 Hello everyone, welcome to this course on supply chain digitization. This course is being offered by IIM Mumbai. I am going to cover the second module and the fifth lecture which is based on supply chain segmentation. If you remember in our previous sessions we started on supply chain segmentation by understanding the different supply chain challenges and then we have seen that supply chain segmentation is one of the solutions we have explored and discussed about different strategies for having supply chain segmentation and going forward we are trying to explore now those ways in detail for looking into a supply chain in a different view. In the last session we discussed about product segmentation which was based on demand variability and also your weekly sales information, how you can divide your SKUs based on these two characteristics and depending upon the combinations of the demand variability and the weekly sales the four quadrants were proposed. Depending on all these four quadrants different type of strategies can be implemented which can be either a pure pull strategy or a very pure push strategy whereas in some quadrants the pure strategies do not work out and you have to find a combination of push and pull strategies. So depending upon the type of product the level of customization required and the customer needs the push and pull boundary can be decided in a supply chain for the remaining two quadrants of this solution. Now in today's session we are going to take our discussion forward and we will see that how the inventory segmentation can be done. To understand this process we are going through a case and this will be using some basic methods which are already available in literature. So let us first look into this topic of inventory segmentation, what is meant by that and what is the importance of it. When we talk about inventory segmentation a company is owning a different type of inventory and if there is a way by which we can categorize them or segment them based on certain criteria the business knows very well that how to manage those particular segments. The inventory segmentation as we understand will help in optimizing the cost of the supply chain. The service level for each SKU can be improved accordingly the type of storage system which is required for that particular inventory can be decided and depending upon their frequency and other characteristics the picking strategy for that particular component can also be seen in any warehouse or any distribution center and there are many other operational performance which are related to this inventory can be effectively managed which are all the additional advantage of following an inventory segmentation in practice. So let us see that what are some of the popular approaches for inventory segmentation which are already being used by industries in their day to day operation. One of it is ABC analysis which is very popular in the business and is being adopted by the different companies as well. So here the inventory available is classified into A, B and C category and this is generally done based on the revenue that they are generating. So we can say that the segment A will be having approximately 20% of the total products and in terms of the revenue that is being generated by them corresponds to the 80%. Similarly, the segment B in terms of the volume is little higher than segment A but the revenue generation is considerably low whereas the segment C is the lowest of the above too and it in terms of the volume it captures the major volume but the revenue generated by this segment is the lowest. There is another way of classifying the inventory which is referred in literature as FSN analysis where F stands for fast moving, S stands for slow moving products and N stands for non-moving products. Again these are all indicated numbers over here which says that around 10% of the inventory items are fast moving and less than 10% of the average time cumulative stay is associated with it and similarly the slow moving and non-moving items are also considered. The third way of classifying the inventory which is again quite popular is VED analysis where V stands for vital, E stands for essential and D stands for desirable. Vital items are those items which are very crucial for any business. The essential items are again very critical but they have importance next to the vital for the business whereas desirable items are goods and they are those goods which are not necessary to run business operations but of course they are needed. Then we have the fourth way of classification which is being used in today's time because of the availability of data and that is called as XYZ analysis which is helping us in classifying the inventory based on demand variability. We already have discussed about demand variability in our previous sessions. From there we can say that the segment X have little or no variation in demand. The segment Y have got unsteady demand but their demand can be predicted to certain extent. Segment Z have got very high variation in demand that is one of the critical challenge and that is why it becomes very difficult to predict the requirement of segment Z type of product because there is no trend or any predictable causal factors which are associated with this category of product. So we can see that these are the quite popular approaches for inventory segmentation which are being used by the industries and on their depending upon the policies that they have adopted. But there is a limitation of these approaches. So let us look into the limitation. First of all the segmentation which are proposed are typically focusing on a single criteria majorly and sometimes they are also looking for more than one criteria as well. But if you have, if you look into your real business you always