore. These are all pitfalls in metrics. And I'm sure you've been in organizations where you've seen metrics that went awry that really didn't measure what you were hoping for. Well that leads us to think a little bit about metrics theory. What is it that makes a good metric?
Metrics Theory
There's no perfect measure because in the end what you're trying many times to do is to measure kind of not just what's happening now, but where is it going, what is the future? Are we going in the right direction or in the wrong direction? The stock market's always trying to figure out not just what happened last month or last quarter -what were sales last quarter. Well everyone wants to know that, but they more importantly - they also want to know which way is it going? Are sales declining? Are they increasing? How fast are they increasing? Are they reducing inventory? Have they reduced it a lot, a little? Will they reduce it more next quarter? These are the kinds of questions they're asking and it's very hard to measure that. We typically measure historical things, things that have happened and try to use then to project the future. That's tough. But there are other things about metrics that you need to know. One is that any metric that you put in place at first may be very effective but it will slowly decay as the longer it's in use. The reason it decays is that it loses its punch and the punch is variance. That is, typically when you put a new metric in place, there's a big difference between some of the best and the worst. If you're measuring a bunch of distribution centers or a bunch of different employees or a bunch of different supply chain managers or different factories, or you against your competitor, there's ones that are - some that are doing well and some that aren't doing well and the ones that aren't doing well will work hard to improve. But over time, they'll improve and eventually everybody will kind of be pretty good on that metric. Now, it doesn't mean the metric is not useful anymore but at that point, it really becomes a maintenance thing. You're just kind of keeping things the way they are. They're not getting worse. They're not getting better. But you really need to start thinking about what will be the next thing you want to measure to try to really push the organization or the supply chain forward.
One of the things that's really true is metrics accumulate. It's much easier to place a new metric into an organization than it is to take one away. People get used to them. They get their bonuses on them. And it's really hard once you've instated some new ones - it's very hard to get one out of there. And as you get more metrics, everyone puts their new favorite metric in place, accountability and focus decline because you have all these different metrics - it's hard to know which one's important.
Slide 16: Characteristics of Good Metrics
What are we really focusing on here? Well, what's a good metric? How do you get a good metric? Good metrics exhibit a few different things. They exhibit variability. That is, between the best and the worst players. The second thing is good metrics exhibit independence. You want metrics that measure different things. That sounds kind of stupid but yet a lot of companies will measure two or three different metrics - they're more or less measuring the same thing. Like for example, they're measuring inventory turns and days of supply. Well, those are both measuring how much inventory you have in the supply chain and you really don't need both. Better to probably have one or the other rather than to have them accumulate. The last thing that's true is that you need metrics that are valid. What that really means is that you want to make sure the metric is connected to the business. If the business objective is to increase shareholder value, then how does that metric - improving some supply chain metric, improve shareholder value? Say you're measuring the amount of inventory you have in your supply chain. Well, it's pretty easy to see how that's linked to shareholder value - because inventory is an asset and return on assets is a very important measure that Wall Street often uses. And it often will be very closely connected to stock price. So reducing inventory would improve return on assets, improving return on assets likely will be rewarded by the stock market. But other metrics it's not always so clear whether they have validity. They really do they really end up helping you get to a desirable outcome. And so you have to think carefully does this really happen?
Metrics Theory & Types
Slide 17: Types of Metrics
Now there's different types of metrics - there are categories of metrics. Some are what we call disaggregating metrics. Now this is a complicated one. But what it basically means is that you can take this metric and roll it up in the organization. That is, return on assets is a good one that's disaggregated. That is, you can measure at the divisional level and it's easy to roll that thing right up to a corporate level and come up with the corporate ROA. In contrast to that, many metrics are what we call disintegrating. What that really means is that the relationship to the overall organizational performance is not really clear. Many service metrics fall into that category. That is, if you take two different divisions and try to roll up their service metric into a corporate one, two different fill rates, well that's not always so easy - because they have different businesses, different business practices, different customer expectations, different ways of measuring customer service, different volumes all these things make it hard to roll them up.
It turns out that many order fulfillment metrics are in fact disintegrating service metrics. And so it's sometimes harder to measure them. These will be things that you want to be thinking about when you're putting together your metrics. Can you roll them up? How easy is it to roll them up? If you can't roll them up, how can you explain the differences between two different organizations?
Metrics Issues
The last thing we want to talk about then is back to our order fulfillment and we said the order fulfillment is hard to measure and we have these different metrics - fill rate, item fill rate, availability and so forth. What about for an order driven system? Fill rate and availability sound great for a made to stock system, but what about an order driven system? Here you have to be thinking about a metric that's really linked to the customer. When did the customer really want the order? Early can be as bad as late. Many retailers now may penalize you if you ship early to them because it means they have to hold the inventory and they don't want it. So being early can be as bad as late. And so you have to start thinking about, well how long did a typical customer have to wait? And how long are they willing to wait? And how much variability is there between these different order experiences?
Slide 19: Order Fulfillment - Aging
One metric that I think is particularly useful in that regard is this one called order aging. Looking at orders and comparing them to when they were expected. To make the order system, you might quote a delivery time, even if you're not making it, but you're shipping it, you may - a customer placed an order and expect it in a certain amount of time. And here what you're doing is you're looking at what percentages of the orders were early, what percentages were late, and drawing a cumulative curve that shows how you performed against the customer expectations. This is order aging and it's really much more effective than any single point measure like fill rate. Here's two order aging curves. You see those two things, but which you notice that they both have the same percentage of orders filled - or you could think of this as items filled or orders filled or line items filled. It would be true in any of those cases, but the example would still hold, and you can see it's the same that zero days right there. They're the same. They cross each other. But if you were managing these, which would you rather have? A or B? I think if you think about it a little bit, you'd be much happier with A because look what happens there.
A high percentage of the customer orders are shipped within that little gray band of on time, very close to being on time, only a few shipped early, not very many were shipped late. Very compact. And the order aging curve shows you that. It shows you how you're doing over that whole experience. How is it really going.
Summary
Slide 21: Summary
So these are some lessons in order fulfillment and you'll see that we've been thinking about measuring inventory, about these trade offs, about metrics and the fact that metrics require ongoing management attention and that finding the right mix of metrics is really the key. The order aging idea is one powerful way to manage the service component of supply chain performance because it allows you to see how much variability there is in your service. And we all know variability really is the enemy of the supply chain manager.So order aging allows us to capture that. So I hope you have found this segment on metrics to be really useful today. I know that I find in organizations that I've studied that metrics in many ways are really the most important thing. If we could get the metrics right, many other supply chain problems would become much easier because metrics drive performance. People really do what they're measured by. They want those bonuses. They want to move ahead and when you start measuring things, you'll often get what you measure, so make sure you have the right measures.
No hay comentarios:
Publicar un comentario