have to consider two or more than two criteria to create your segmentation and that is why the actual business requirement is quite different and quite complex. It is not focusing only on one or two criteria or maybe a multiple combination of criteria are required which should be considered for creating the segmentation of the inventory to be managed. So here if we have certain methods which can take care of this requirement of considering multiple criteria for developing for inventory segmentation this is quite useful in today's time. So what are some of the advanced inventory segmentation approaches which are available in today's time? There are numerous methods which are being used in today's time. For example there is mathematical programming which is being used and in this method linear programming or non-linear programming can easily be used for formulating your problems which consider these criteria and help us in classifying the inventory. Similarly metaheuristics like genetic algorithm, particle swarm optimization, simulated handling and many other similar metaheuristics are also being proposed for inventory segmentation. Going forward because of the availability of the data the AI and ML part as you all know and this is also the focus of our course is heavily used for inventory segmentation and various algorithms like artificial neural network, support vector machines, back propagation networks, KNIRS neighbor, regression and many other similar algorithms can be used for inventory segmentation in AI, ML and then we also have another popular approach called as multi-criteria decision making which includes methods like analytical hierarchy process that is AHP, fuzzy AHP, ANP and many other similar methods can also be used which depends on the experts opinion and different factors are being considered for classifying the inventory. But interestingly in literature has also proposed some hybrid approaches where the combination of these algorithms are also being used for inventory segmentation approaches as well. So we are not going into the detail of it but just to get an idea that in today's time because of the availability of the data and because of high computational processes available with us we can apply many advanced techniques for inventory segmentation as well. So here we are going to cover a simple example which will be based on using your analytical hierarchy process that is AHP method and how we will use this method for classifying the inventory it will be shown right now with a simple example. As we understand AHP is a popular method which is proposed by Sati and which is also being used in different possible in other areas as well like in supplier selection or in procurement decisions sometimes for product or service design and improvement maybe for resource allocation and for many other selection things this method is very popularly used. So let us see into this method quickly as it is a very standard method and very popular we will be just running through the steps involved with this method. AHP is MCDM method as already shared it what it does it tries to organize the complex problem in a heretical manner and then it does a pair wise comparison of all the elements or all the factors in order to understand the relative importance of each factor with respect to each other there are certain steps of mathematical calculations which tries to normalize them and helps us in deriving the prioritized outcomes of these factors being considered. So at the end you see that we get a quantitative and consistent decision making by following this process this can be done by a single expert or more than a single expert more than one expert and thereby this method depends on the experts opinion to understand the relative importance of the factors under consideration. This framework is shown over here which is trying to show you the steps or the mathematical steps which is followed for the AHP method. So it starts with the first step where the problem is defined and the objective is clearly stated a heretical framework is developed to show the factors being considered and their relationships and then a pair wise comparison matrix is created for every levels. So if there are single level of factors or there are multiple level of factors based on that the pair wise comparison matrix is being developed and then we do some mathematical calculations to calculate the vector priorities and weights. These calculations are quite standard and available at different platforms. So it can be easily referred and this can be easily followed from there. At the end if we calculate the consistency ratio also called as CR which ensures that the experts who have given the responses to these factors are following a consistency and that is checked through this interesting formula which finds that if the calculated value of consistency ratio is less than 0.1 then the given pair wise comparison of these factors are consistent in nature and thereby you can proceed for calculating the criterion weights in case this consistency ratio is more than 0.1 then the consistency is not there in giving the relative importance or in giving the in doing the pair wise comparison and in such case the process needs to be repeated and it is it goes back to the step of constructing the pair wise comparison matrix for each level and again the calculation of the factor priorities and weights are done again the consistency ratio is being calculated and you can see from here that the consistency ratio is again compared. So unless you get the consistency ratio less than 0.1 you cannot stop this process and this follows a repetitive procedure. So let us try to apply this inventory segmentation in a very simple example with the given case. If there is a company let us imagine about a company XYZ and it is a e-commerce retailer which is operating on a very popular e-commerce website and now this company has entered into a three year agreement with the e-commerce company where it is trying to secure a fixed amount of space in their fulfillment center. Now here the XYZ e-tail follows a practice of listing a product as available only when it is physically present in their warehouse inventory. So this is again very interesting thing or interesting practice you can see it from here and that is as you all know and you all are regular users of e-commerce in today's time we can see that how it is ensured that customers are always satisfied with services being provided and that is done through this option where the product is so shown only if it is available otherwise it is shown as unavailability. And remember that any availability of an SKU is now leading to a direct leading to the loss of sales and again it is not a good practice for the company. Now there is further a new complication is being added over here. The management of the XYZ e-tail has decided to expand their product portfolio. So already they were handling enough good number of varieties but now they want to expand their product portfolio. But they also have a challenge of the inventory space and as you can see from here that this space is limited and they also have to maintain the previously achieved service level of 95 percent of all the SKUs in the which they have previously managed. Now as soon as they have deciding to introduce a new product portfolio because of this the service level will be difficult to manage in a long run. So here they have decided to categorize the SKUs into some three distinct groups where each will be decided with a designated service level of 95 percent 90 percent and 85 percent. And for that process now they are actually trying to do the ABC analysis only so that they can have certain group of products which for which they have they will be following a 95 percent service level whereas the other group of products can be having lower service levels. So for this purpose they are proposing to use AHV method and how the SKUs are getting clusters into these three ABCs. So here we have been provided with the SKU wise data and corresponding to these six factors the pair wise importance of the factors is also shown. So let us see this into this calculation again. So here we have we can see that these are the six factors which are being considered for inventory segmentation. The first is some monthly demand that is units per month. This is the priority of product category which scores between 1 to 10. This is about the average suppliers reliability which is capturing the percentage of perfect orders. The next factor is about profit margin which is again expressed in percentage. Similarly we have lead time which is in hours and then we also have very interesting feature which is likelihood of return and also expressed in percentage. This is interesting because nowadays the e-commerce is also linked with what is the ease of return or what is that service that they are offering. So this again has to be considered when you are planning for inventory segmentation. So we have already been given with this information for 25 SKUs and one very interesting feature as you can see from here is these all these factors which I introduced over here are having different units. So how do we consider these together to decide about the importance of the SKU is again a critical challenge. So some of these factors are in hours, some are in percentage, some are in scores. So we can see that they are not having consistent units. Again this becomes a quite big challenge for us to manage. Going forward we are going to use the SATI scale which is on a scale of 1 to 9 to prioritize or to do the pair wise comparison between the two factors where one is indicating equally important and 9 indicates extremely important compared to the other. So when we say monthly demand versus monthly demand is 1, yes both are having the similar characteristics but when we compare monthly demand versus priority of product category it means that monthly demand is having 3 times more importance than the priority of product category and so on. So in this way you can fill up the whole this given matrix which can be provided by the experts in the given domain and that this is going to work as the input matrix for our relative weightage calculations. To understand this example little better let us shift to our excel sheet which is going to handle the SKU wise data given to us and then also we will calculate the AHP weightage to find out the overall scores of these SKUs. So as you can see from this excel sheet we have already discussed about the pair wise comparison matrix for the 6 factors which is already given to us and we have also discussed about the 6 factors and the values corresponding to it for the given 25 SKUs. Now let us see the AHP calculation as this is the very very standard method we will be not going into the detail of this and you can follow some standard literature on AHP and you can take the calculations from there. The excel sheet will be provided so you can also refer to the sheets to see the calculations of AHP. So here you can see all the factors the step 1 is done which talks about normalizing the decision matrix and with respect to this the additions and the subtractions and the some ratio calculations are done. So normalize you can see the resultant normalize decision matrix and the calculation of the criterion weights are shown over here. In the second step the consistency is checked and as we already discussed about the consistency ratio so using this consistency matrix calculation we can calculate our consistency ratio from here. We can see that the consistency ratio is coming around 0.08 as CR is less than 0.1 which is the requirement we can say that the given pair wise matrix is reasonably consistent. Now from here the criterion weight is coming for all the given considered criteria and we can easily see the weights for each 6 criterion from here. So remember these weights and let us see how are we going to use these weights for getting a score for each SKU. Now we have as we already are knowing the data is already provided to us for these 25 SKUs and for all these 6 factors as we can see the given 6 factors are not of similar units and if you observe them closely the meaning of these factors are also not same. For example a demand is expected to be higher the priority of the product category can again be higher the supplier reliability expectation is always to be higher the profit margin again is expected to be higher. But when it talks about leap time this is lower the better likelihood of return is again you can see is also not expected to be higher. So it means that the factors which are considered over here some of them are like the good the better some of them are like the less the better. So what we have to do now because these factors are available in different units we need to normalize them and for that a very simple method of normalization is followed where the given criterion is compared with the minimum value and is divided by the maximum minus minimum and in this way the whole data set is normalized for those criteria which are expected to be higher the better. Whereas those criteria which are the opposite of it which means that the lower is the lead time the better it is it is just subtracted with one so that you get the you can convert that minimum requirement to a maximum requirement. So in this way we have got the relative the normalized scores for all these 25 SKUs and now you can also see that the criterion weight are also shown over here in green using the AHV method that we have already discussed. So now we need to have a score for all of these SKUs which considers the importance of the weight and also considers the weightage for the normalized scores considering these particular factors. So now we have just multiplied these weights for the criterion and also along with the respective weights respective normalized scores and if we add them together we will get the combined SKU score. So in this way we have calculated the combined SKU score for all the 25 SKUs as you can see in this table and now what we have done next is that we have used this SKU score and we have first arranged them in the descending order. The next step is we have just calculated the cumulative score of all of these SKUs score and then we have calculated the percentage of the cumulative scores. Now from here we have decided certain cutoff that helps us in deciding the, that helps us in deciding the ABC classification. So when we have fixed that the 28 percentage 7 items or the 25 items are belonging to the A category and this is around 60 percent of the cumulative SKU score is kept as the cutoff. So you can see from the cumulative score till this part from 0.568 to 0.610 because 60 percent is the cutoff. So any SKU having higher cumulative score than 60 percent is considered as in A category. Similarly the next 25 percent is considered into B category and the last 15 percent is considered as C category. So from this SKU you can say that around 8 items are coming into B category and around 10 items are coming into the C category. So in this way we have classified our inventory into 3 different classification classes that is AB and C. So in this way we have we can see that we have classified our inventory into 3 classes that is AB and C and we can see that how interestingly we have used the 6 components, 6 factors which are required to be considered for inventory segmentation. We have used a very simple method and very well known method that is AHP method for doing the classification of the available SKUs and from here we can see that how interestingly we have divided them into 3 classes where A class belong to those items which can which require a higher service level of 95 percentage, B class items are those items which requires a 90 percent of service level and the C A class items are those items which requires 85 percent of the service levels. So this is again another approach of inventory segmentation but this is not the only method this is just the differentiation of one single method. There are various other methods which can be simply developed and which can be used for doing the inventory segmentation and can be accordingly applied by the business for their respective usage. So in today's class just to summarize we have focused on inventory segmentation and what are the popular approaches of inventory segmentation, what are the other ways of following the inventory segmentation apart from the popular methods. There we have seen very complex approaches like mathematical programming, metaheuristics, machine learning are quite popular in today's time for inventory segmentation and lastly we have understood about the classification of inventory into ABC classification by following a very simple method of AHP which considers multiple factors into the decision and decide about the priorities for every SQU. So with this we will end the second module where we have finished about understanding on supply chain segmentation, the need for supply chain segmentation, how it can help in understanding and managing the different supply chain challenges and going forward we have seen the different approaches for doing supply chain segmentation. Some of these approaches are standard approaches, some of these approaches we have seen through some cases and with this we will be ending our second week course element. Thank you, have a great day.